Loading…

Worked solutions

IB Computer Science HL · Paper 2

Computer Science, Higher Level

Session: May 2017
Marks: 260
Duration: 1 hour 20 min
Calculator: Not allowed
Questions: 16

Q1 Q2 Q3 Q4 Q5 Q6 Q7 Q8 Q9 Q10 Q11 Q12 Q13 Q14 Q15 Q16

Keep your copy of the paper open alongside. We don't reproduce the questions: each part below is labelled to match the May 2017 Global paper, which your school or the IB can provide.

Practice mode

Hide our working until you've had a go.

Option A: Databases · Option

Questions 1–4 · 65 marks

Question 1

13 marks

School parent database system, input validation, and concurrent online lunch payments

Databases

(a)

2 marks

Identify two operational advantages of deploying a Database Management System (DBMS) over flat files or spreadsheets:
1) Structured querying and reporting: Enables rapid filtering, sorting, and extraction of parent records to generate customized invoices or administrative reports.
2) Data security and access control: Centralizes data organization with built-in authentication, validation constraints, and backup routines that minimize the risk of data loss.

Mentor tip

Focus on data querying, security, or modular extensibility rather than raw physical disk capacity.

(b)

1 mark

State one relevant parent attribute stored in the school database:
Parent full name, residential address, telephone number, or email address.

Mentor tip

Ensure the attribute specifically pertains to a parent record in a school context.

(c)

2 marks

Identify an input error-reduction technique: Data validation (or data verification).
Elaborate on its mechanism: Applies automated rules such as type checks, length checks, or format checks (for example, verifying that a phone number contains only digits or an email contains an '@' symbol) before permitting the record to be saved.

Mentor tip

State the specific validation technique and provide a concrete example of how it restricts malformed input.

(d)

5 marks

Trace the sequential interaction between the parent and the DBMS during an online lunch payment:
1) Notification retrieval: The DBMS queries parent contact details and outstanding balances to trigger the automated billing email.
2) Authentication: The parent enters login credentials, which the DBMS validates and verifies against stored user accounts.
3) Partitioned data access: Upon verification, the DBMS grants restricted read access to the specific child's lunch balance and activity records.
4) Transaction execution and update: The parent submits payment credentials; once processed, the DBMS updates the accounting balance in an atomic transaction.
5) Confirmation and session termination: The DBMS passes session details to the mail server to dispatch a payment receipt and closes the database session upon logout or timeout.

Mentor tip

Follow the chronological lifecycle of the transaction from login authentication through balance update to email confirmation.

(e)

3 marks

Explain how the DBMS manages simultaneous online payments from multiple parents:
1) Concurrent multi-user access: The DBMS supports concurrent read and write operations across multiple active sessions.
2) Record partitioning and row-level locking: Each parent is restricted to their own family's records, and the DBMS locks only the specific rows being updated.
3) Transaction isolation: Because updates occur on non-overlapping partitions of the table, transactions execute in isolation without race conditions or deadlocks.

Mentor tip

Highlight how row-level locking and workspace partitioning allow simultaneous transactions without data collisions.

Question 2

12 marks

Relational database fundamentals, entity-relationship modelling, and referential integrity

Databases

(a)

1 mark

Define a relational database:
A collection of structured tables (relations) linked together via common attributes (primary and foreign keys).

Mentor tip

Mention both multiple tables and the key relationships that link them.

(b)

4 marks

Explain how Entity-Relationship Diagrams (ERDs) support data modelling:
1) Visual abstraction: Provides a compact graphical representation of entities, attributes, and the cardinality of relationships between them.
2) Implementation independence: Facilitates database planning and schema extension independently of the specific DBMS software used.
3) Logical blueprint: Informs the logical structure of the relational system, guiding table creation and foreign key placement.
4) Team collaboration: Enables systematic, standardized communication across development teams and simplifies future system upgrades.

Mentor tip

Explain why an ERD is useful for database designers rather than merely listing the shapes drawn in a diagram.

(c)

2 marks

Identify the anomaly in the provided tables:
1) Referential integrity violation: In TABLE_2, the third row (Member_ID 11) references Student_ID = 21, which does not exist in TABLE_1.
2) Operational consequence: Any query performing a JOIN between TABLE_1 and TABLE_2 to retrieve the email addresses of all chess club members will fail to resolve an email for Student_ID 21.

Mentor tip

Reference the exact orphan foreign key value (Student_ID = 21 in TABLE_2) and its impact on table joins.

(d)(i)

2 marks

Describe the issue caused by deleting Student_ID = 12 from TABLE_1:
1) Orphaned foreign key: The record for the chess club Captain (Member_ID = 1) in TABLE_2 still references Student_ID = 12.
2) Query failure or false linkage: Queries attempting to look up the Captain's email address in TABLE_1 will return no match, or if Student_ID 12 is later reassigned to a new student, that new student will be falsely linked as the chess club Captain.

Mentor tip

Connect the deletion in TABLE_1 directly to the remaining Captain record in TABLE_2.

(d)(ii)

3 marks

Outline the relational property that prevents unwanted deletion anomalies:
1) Property identification: Referential integrity.
2) Rule definition: Requires that every foreign key value in a referencing table (Student_ID in TABLE_2) must match an existing primary key value in the referenced table (Student_ID in TABLE_1).
3) DBMS enforcement mechanism: The DBMS prevents orphan records either by blocking the deletion in TABLE_1 if dependent records exist, or by executing a cascading delete (or setting NULL) on matching Student_ID rows in TABLE_2.

Mentor tip

Define referential integrity clearly and state how cascading deletes or deletion restrictions enforce it.

Question 3

20 marks

Telecom SMS data structures, normalization to 3NF, SQL querying, and DBA ethics

Databases

(a)(i)

2 marks

Outline a limitation of storing the raw SMS sequences directly as rows in a table:
1) Unbounded record length: Because a subscriber can send an arbitrary number of messages, the sequence has no fixed length.
2) Schema incompatibility: A relational table requires a fixed, predetermined number of columns, whereas repeating groups of receiver and message details would demand a variable number of columns per row.

Mentor tip

Point out that relational tables cannot dynamically add columns as a subscriber sends more text messages.

(a)(ii)

4 marks

Construct a single unnormalized/1NF table storing each SMS message as a separate row with a unique primary key (or composite key):
Columns: MsgCode (Primary Key) | SenderPhoneNumber | SenderName | ReceiverPhoneNumber | ReceiverName | TextMessage
Row 1: A0001 | 1111 | fred | 1234 | smith | Hi there
Row 2: A0002 | 1111 | fred | 1234 | smith | Call me
Row 3: A0003 | 1234 | smith | 1111 | fred | Later

Mentor tip

Include a clear primary key (such as MsgCode), separate columns for sender and receiver attributes, and all three sample messages.

(b)

2 marks

Define data redundancy:
1) The unnecessary repetition or duplication of identical data attributes or values within a database.
2) Occurs either across multiple rows within a single unnormalized table or across separate tables when the schema is not properly normalized.

Mentor tip

Distinguish between storing duplicate values unnecessarily and linking tables via foreign keys.

(c)

4 marks

Decompose the table from part (a)(ii) into two normalized relations to eliminate redundancy:
1) SUBSCRIBERS table: Stores subscriber details once.
Schema: SUBSCRIBERS(PhoneNumber [PK], Name)
2) SMS table: Stores individual message transactions using foreign keys referencing SUBSCRIBERS.
Schema: SMS(MsgCode [PK], SenderPhoneNumber [FK], ReceiverPhoneNumber [FK], TextMessage)

Mentor tip

Ensure subscriber names are removed from the SMS table and placed exclusively in the SUBSCRIBERS lookup table with underlined primary keys.

(d)

3 marks

Suggest the procedural query steps to retrieve the receiver's name for a given text message:
1) Join tables: Perform an inner join between the SMS table and the SUBSCRIBERS table where SMS.ReceiverPhoneNumber = SUBSCRIBERS.PhoneNumber.
2) Filter by message: Apply a selection condition matching the target message text (WHERE SMS.TextMessage = 'target_text').
3) Project attribute: Extract and output the Name field from the joined SUBSCRIBERS record.

Mentor tip

You can express the steps either in structured English or as an SQL SELECT ... FROM ... JOIN ... WHERE statement.

(e)

2 marks

Outline one routine responsibility of a Database Administrator (DBA):
1) Regular database backups: Executing scheduled backups and testing recovery procedures to prevent catastrophic data loss.
2) Alternative: Routine consistency/index maintenance, managing role-based access permissions, or applying security patches to prevent vulnerabilities.

Mentor tip

Pair the administrative task with the specific operational benefit or risk it mitigates.

(f)

3 marks

Discuss the ethical implications of external agencies analysing subscriber SMS records:
1) Privacy violation: Analysing private text messages without informed consent infringes upon individual privacy and anonymity.
2) Profiling and exploitation risks: External parties can aggregate message contents to infer sensitive habits (such as financial distress, health conditions, or physical locations), enabling targeted spam, surveillance, or identity fraud.
3) Competing public interest: Conversely, controlled access by law enforcement under judicial oversight may be ethically argued as necessary to counter severe national security threats or organized crime.

Mentor tip

Balance the individual right to privacy and protection from profiling against potential public safety justifications.

Question 4

20 marks

Educational authority data warehouse, ETL pipelines, and data mining analytics

Databases

(a)

2 marks

Define a data warehouse:
1) A centralized repository that integrates large volumes of historical and current data collected over time from multiple disparate sources.
2) Structured as non-volatile, time-stamped storage optimized for complex analytical querying, trend exploration, and decision support rather than routine transaction processing.

Mentor tip

Include key characteristics: central repository, integrated from multiple sources, time-variant/historical, and non-volatile.

(b)

4 marks

Describe the Extract, Transform, Load (ETL) process for migrating school records into the central data warehouse:
1) Extraction: Identify and select the relevant student and examination records from each of the 1,000+ local school databases, staging them temporarily.
2) Transformation (Cleaning and Standardization): Reconcile differing local schemas into a unified global format, scrub missing or corrupted entries, standardize data types (e.g., date formats or grade scales), and deduplicate records.
3) Loading: Transmit the transformed datasets into the central data warehouse repository, typically during scheduled off-peak windows so daily school operations are not disrupted.

Mentor tip

Explicitly name and describe all three stages: Extract, Transform (giving a concrete cleaning example), and Load.

(c)

4 marks

Explain the analytical process to identify schools with unusually low mathematics scores for a specific age group:
1) Data selection: Query the data warehouse to filter mathematics test scores for the target student age group and time period.
2) Aggregation: Compute the mean mathematics score for each individual school alongside the overall national benchmark average.
3) Deviation detection: Run an outlier/deviation detection algorithm (comparing each school's mean against a minimum threshold or standard deviation below the national mean).
4) Reporting: Output the list of school identifiers flagged as low-performing outliers.

Mentor tip

Focus on the data mining workflow (filtering, aggregating averages, and applying deviation detection) rather than writing raw SQL syntax.

(d)

2 marks

Define data mining:
1) The automated process of analysing massive datasets to discover previously unknown, implicit patterns, correlations, and trends.
2) Transforms raw archived data into actionable knowledge for strategic decision-making.

Mentor tip

Emphasize knowledge discovery and pattern recognition across large datasets.

(e)

4 marks

Compare cluster analysis and association analysis for evaluating 10-year grade 6 reading test distributions:
1) Cluster analysis mechanism: Partitioning technique that groups schools or students exhibiting similar reading score profiles without predefined labels.
2) Cluster application to scenario: Reveals geographic or demographic groupings of grade 6 reading performance over the decade (e.g., identifying regions where reading proficiency has steadily improved or stagnated).
3) Association analysis mechanism: Rule-based technique that discovers co-occurring variables and correlations across attributes.
4) Association application to scenario: Uncovers links between grade 6 reading scores and non-academic variables across the dataset (e.g., whether students with specific health record indicators or gender attributes correlate with higher reading scores).

Mentor tip

Clearly define both techniques and anchor each one to the grade 6 reading test scenario.

(f)

4 marks

Suggest negative social impacts of publishing the cluster and association mining results:
1) Social stigmatization and morale loss: Publishing cluster rankings can label certain schools or neighbourhoods as chronically failing, demoralizing students and teachers.
2) Impact of cluster analysis: Grouping schools by low performance can trigger 'white flight' or reduced enrolment, starves local schools of community support, and depresses neighbourhood property values.
3) Impact of association analysis (correlation vs. causation): Association rules identify statistical correlations rather than causal proof.
4) Policy misdirection and stereotyping: Publicizing raw correlations (such as linking gender or health traits to reading ability) can reinforce harmful social stereotypes or lead policymakers to implement misguided educational interventions.

Mentor tip

Address the distinct social harm of both cluster analysis (regional stigmatization) and association analysis (confusing correlation with causation).

Option B: Modelling and simulation · Option

Questions 5–8 · 65 marks

Question 5

19 marks

Sports league probability matrix, spreadsheet macros, and mathematical modelling vs simulation

Modelling and simulation

(a)

1 mark

Locate row 3, column 2 in the PROB array (PROB[3, 2]):
0.1

PROB [3, 2] = 0.1

Mentor tip

Row index represents the winning team (team 3) and column index represents the opponent (team 2).

(b)

2 marks

In any game between team i and team j, the three mutually exclusive outcomes are: team i wins, team j wins, or they draw.
Subtract the win probabilities of both teams from 1:
1 - (PROB[i, j] + PROB[j, i])

P (draw) = 1 - (PROB [i, j] + PROB [j, i])

Mentor tip

Remember to subtract both team i's win probability and team j's win probability from 1.

(c)

1 mark

Each of the 4 teams plays against 3 other opponents, playing 10 games against each opponent:
3 × 10 = 30 games.

3 \times 10 = 30

Mentor tip

A team does not play against itself, so multiply 3 opponents by 10 games.

(d)

3 marks

Demonstrate how the spreadsheet macro calculates expected draws along the main diagonal (row 1 col A, row 2 col B, row 3 col C, row 4 col D):
1) Sum wins and losses: For a given team k, sum all expected wins across row k and all expected losses down column k.
2) Subtract from total games: Subtract that combined sum of wins and losses from 30 (total games played by team k) and store the result in the diagonal cell.
3) Concrete illustration for Team 1 (top-left diagonal cell, row 1 col A): Row 1 wins = 5 + 0 + 5 = 10. Column A losses = 2 + 9 + 3 = 14. Expected draws in row 1 col A = 30 - (10 + 14) = 6. Repeat similarly for Team 2 (= 4), Team 3 (= 4), and Team 4 (= 4).

Draws_{1, 1} = 30 - (SUM(row 1 wins) + SUM(col A losses)) = 30 - (10 + 14) = 6

Mentor tip

Include an explicit spreadsheet formula or numerical walkthrough showing how row wins and column losses are subtracted from 30.

(e)

4 marks

Construct the algorithm to update the PLAYED table given match result (i, j, m, n) where team i scored m and team j scored n:
```
if m > n then
PLAYED[i, j] = PLAYED[i, j] + 1
else if n > m then
PLAYED[j, i] = PLAYED[j, i] + 1
else
PLAYED[i, i] = PLAYED[i, i] + 1
PLAYED[j, j] = PLAYED[j, j] + 1
end if
```

Mentor tip

When the match is a draw (m == n), ensure that the diagonal draw counters for BOTH team i and team j are incremented.

(f)

4 marks

Distinguish between a mathematical model and a simulation in the context of the sports league:
1) Mathematical model definition: A formal representation of a system using mathematical variables, probabilities, and formulas.
2) Scenario link (Model): The PROB array, which codifies historical win/loss ratios into fixed probability rules.
3) Simulation definition: The execution or application of a model over time or across trials to observe and predict expected outcomes.
4) Scenario link (Simulation): Applying the PROB rules across a 10-game schedule to generate the predicted season results in PROB10.

Mentor tip

Explicitly connect the mathematical model to the PROB array and the simulation to generating the PROB10 season predictions.

(g)

4 marks

Discuss the benefits of comparing predicted results (PROB10) against actual results (PLAYED):
1) Detecting real-world shifts: Significant discrepancies reveal that underlying conditions have changed since previous seasons (e.g., new coaching staff, player injuries, or transfers).
2) Performance evaluation: Identifies specific teams that significantly outperformed or underperformed relative to historical expectations.
3) Model recalibration: Prompts analysts to update the probability matrix for the following season using current-year statistics.
4) Validating model utility: Helps determine whether historical form is a reliable predictor or if the model's assumptions are flawed.

Mentor tip

Discuss both real-world domain insights (team changes) and improvements to the predictive model itself (recalibrating probabilities).

Question 6

17 marks

Town water reservoir simulation, variable bounds, graphical interpretation, and CAD engineering

Modelling and simulation

(a)

1 mark

State the mathematical inequality between reservoir volume V and danger threshold D:
V <= D (or V < D)

V \leq D

Mentor tip

Any equivalent inequality showing V does not exceed D is valid.

(b)

6 marks

Describe the factors considered when establishing simulation parameter ranges for R, T, and E:
1) River and rainfall inflow (R) — Factor: Historical meteorological and hydrological fluctuations across seasons.
2) River and rainfall inflow (R) — Range determination: Analyse multi-year climate records to establish baseline averages, drought minimums, and extreme storm surge outliers.
3) Town consumption outflow (T) — Factor: Temporal variations in municipal water demand.
4) Town consumption outflow (T) — Range determination: Measure town usage across peak vs off-peak hours, weekdays vs weekends, and summer vs winter months.
5) Excess outlet capacity (E) — Factor: Derived net surplus between peak inflow R and minimum town consumption T when V approaches D.
6) Excess outlet capacity (E) — Range determination: Calculate the maximum possible excess volume required to drain water fast enough during worst-case flood scenarios so V never breaches D.

Mentor tip

Address all three variables (R, T, and E) separately, giving both the physical factor and how its numerical bounds are determined.

(c)

6 marks

Discuss how time-series graphs of the reservoir simulation are constructed and interpreted:
1) Graph construction — Governing equation: Compute reservoir volume at each time step using V_current = V_previous + R - T - E.
2) Graph construction — Parameter sweeps: Plot multiple time-series curves of V over time while varying R from drought minimums to flood maximums against different fixed outlet capacities E.
3) Graph construction — Threshold marker: Superimpose a horizontal reference line at V = D to visually demarcate the danger level.
4) Graph interpretation — Identifying failure modes: Locate specific combinations of high inflow R and low consumption T where the plotted curve crosses above the horizontal danger line D.
5) Graph interpretation — Sizing overflow capacity: Determine the minimum required outlet flow rate E that keeps the peak of every simulated curve safely below D.
6) Graph interpretation — Duration analysis: Assess how rapidly V rises during storm spikes to determine how quickly the emergency outlets must open.

V_{new} = V_{old} + R - T - E

Mentor tip

Divide your discussion equally between how the graphs are generated (equations, axes, reference line D) and how engineers read them.

(d)

4 marks

Explain how simulation outputs directly inform the physical design of the reservoir:
1) Sizing emergency outlets: The peak simulated value of E dictates the physical diameter and number of overflow spillways required.
2) Positioning outlets: Time-series rise rates indicate the exact elevation levels along the dam wall where relief outlets should be placed to trigger in time.
3) Material stress specifications: Knowing the maximum sustained water volume and flow velocities guides engineers in selecting appropriate structural materials and wall thicknesses.
4) Future-proofing capacity: Simulating projected town population growth (increased T) ensures the reservoir dimensions remain adequate for decades.

Mentor tip

Connect specific simulation metrics (like maximum E and water level heights) to physical engineering choices (outlet size, placement, and materials).

Question 7

9 marks

2D vs 3D visualization, CAD modelling advantages, and 3D rendering pipelines

Modelling and simulation

(a)

2 marks

Outline a scenario where 2D visualization is preferable to 3D visualization:
1) Scenario: Displaying a navigation route map on a mobile device (or plotting a 2D scientific velocity-time graph).
2) Justification: A 2D map communicates route directions clearly without visual clutter, requires significantly less memory and bandwidth, and renders rapidly on mobile hardware.

Mentor tip

Pair a concrete 2D use case with a technical advantage such as lower computational overhead or clearer readability.

(b)

3 marks

Outline three advantages of using 3D CAD software over hand-drawn sketches when designing a statue:
1) 360-degree viewing: A single 3D wireframe model can be freely rotated, zoomed, and inspected from any angle, whereas sketches only show fixed perspectives.
2) Non-destructive editing: The artist can easily modify proportions, undo changes, or scale dimensions instantly without erasing and redrawing physical paper sketches.
3) Digital durability and sharing: CAD files do not degrade physically and can be easily duplicated to test multiple design variations side by side.

Mentor tip

Focus on specific CAD capabilities like 3D rotation, instant scaling/undoing, and wireframe manipulation.

(c)

4 marks

Outline the rendering steps required to convert the 3D model into a realistic visualization of the statue:
1) Wireframe geometry setup: Start with the 3D polygon mesh or wireframe representation defining the statue's surface geometry.
2) Texture and material mapping: Apply surface textures, colours, and material properties (such as bronze, marble, or stone reflectance) to the mesh.
3) Lighting and ray tracing: Position virtual light sources and execute rendering algorithms (such as ray tracing or shading) to calculate reflections, shadows, and ambient occlusion.
4) Rasterization to display: Project the lit, textured 3D scene onto a 2D pixel viewport so the artist can inspect a photorealistic preview of the finished statue.

Mentor tip

Mention wireframe geometry, applying surface textures, lighting/ray tracing calculations, and producing the final realistic view.

Question 8

20 marks

Natural language processing (NLP), neural network training, fitness functions, and unsupervised learning

Modelling and simulation

(a)

4 marks

Outline two advantages of applying Natural Language Processing (NLP) to 600,000 academic posts:
1) Scale and processing speed: The volume of 600,000 posts is far too vast for human researchers to read and summarize manually within a feasible timeframe; NLP automates ingestion and parsing rapidly.
2) Complex pattern extraction: Academic opinions span numerous overlapping technological themes; NLP algorithms can systematically classify multi-topic relationships and sentiment patterns that human readers would overlook.

Mentor tip

Develop two distinct points: one addressing the sheer volume of data and the second addressing the complexity of multi-group classification.

(b)

2 marks

Identify two data quality challenges when ingesting text posts from diverse online sources:
1) Typographical errors, misspellings, and informal abbreviations/acronyms.
2) Regional dialects, multilingual posts requiring translation, or ambiguous slang.

Mentor tip

List two concrete linguistic or formatting obstacles encountered in raw user-generated text.

(c)

2 marks

Suggest how to develop a fitness function before running the full dataset through the system:
1) Sample analysis: Manually inspect and label a small, representative subset of the academic posts.
2) Metric formulation: Define target evaluation criteria (such as the frequency and co-occurrence of key technology terms) from this sample to construct the mathematical fitness function.

Mentor tip

Explain that a small human-verified sample is analysed first to establish the target criteria for the fitness function.

(d)

6 marks

Explain, with a diagram structure, how the neural network uses the fitness function to analyse the complete dataset:
1) Network architecture diagram: Depict an Input Layer (receiving tokenized text features), at least two Hidden Layers of interconnected nodes, and an Output Layer (producing summarized topic classifications).
2) Weighted connections: Label edges between layers with synaptic weights that scale signals passing from input nodes through hidden nodes.
3) Forward propagation: Input posts are fed through the hidden layers to generate a trial classification output.
4) Fitness evaluation: The output is evaluated against the fitness function (or error/cost metric) to measure how closely it matches desired summary criteria.
5) Backpropagation / weight adjustment: The error signal is propagated backward through the network (or evolved via a genetic algorithm) to adjust the connection weights.
6) Iterative convergence: The cycle repeats across the dataset until the network's output satisfies the fitness function within a predefined tolerance.

Mentor tip

Ensure your diagram clearly labels the Input Layer, at least two Hidden Layers, the Output Layer, weights, and a feedback loop for backpropagation.

(e)

6 marks

Discuss why unsupervised learning may be superior to supervised learning for summarizing future technology opinions:
1) Supervised learning mechanism: Requires pre-labelled training data and a known target outcome, mapping inputs into predefined categories.
2) Supervised limitation here: Because the survey asks about novel future technological advances, researchers may not know all emerging categories in advance, making it nearly impossible to craft an accurate fitness function for 600,000 diverse posts.
3) Unsupervised learning mechanism: Operates without predefined labels or target outputs, autonomously clustering data based on intrinsic similarities and hidden structures.
4) Unsupervised advantage here: Discovers unforeseen connections, emerging research themes, and latent sentiment groupings across the 600,000 posts that human researchers did not anticipate.
5) Scalability: Avoids the bottleneck and bias of manually labelling a training sample before processing the massive corpus.

Mentor tip

Contrast supervised learning (known target labels) with unsupervised learning (autonomous pattern discovery) and tie both to the unpredictability of future technology trends.

Option C: Web science · Option

Questions 9–12 · 65 marks

Question 9

16 marks

X-Sprintz e-commerce layout, HTML forms, server-side PHP/MySQL processing, and client-side scripts

Web science

(a)

4 marks

Outline two reasons for placing a consistent banner across the top of all X-Sprintz web pages:
1) Brand identity and recognition: Displaying the company logo and styling consistently at the top of every page ensures users immediately recognize they are browsing the authentic X-Sprintz store.
2) Intuitive navigation: The top banner conventionally acts as a clickable hotspot linking back to the homepage, appearing first on all device viewports (desktop and mobile) for effortless navigation.

Mentor tip

State both a branding benefit (visual consistency/recognition) and a functional usability benefit (homepage navigation link).

(b)

1 mark

Identify one item typically placed in an e-commerce page footer:
Copyright notice, privacy policy link, terms and conditions, or company contact details.

Mentor tip

Any standard legal or corporate footer element is suitable.

(c)(i)

1 mark

Identify the markup language of the form snippet:
HTML (or XHTML).

Mentor tip

Recognize standard tags such as <html>, <body>, and <form>.

(c)(ii)

3 marks

Describe the exact chronological sequence of user actions to complete the search form:
1) Select product category: The user chooses one option ('Spikes', 'Trail', or 'Endurance') from the drop-down `<select>` menu.
2) Enter shoe size: The user types their shoe size into the text `<input>` field.
3) Submit form: The user clicks the 'Send it' submit button to transmit the form data.

Mentor tip

List the three user interactions in the exact order they appear in the form.

(d)(i)

5 marks

Explain the server-side processing enabled by the HTML form and `process.php`:
1) Parameter reception: Upon submission, the server executes `process.php` and retrieves the `item` and `size` values from the HTTP `$_POST` array.
2) Server-side input validation: The `check_input()` function validates that `size` is numeric (`is_num`) and falls within the inclusive range [34, 46], echoing an error message if validation fails.
3) Database authentication: The script establishes an authenticated connection (`mysqli_connect`) to the local `CatalogDB` database using a username and password.
4) SQL query execution: The server executes a `SELECT` query filtering `CatalogDB` for records matching both the submitted `item` (`it`) and `size` (`sz`).
5) Result rendering and cleanup: If zero rows match, it echoes an out-of-stock message; otherwise, it iterates through the result set (`mysqli_fetch_assoc`) to output the product image, item, size, and price before closing the database connection (`mysqli_close`).

Mentor tip

Trace the full PHP lifecycle: receiving POST variables, validating numeric range [34, 46], connecting to CatalogDB, running the SELECT query, and formatting the output.

(d)(ii)

2 marks

Outline an example of a client-side script that could enhance the form:
1) Script example: A JavaScript function embedded via `<script>` tags that performs immediate client-side validation (verifying the size field is not blank and contains a number between 34 and 46 before submission) or dynamically adjusts UI styling/text magnification.
2) Client-side rationale: Executing validation or UI adjustments directly in the user's browser provides instant feedback and avoids unnecessary round-trip HTTP requests that would waste server bandwidth.

Mentor tip

Provide a concrete client-side JavaScript feature and explain why running it in the browser reduces server load.

Question 10

11 marks

Blogs vs discussion forums, keyword tagging for web crawlers, and search engine ranking

Web science

(a)

4 marks

Distinguish between a blog and a discussion forum with respect to posting and commenting:
1) Posting in a blog: Only the blog owner or authorized author (the scientific journalist) can publish primary articles and initiate new discussion topics.
2) Posting in a forum: Any registered community member can create a new thread and initiate a discussion topic.
3) Commenting in a blog: Readers may submit comments on the journalist's article, but comments are typically moderated by the blog owner to enforce editorial standards and filter out off-topic remarks.
4) Commenting in a forum: Registered participants converse horizontally across threads with minimal pre-moderation, fostering an open peer-to-peer community debate.

Mentor tip

Explicitly contrast both 'posting' (who can start a thread/article) and 'commenting' (editorial moderation vs open peer replies).

(b)

5 marks

Discuss how visible keyword tags on blog articles influence user navigation and web crawler indexing:
1) User navigation: Tags group related articles by theme, allowing readers to click a tag and browse all posts sharing that subject.
2) Internal hyperlink generation: Visible tags function as anchor links connecting pages across the blog's internal architecture.
3) Crawler discovery: When a web crawler visits a blog post, it parses the visible HTML tags and follows their hyperlinks to discover related articles.
4) Preventing orphan pages: This dense internal linking ensures older or deeply nested blog entries are reached and indexed rather than remaining isolated.
5) Semantic classification: The tag keywords provide explicit topical signals that help the search engine categorize the blog's subject matter accurately in its index.

Mentor tip

Connect visible tags to internal hyperlinks, explaining how crawlers follow those links to prevent unindexed orphan pages.

(c)

2 marks

State two reasons why a newly launched discussion forum initially receives a low search engine ranking:
1) Lack of inbound links (backlinks) and low visitor traffic compared to established sites.
2) Unproven domain authority/reputation or absence of syndication feeds (such as RSS) and active user content.

Mentor tip

Focus on PageRank/authority factors such as few external backlinks and low user traffic.

Question 11

18 marks

Assisted living ubiquitous computing, public vs private cloud storage, and healthcare protocols

Web science

(a)(i)

1 mark

Define ubiquitous computing (pervasive computing):
The integration of microprocessors and computational capability into everyday objects and environments so that computing occurs seamlessly anytime and anywhere.

Mentor tip

Emphasize computing embedded invisibly into everyday physical objects.

(a)(ii)

4 marks

Outline two examples of ubiquitous computing in the elderly assisted living project:
1) Wearable biometric smartwatch — Embedded device: A wristwatch equipped with heart-rate and accelerometer sensors. Healthcare use: Continuously monitors vital signs (pulse, blood pressure) and transmits alerts if an arrhythmia is detected.
2) Smart pressure-sensitive flooring — Embedded device: Floor tiles or bathroom mats containing pressure sensors. Healthcare use: Detects if an elderly resident has fallen and remains immobile, automatically alerting emergency medical staff.

Mentor tip

For each example, state both the physical object containing the embedded sensor and its specific healthcare monitoring function.

(b)

5 marks

Discuss the trade-offs between public and private clouds for storing patient healthcare data:
1) Public cloud characteristics: Infrastructure is shared among multiple tenants (multi-tenancy), managed by major commercial providers with high scalability and lower upfront cost.
2) Public cloud security perspective: While commercial providers employ dedicated security experts and advanced patching, shared infrastructure increases exposure if a multi-tenant breach occurs.
3) Private cloud characteristics: Dedicated infrastructure operated exclusively for a single healthcare organization, offering granular control over physical and logical access.
4) Private cloud security perspective: Isolating patient records on dedicated hardware reduces external attack surfaces, though the healthcare provider must fund and maintain its own specialist IT security staff.
5) Conclusion: Because patient medical telemetry is highly confidential and legally protected, a private cloud (combined with end-to-end encryption) is generally preferred to guarantee strict data sovereignty and privacy.

Mentor tip

Address both public and private clouds in detail and finish with a justified conclusion tailored to sensitive medical records.

(c)

2 marks

Outline one risk posed to elderly residents by relying on assisted living technologies:
1) Hardware or network failure: If a wearable sensor malfunctions or the home Wi-Fi drops, a medical emergency (such as a fall) may go unreported while caregivers assume the resident is safe.
2) Alternative: Loss of personal privacy/autonomy (continuous surveillance of daily habits) or forgetting to wear/charge a biometric device.

Mentor tip

Focus on a direct risk to the elderly person (such as undetected emergencies during device failure or loss of domestic privacy) rather than generic cloud storage security.

(d)

1 mark

State a wireless or mobile communication protocol:
Wi-Fi (IEEE 802.11), Bluetooth, 4G/LTE, 3G, GSM, or WiMAX.

Mentor tip

Name any standard wireless or cellular transmission protocol.

(e)

5 marks

Explain the importance of standardized protocols in the assisted living healthcare ecosystem:
1) Interoperability across heterogeneous devices: Assisted living combines wearables, floor sensors, mobile apps, and cloud servers from different manufacturers; open standards ensure these subsystems communicate seamlessly.
2) Secure data transmission: Cryptographic transport protocols (such as TLS/SSL) encrypt vital signs in transit from home sensors to the cloud, preventing eavesdropping or man-in-the-middle tampering.
3) Data integrity and reliability: Packet error-checking protocols guarantee that critical medical alerts arrive intact and uncorrupted so clinicians do not make decisions on faulty telemetry.
4) Future extensibility and backward compatibility: Adhering to industry standards allows the project to integrate newer medical sensors in the future without replacing the entire existing infrastructure.
5) Regulatory compliance: Standardized access control and logging protocols help healthcare providers audit data access and protect patient safety.

Mentor tip

Cover three dimensions: device interoperability/extensibility, encrypted transmission security (TLS), and data integrity for patient safety.

Question 12

20 marks

Semantic web vs traditional web, web graphs, audio search engines, P2P power laws, and collective intelligence

Web science

(a)

3 marks

State three features that distinguish the Semantic Web from the traditional World Wide Web:
1) Machine-readable meaning: The Semantic Web structures data so software agents can automatically interpret meaning and relationships, whereas the traditional web is designed for human reading.
2) Data interchange vs document interchange: The Semantic Web links granular data resources using URIs, RDF, and ontologies (OWL), whereas the traditional web links HTML documents via URLs.
3) Automated querying and inference: Users and agents can execute logical queries (e.g., SPARQL) to extract synthesized knowledge across sources, rather than merely browsing pages via keyword matching.

Mentor tip

Contrast HTML document browsing for humans with RDF/OWL structured data and automated machine reasoning.

(b)

2 marks

Outline the difference between a web graph and a sub-graph with an example:
1) Structural distinction: The web graph is a directed graph modelling the entire World Wide Web (where vertices represent web pages/URLs and directed edges represent hyperlinks between them), whereas a sub-graph is a mathematical subset of those vertices and directed edges.
2) Concrete example: The complete World Wide Web forms the full web graph, while all pages and internal links within the Wikipedia domain (or an university portal) form a sub-graph.

Mentor tip

Mention that edges are directed hyperlinks and provide a clear domain-level example of a sub-graph.

(c)

2 marks

Define the multimedia-web:
1) An evolution of the web specialized in indexing, delivering, and interacting with rich audiovisual media (audio, video, images) rather than plain text.
2) Enables non-textual search queries where users can search using multimedia inputs (such as an audio clip or image) to retrieve matching media.

Mentor tip

Highlight both multimedia content delivery and querying via non-text media inputs.

(d)

4 marks

Explain how an audio-input music search engine identifies a song:
1) Audio capture: The system records an incoming soundwave (or MIDI stream) hummed or played by the user.
2) Feature extraction: Signal processing algorithms convert the raw audio waveform into a digital acoustic fingerprint, extracting physical characteristics such as pitch contour, melody, rhythm, beat, or harmonic frequencies.
3) Database matching: The extracted acoustic signature is compared against a repository of indexed audio fingerprints in a music archive.
4) Ranked retrieval: The engine returns the closest matching tracks or audio excerpts ranked by similarity score.

Mentor tip

Focus on acoustic feature extraction (melody, pitch, rhythm, spectrogram fingerprinting) from soundwaves rather than typing song lyrics.

(e)

5 marks

Discuss how power laws (Zipf's law) explain why identically sized movies download at vastly different speeds on a P2P network:
1) Power-law popularity distribution: According to Zipf's law, movie download frequency is inversely proportional to its popularity rank (a few blockbuster movies account for the vast majority of downloads, while niche films lie in the long tail).
2) Distributed peer caching: In a P2P network, peers cache fragments or complete copies of files they download and serve them to other peers.
3) Fast parallel downloads for popular movies: A top-ranked movie is cached across thousands of active peer nodes simultaneously, allowing a client to download multiple chunks in parallel from nearby peers at high speed.
4) Scarcity of niche movies: A rarely requested movie in the long tail is held by very few (or zero) active peers in the P2P swarm.
5) Bottlenecks for rare files: Locating a rare movie takes longer, and downloading from a single distant peer or fallback repository server creates a bandwidth bottleneck, greatly increasing total download time.

f (r) \propto \frac{1}{r}

Mentor tip

Link Zipf's law (popularity rank) to the number of peers caching file fragments and the ability to download chunks in parallel.

(f)

4 marks

Explain how collective intelligence shapes the online purchasing habits of the general public:
1) Concept definition: Collective intelligence emerges when large groups of individuals share knowledge, ratings, and feedback online, producing aggregated insights superior to individual judgment.
2) Concrete e-commerce example: Online shoppers contributing verified product reviews, star ratings, and folksonomy tags on marketplaces, or backing niche creators via crowdfunding platforms.
3) Influence on consumer trust and trends: Aggregated user reviews and collaborative filtering recommendations guide prospective buyers toward high-rated items and steer them away from poor products.
4) Market and economic impact: Niche ethical or interest communities (such as zero-waste groups or indie gaming communities) can pool demand to create viable markets, drive economies of scale, and pressure vendors to alter pricing or product policies.

Mentor tip

Define collective intelligence, give a specific online shopping example (like crowdsourced reviews or crowdfunding), and explain how it alters buyer decisions and market trends.

Option D: Object-oriented programming · Option

Questions 13–16 · 65 marks

Question 13

10 marks

OOP class vs instantiation, inheritance hierarchies, and software libraries

Object-oriented programming

(a)

3 marks

Distinguish between a class and an instantiation of a class using the company scenario:
1) Class definition: A class is an abstract blueprint or template that defines the attributes (variables) and behaviours (methods) common to a category of objects.
2) Instantiation definition: Instantiation is the runtime creation of a concrete object in memory with its own specific attribute values based on the class blueprint.
3) Scenario example: `Employee` (or `SalesPerson`) is a class defining properties like `id` and `salary`, whereas creating a specific record for employee `'John Smith'` with `id = "101"` is an instantiation of that class.

Mentor tip

Contrast the abstract blueprint (class) with the concrete runtime object in memory (instantiation) and ground both in the company scenario.

(b)

4 marks

Explain two ways inheritance can be incorporated into the company's administration software:
1) GUI module hierarchy: Define a `BaseGUI` superclass containing standard window components (menus, company branding, login bar), and derive specialized subclasses (`SalesGUI`, `PayrollGUI`) that inherit the common layout while adding module-specific controls.
2) Staff hierarchy: Define an `Employee` superclass storing shared attributes (`id`, `name`, `baseSalary`), and create subclasses (`Manager`, `OfficeStaff`, `SalesPerson`) that inherit these fields while adding role-specific attributes like `commissionRate` or `bonus`.

Mentor tip

Provide two distinct superclass/subclass examples from the prompt (one for the GUI modules and one for the staff roles).

(c)

3 marks

Describe how pre-written code libraries facilitate software development:
1) Reduced development time: Developers can import standard, pre-built classes (such as GUI widgets, sorting routines, or `LinkedList` collections) rather than coding complex algorithms from scratch.
2) High reliability: Library routines have already been rigorously tested and debugged by the wider community, reducing software defects.
3) Modularity and team focus: Programmers can focus on the company's core business logic while relying on standardized library interfaces for low-level tasks.

Mentor tip

Highlight time saved, pre-tested reliability (fewer bugs), and reuse of standard functionality.

Question 14

35 marks

SalesPerson and Sales classes: constructors, encapsulation, UML, array algorithms, and polymorphism

Object-oriented programming

(a)

2 marks

Complete the `SalesPerson(String id)` constructor by assigning the ID parameter and instantiating the `salesHistory` array:
```java
public SalesPerson(String id) {
this.id = id;
this.salesHistory = new Sales[100]; // any reasonable array capacity
}
```

Mentor tip

Initialize both `this.id` and the `salesHistory` array (`count` is already initialized to 0 at declaration).

(b)

3 marks

Explain why accessor (getter) methods are required in `SalesPerson`:
1) Private encapsulation: The instance variables (`id`, `salesHistory`, `count`) are declared with the `private` access modifier.
2) Restricted direct access: External classes (such as `Main` or `Payroll`) are prevented from reading or corrupting these fields directly.
3) Controlled read interface: `public` accessor methods (`getId()`, `getCount()`) provide a safe, read-only mechanism for external classes to retrieve attribute values.

Mentor tip

Use exact OOP terminology: `private` instance variables, encapsulation, and `public` read access from outside the class.

(c)(i)

4 marks

Construct the UML class diagram showing the relationship between `SalesPerson` and `Sales`:
1) `SalesPerson` 3-compartment box:
- Attributes: `- id : String`, `- salesHistory : Sales[]`, `- count : int`
- Methods: `+ calcTotalSales() : double`, `+ largestSale() : Sales`
2) `Sales` 3-compartment box:
- Attributes: `- itemId : String`, `- value : double`, `- quantity : int`
- Methods: (accessors/constructors optional per prompt)
3) Relationship arrow: Draw a `has-a` aggregation/composition arrow (or 1-to-many `1..*` association line) pointing from `SalesPerson` to `Sales`.

Mentor tip

Include visibility symbols (`-` for private, `+` for public), data types for all attributes/methods, and a clear 'has-a' relationship link.

(c)(ii)

2 marks

Outline a negative consequence of modifying the `Sales` class design later:
1) Tight coupling / ripple effects: Because `SalesPerson` depends directly on `Sales` methods, altering method signatures or attribute types in `Sales` can break existing code in `SalesPerson`.
2) Concrete example: If `getValue()` or `getQuantity()` in `Sales` is renamed or changed, `calcTotalSales()` and `largestSale()` in `SalesPerson` will fail to compile and must be rewritten.

Mentor tip

Give a specific example showing how changing a method in `Sales` breaks `calcTotalSales()` in `SalesPerson`.

(d)

4 marks

Trace the execution of the `Main` code snippet line by line:
1) `salesPeople[2].getId()` -> Outputs the ID of the third salesperson: `102`
2) `salesPeople[0].getCount()` -> Two `Sales` objects were added to `salesPeople[0]`: `2`
3) `salesPeople[1].getSalesHistory(0).getValue()` -> Unit value of the first sale for `salesPeople[1]`: `2550.40` (or `2550.4`)
4) `salesPeople[0].calcTotalSales()` -> `(300.00 * 10) + (1000.00 * 2) = 3000.00 + 2000.00`: `5000.00` (or `5000.0` / `5000`)

300.00 \times 10 + 1000.00 \times 2 = 5000.00

Mentor tip

Write each of the four printed values clearly on its own line.

(e)

5 marks

Construct the `calcTotalSales()` method inside `SalesPerson`:
```java
public double calcTotalSales() {
double total = 0.0;
for (int i = 0; i < count; i++) {
total = total + (salesHistory[i].getValue() * salesHistory[i].getQuantity());
}
return total;
}
```

Mentor tip

Loop only up to `count` (the number of initialized sales), multiply `getValue()` by `getQuantity()` for each sale, and return the accumulated `total`.

(f)

5 marks

Construct the `highest()` method to return the ID of the salesperson with the maximum total sales across 100 `salesPeople`:
```java
public String highest() {
double maxTotal = -1.0;
int bestIndex = 0;
for (int i = 0; i < 100; i++) {
if (salesPeople[i].calcTotalSales() > maxTotal) {
maxTotal = salesPeople[i].calcTotalSales();
bestIndex = i;
}
}
return salesPeople[bestIndex].getId();
}
```

Mentor tip

Reuse `calcTotalSales()` and `getId()` via dot notation on each element of `salesPeople`.

(g)

4 marks

Construct `addSales(Sales s, String id)` in `Main` to append sale `s` to the matching salesperson:
```java
public void addSales(Sales s, String id) {
for (int i = 0; i < salesPeople.length; i++) {
if (salesPeople[i] != null && salesPeople[i].getId().equals(id)) {
salesPeople[i].setSalesHistory(s);
break;
}
}
}
```

Mentor tip

Search the `salesPeople` array for the matching `getId()`, then invoke `setSalesHistory(s)` on that object.

(h)

3 marks

Suggest code modifications needed so `Payroll` can calculate monthly commissions:
1) Add a `date` (or `month`) instance variable to the `Sales` class and include it in the `Sales` constructor.
2) Add a public `getDate()` (or `getMonth()`) accessor method in the `Sales` class.
3) Overload or update `calcTotalSales(int targetMonth)` in `SalesPerson` to filter and sum only `Sales` objects whose date matches the requested month.

Mentor tip

State the new attribute in `Sales`, its getter/constructor update, and the filtering adjustment in `calcTotalSales()`.

(i)

3 marks

Discuss how polymorphism is demonstrated in the provided classes:
1) Constructor overloading: The `SalesPerson` class defines two constructors sharing the same name: `SalesPerson(String id)` and `SalesPerson(String id, Sales[] s, int c)`.
2) Compile-time resolution: Having multiple methods/constructors with the same identifier but different parameter lists is static polymorphism (overloading).
3) Parameter matching: When a `new SalesPerson(...)` object is instantiated, the compiler selects which constructor to execute based on the number and types of arguments passed.

Mentor tip

Identify the overloaded constructors in `SalesPerson` and explain how the compiler dispatches the call based on argument signatures.

Question 15

7 marks

Replacing arrays with LinkedList<Sales> collections and finding the largest sale

Object-oriented programming

(a)

2 marks

Write the attribute declaration replacing `private Sales[] salesHistory;` with a generic `LinkedList`:
```java
private LinkedList<Sales> salesHistory = new LinkedList<Sales>();
```

Mentor tip

Include both the `LinkedList` class type and the generic `<Sales>` type parameter.

(b)

5 marks

Construct `largestSale()` in `SalesPerson` using `LinkedList` methods to return the `Sales` object with the highest total value:
```java
public Sales largestSale() {
double maxVal = -1.0;
Sales bestSale = null;
for (int i = 0; i < salesHistory.size(); i++) {
Sales current = salesHistory.get(i);
double saleTotal = current.getValue() * current.getQuantity();
if (saleTotal > maxVal) {
maxVal = saleTotal;
bestSale = current;
}
}
return bestSale;
}
```

Mentor tip

Use `salesHistory.size()` and `salesHistory.get(i)` (or an `Iterator<Sales>`) to traverse the list, comparing `getValue() * getQuantity()` and returning the `Sales` object.

Question 16

13 marks

Binary search tree Node references, post-order recursive trace, and finding the minimum value

Object-oriented programming

(a)

3 marks

Describe how object references function as a data type in the `Node` class:
1) Self-referential fields: The variables `left` and `right` are declared of reference type `Node`.
2) Memory address pointers: Rather than holding raw primitive integers, `left` and `right` store the memory addresses (references) pointing to other instantiated `Node` objects (the left and right children).
3) Null termination: When a node has no child on a given branch, its reference variable holds `null`.

Mentor tip

Identify `left` and `right` as reference variables that store memory pointers to other `Node` instances or `null`.

(b)

5 marks

Trace the post-order recursive execution of `print(root)` on the tree (`node1` = 33, `node2` = 12, `node3` = 14, `node4` = 50):
1) `print(node1 [33])` calls `print(node1.left)` -> enters `print(node2 [12])`.
2) `print(node2 [12])` calls `print(node2.left)` -> `null` (base case returns immediately).
3) `print(node2 [12])` calls `print(node2.right)` -> enters `print(node3 [14])`.
4) `print(node3 [14])` calls `print(node3.left)` -> `null` (returns) and `print(node3.right)` -> `null` (returns), then outputs `14`.
5) Control returns to `node2 [12]`, which completes its right call and outputs `12`.
6) Control returns to `node1 [33]`, which calls `print(node1.right)` -> enters `print(node4 [50])`.
7) `print(node4 [50])` calls `print(node4.left)` -> `null` (returns) and `print(node4.right)` -> `null` (returns), then outputs `50`.
8) Control returns to `node1 [33]`, which outputs `33`.
Final output sequence: `14 12 50 33`

Mentor tip

Clearly show each recursive call level, the `null` base case returns, and the exact post-order output sequence: 14, 12, 50, 33.

(c)

5 marks

Construct the `smallest(Node node)` method to return the minimum integer in a non-empty Binary Search Tree (located at the leftmost node):
Iterative solution:
```java
public int smallest(Node node) {
Node current = node;
while (current.left != null) {
current = current.left;
}
return current.data;
}
```
Alternative recursive solution:
```java
public int smallest(Node node) {
if (node.left == null) {
return node.data;
}
return smallest(node.left);
}
```

Mentor tip

In a binary search tree, smaller values are always placed in the left subtree, so traverse `.left` references until `left == null`.

Our analysis

Where the marks are

Each part sorted by the topic it mostly tests. Databases carries the most marks on this paper.

Databases65 marks · 25%
Q1, Q2, Q3, Q4
Modelling and simulation65 marks · 25%
Q5, Q6, Q7, Q8
Web science65 marks · 25%
Q9, Q10, Q11, Q12
Object-oriented programming65 marks · 25%
Q13, Q14, Q15, Q16

Mentor insights

Traps to watch for

01
Databases: Telecom SMS data structures, normalization to 3NF, SQL querying, · Q3
Point out that relational tables cannot dynamically add columns as a subscriber sends more text messages.
02
Modelling and simulation: Natural language processing (NLP), neural network training, fitne · Q8
Develop two distinct points: one addressing the sheer volume of data and the second addressing the complexity of multi-group classification.
03
Web science: Semantic web vs traditional web, web graphs, audio search engines · Q12
Contrast HTML document browsing for humans with RDF/OWL structured data and automated machine reasoning.
04
Object-oriented programming: SalesPerson and Sales classes: constructors, encapsulation, UML, · Q14
Initialize both `this.id` and the `salesHistory` array (`count` is already initialized to 0 at declaration).

Solutions and tips are GeorgeTheTutor's own; the questions are not reproduced here. Not affiliated with or endorsed by the International Baccalaureate Organization.

Want someone to go through your paper with you?

Our maths mentors took these exams recently. One-to-one, they'll find where your marks are going and fix it before May.

Find your maths mentor More past papers

Worked solutions

IB Computer Science HL · Paper 2

Computer Science, Higher Level

Session: May 2017
Marks: 260
Duration: 1 hour 20 min
Calculator: Not allowed
Questions: 16

Q1 Q2 Q3 Q4 Q5 Q6 Q7 Q8 Q9 Q10 Q11 Q12 Q13 Q14 Q15 Q16

Keep your copy of the paper open alongside. We don't reproduce the questions: each part below is labelled to match the May 2017 Global paper, which your school or the IB can provide.

Practice mode

Hide our working until you've had a go.

Option A: Databases · Option

Questions 1–4 · 65 marks

Question 1

13 marks

School parent database system, input validation, and concurrent online lunch payments

Databases

(a)

2 marks

Identify two operational advantages of deploying a Database Management System (DBMS) over flat files or spreadsheets:
1) Structured querying and reporting: Enables rapid filtering, sorting, and extraction of parent records to generate customized invoices or administrative reports.
2) Data security and access control: Centralizes data organization with built-in authentication, validation constraints, and backup routines that minimize the risk of data loss.

Mentor tip

Focus on data querying, security, or modular extensibility rather than raw physical disk capacity.

(b)

1 mark

State one relevant parent attribute stored in the school database:
Parent full name, residential address, telephone number, or email address.

Mentor tip

Ensure the attribute specifically pertains to a parent record in a school context.

(c)

2 marks

Identify an input error-reduction technique: Data validation (or data verification).
Elaborate on its mechanism: Applies automated rules such as type checks, length checks, or format checks (for example, verifying that a phone number contains only digits or an email contains an '@' symbol) before permitting the record to be saved.

Mentor tip

State the specific validation technique and provide a concrete example of how it restricts malformed input.

(d)

5 marks

Trace the sequential interaction between the parent and the DBMS during an online lunch payment:
1) Notification retrieval: The DBMS queries parent contact details and outstanding balances to trigger the automated billing email.
2) Authentication: The parent enters login credentials, which the DBMS validates and verifies against stored user accounts.
3) Partitioned data access: Upon verification, the DBMS grants restricted read access to the specific child's lunch balance and activity records.
4) Transaction execution and update: The parent submits payment credentials; once processed, the DBMS updates the accounting balance in an atomic transaction.
5) Confirmation and session termination: The DBMS passes session details to the mail server to dispatch a payment receipt and closes the database session upon logout or timeout.

Mentor tip

Follow the chronological lifecycle of the transaction from login authentication through balance update to email confirmation.

(e)

3 marks

Explain how the DBMS manages simultaneous online payments from multiple parents:
1) Concurrent multi-user access: The DBMS supports concurrent read and write operations across multiple active sessions.
2) Record partitioning and row-level locking: Each parent is restricted to their own family's records, and the DBMS locks only the specific rows being updated.
3) Transaction isolation: Because updates occur on non-overlapping partitions of the table, transactions execute in isolation without race conditions or deadlocks.

Mentor tip

Highlight how row-level locking and workspace partitioning allow simultaneous transactions without data collisions.

Question 2

12 marks

Relational database fundamentals, entity-relationship modelling, and referential integrity

Databases

(a)

1 mark

Define a relational database:
A collection of structured tables (relations) linked together via common attributes (primary and foreign keys).

Mentor tip

Mention both multiple tables and the key relationships that link them.

(b)

4 marks

Explain how Entity-Relationship Diagrams (ERDs) support data modelling:
1) Visual abstraction: Provides a compact graphical representation of entities, attributes, and the cardinality of relationships between them.
2) Implementation independence: Facilitates database planning and schema extension independently of the specific DBMS software used.
3) Logical blueprint: Informs the logical structure of the relational system, guiding table creation and foreign key placement.
4) Team collaboration: Enables systematic, standardized communication across development teams and simplifies future system upgrades.

Mentor tip

Explain why an ERD is useful for database designers rather than merely listing the shapes drawn in a diagram.

(c)

2 marks

Identify the anomaly in the provided tables:
1) Referential integrity violation: In TABLE_2, the third row (Member_ID 11) references Student_ID = 21, which does not exist in TABLE_1.
2) Operational consequence: Any query performing a JOIN between TABLE_1 and TABLE_2 to retrieve the email addresses of all chess club members will fail to resolve an email for Student_ID 21.

Mentor tip

Reference the exact orphan foreign key value (Student_ID = 21 in TABLE_2) and its impact on table joins.

(d)(i)

2 marks

Describe the issue caused by deleting Student_ID = 12 from TABLE_1:
1) Orphaned foreign key: The record for the chess club Captain (Member_ID = 1) in TABLE_2 still references Student_ID = 12.
2) Query failure or false linkage: Queries attempting to look up the Captain's email address in TABLE_1 will return no match, or if Student_ID 12 is later reassigned to a new student, that new student will be falsely linked as the chess club Captain.

Mentor tip

Connect the deletion in TABLE_1 directly to the remaining Captain record in TABLE_2.

(d)(ii)

3 marks

Outline the relational property that prevents unwanted deletion anomalies:
1) Property identification: Referential integrity.
2) Rule definition: Requires that every foreign key value in a referencing table (Student_ID in TABLE_2) must match an existing primary key value in the referenced table (Student_ID in TABLE_1).
3) DBMS enforcement mechanism: The DBMS prevents orphan records either by blocking the deletion in TABLE_1 if dependent records exist, or by executing a cascading delete (or setting NULL) on matching Student_ID rows in TABLE_2.

Mentor tip

Define referential integrity clearly and state how cascading deletes or deletion restrictions enforce it.

Question 3

20 marks

Telecom SMS data structures, normalization to 3NF, SQL querying, and DBA ethics

Databases

(a)(i)

2 marks

Outline a limitation of storing the raw SMS sequences directly as rows in a table:
1) Unbounded record length: Because a subscriber can send an arbitrary number of messages, the sequence has no fixed length.
2) Schema incompatibility: A relational table requires a fixed, predetermined number of columns, whereas repeating groups of receiver and message details would demand a variable number of columns per row.

Mentor tip

Point out that relational tables cannot dynamically add columns as a subscriber sends more text messages.

(a)(ii)

4 marks

Construct a single unnormalized/1NF table storing each SMS message as a separate row with a unique primary key (or composite key):
Columns: MsgCode (Primary Key) | SenderPhoneNumber | SenderName | ReceiverPhoneNumber | ReceiverName | TextMessage
Row 1: A0001 | 1111 | fred | 1234 | smith | Hi there
Row 2: A0002 | 1111 | fred | 1234 | smith | Call me
Row 3: A0003 | 1234 | smith | 1111 | fred | Later

Mentor tip

Include a clear primary key (such as MsgCode), separate columns for sender and receiver attributes, and all three sample messages.

(b)

2 marks

Define data redundancy:
1) The unnecessary repetition or duplication of identical data attributes or values within a database.
2) Occurs either across multiple rows within a single unnormalized table or across separate tables when the schema is not properly normalized.

Mentor tip

Distinguish between storing duplicate values unnecessarily and linking tables via foreign keys.

(c)

4 marks

Decompose the table from part (a)(ii) into two normalized relations to eliminate redundancy:
1) SUBSCRIBERS table: Stores subscriber details once.
Schema: SUBSCRIBERS(PhoneNumber [PK], Name)
2) SMS table: Stores individual message transactions using foreign keys referencing SUBSCRIBERS.
Schema: SMS(MsgCode [PK], SenderPhoneNumber [FK], ReceiverPhoneNumber [FK], TextMessage)

Mentor tip

Ensure subscriber names are removed from the SMS table and placed exclusively in the SUBSCRIBERS lookup table with underlined primary keys.

(d)

3 marks

Suggest the procedural query steps to retrieve the receiver's name for a given text message:
1) Join tables: Perform an inner join between the SMS table and the SUBSCRIBERS table where SMS.ReceiverPhoneNumber = SUBSCRIBERS.PhoneNumber.
2) Filter by message: Apply a selection condition matching the target message text (WHERE SMS.TextMessage = 'target_text').
3) Project attribute: Extract and output the Name field from the joined SUBSCRIBERS record.

Mentor tip

You can express the steps either in structured English or as an SQL SELECT ... FROM ... JOIN ... WHERE statement.

(e)

2 marks

Outline one routine responsibility of a Database Administrator (DBA):
1) Regular database backups: Executing scheduled backups and testing recovery procedures to prevent catastrophic data loss.
2) Alternative: Routine consistency/index maintenance, managing role-based access permissions, or applying security patches to prevent vulnerabilities.

Mentor tip

Pair the administrative task with the specific operational benefit or risk it mitigates.

(f)

3 marks

Discuss the ethical implications of external agencies analysing subscriber SMS records:
1) Privacy violation: Analysing private text messages without informed consent infringes upon individual privacy and anonymity.
2) Profiling and exploitation risks: External parties can aggregate message contents to infer sensitive habits (such as financial distress, health conditions, or physical locations), enabling targeted spam, surveillance, or identity fraud.
3) Competing public interest: Conversely, controlled access by law enforcement under judicial oversight may be ethically argued as necessary to counter severe national security threats or organized crime.

Mentor tip

Balance the individual right to privacy and protection from profiling against potential public safety justifications.

Question 4

20 marks

Educational authority data warehouse, ETL pipelines, and data mining analytics

Databases

(a)

2 marks

Define a data warehouse:
1) A centralized repository that integrates large volumes of historical and current data collected over time from multiple disparate sources.
2) Structured as non-volatile, time-stamped storage optimized for complex analytical querying, trend exploration, and decision support rather than routine transaction processing.

Mentor tip

Include key characteristics: central repository, integrated from multiple sources, time-variant/historical, and non-volatile.

(b)

4 marks

Describe the Extract, Transform, Load (ETL) process for migrating school records into the central data warehouse:
1) Extraction: Identify and select the relevant student and examination records from each of the 1,000+ local school databases, staging them temporarily.
2) Transformation (Cleaning and Standardization): Reconcile differing local schemas into a unified global format, scrub missing or corrupted entries, standardize data types (e.g., date formats or grade scales), and deduplicate records.
3) Loading: Transmit the transformed datasets into the central data warehouse repository, typically during scheduled off-peak windows so daily school operations are not disrupted.

Mentor tip

Explicitly name and describe all three stages: Extract, Transform (giving a concrete cleaning example), and Load.

(c)

4 marks

Explain the analytical process to identify schools with unusually low mathematics scores for a specific age group:
1) Data selection: Query the data warehouse to filter mathematics test scores for the target student age group and time period.
2) Aggregation: Compute the mean mathematics score for each individual school alongside the overall national benchmark average.
3) Deviation detection: Run an outlier/deviation detection algorithm (comparing each school's mean against a minimum threshold or standard deviation below the national mean).
4) Reporting: Output the list of school identifiers flagged as low-performing outliers.

Mentor tip

Focus on the data mining workflow (filtering, aggregating averages, and applying deviation detection) rather than writing raw SQL syntax.

(d)

2 marks

Define data mining:
1) The automated process of analysing massive datasets to discover previously unknown, implicit patterns, correlations, and trends.
2) Transforms raw archived data into actionable knowledge for strategic decision-making.

Mentor tip

Emphasize knowledge discovery and pattern recognition across large datasets.

(e)

4 marks

Compare cluster analysis and association analysis for evaluating 10-year grade 6 reading test distributions:
1) Cluster analysis mechanism: Partitioning technique that groups schools or students exhibiting similar reading score profiles without predefined labels.
2) Cluster application to scenario: Reveals geographic or demographic groupings of grade 6 reading performance over the decade (e.g., identifying regions where reading proficiency has steadily improved or stagnated).
3) Association analysis mechanism: Rule-based technique that discovers co-occurring variables and correlations across attributes.
4) Association application to scenario: Uncovers links between grade 6 reading scores and non-academic variables across the dataset (e.g., whether students with specific health record indicators or gender attributes correlate with higher reading scores).

Mentor tip

Clearly define both techniques and anchor each one to the grade 6 reading test scenario.

(f)

4 marks

Suggest negative social impacts of publishing the cluster and association mining results:
1) Social stigmatization and morale loss: Publishing cluster rankings can label certain schools or neighbourhoods as chronically failing, demoralizing students and teachers.
2) Impact of cluster analysis: Grouping schools by low performance can trigger 'white flight' or reduced enrolment, starves local schools of community support, and depresses neighbourhood property values.
3) Impact of association analysis (correlation vs. causation): Association rules identify statistical correlations rather than causal proof.
4) Policy misdirection and stereotyping: Publicizing raw correlations (such as linking gender or health traits to reading ability) can reinforce harmful social stereotypes or lead policymakers to implement misguided educational interventions.

Mentor tip

Address the distinct social harm of both cluster analysis (regional stigmatization) and association analysis (confusing correlation with causation).

Option B: Modelling and simulation · Option

Questions 5–8 · 65 marks

Question 5

19 marks

Sports league probability matrix, spreadsheet macros, and mathematical modelling vs simulation

Modelling and simulation

(a)

1 mark

Locate row 3, column 2 in the PROB array (PROB[3, 2]):
0.1

PROB [3, 2] = 0.1

Mentor tip

Row index represents the winning team (team 3) and column index represents the opponent (team 2).

(b)

2 marks

In any game between team i and team j, the three mutually exclusive outcomes are: team i wins, team j wins, or they draw.
Subtract the win probabilities of both teams from 1:
1 - (PROB[i, j] + PROB[j, i])

P (draw) = 1 - (PROB [i, j] + PROB [j, i])

Mentor tip

Remember to subtract both team i's win probability and team j's win probability from 1.

(c)

1 mark

Each of the 4 teams plays against 3 other opponents, playing 10 games against each opponent:
3 × 10 = 30 games.

3 \times 10 = 30

Mentor tip

A team does not play against itself, so multiply 3 opponents by 10 games.

(d)

3 marks

Demonstrate how the spreadsheet macro calculates expected draws along the main diagonal (row 1 col A, row 2 col B, row 3 col C, row 4 col D):
1) Sum wins and losses: For a given team k, sum all expected wins across row k and all expected losses down column k.
2) Subtract from total games: Subtract that combined sum of wins and losses from 30 (total games played by team k) and store the result in the diagonal cell.
3) Concrete illustration for Team 1 (top-left diagonal cell, row 1 col A): Row 1 wins = 5 + 0 + 5 = 10. Column A losses = 2 + 9 + 3 = 14. Expected draws in row 1 col A = 30 - (10 + 14) = 6. Repeat similarly for Team 2 (= 4), Team 3 (= 4), and Team 4 (= 4).

Draws_{1, 1} = 30 - (SUM(row 1 wins) + SUM(col A losses)) = 30 - (10 + 14) = 6

Mentor tip

Include an explicit spreadsheet formula or numerical walkthrough showing how row wins and column losses are subtracted from 30.

(e)

4 marks

Construct the algorithm to update the PLAYED table given match result (i, j, m, n) where team i scored m and team j scored n:
```
if m > n then
PLAYED[i, j] = PLAYED[i, j] + 1
else if n > m then
PLAYED[j, i] = PLAYED[j, i] + 1
else
PLAYED[i, i] = PLAYED[i, i] + 1
PLAYED[j, j] = PLAYED[j, j] + 1
end if
```

Mentor tip

When the match is a draw (m == n), ensure that the diagonal draw counters for BOTH team i and team j are incremented.

(f)

4 marks

Distinguish between a mathematical model and a simulation in the context of the sports league:
1) Mathematical model definition: A formal representation of a system using mathematical variables, probabilities, and formulas.
2) Scenario link (Model): The PROB array, which codifies historical win/loss ratios into fixed probability rules.
3) Simulation definition: The execution or application of a model over time or across trials to observe and predict expected outcomes.
4) Scenario link (Simulation): Applying the PROB rules across a 10-game schedule to generate the predicted season results in PROB10.

Mentor tip

Explicitly connect the mathematical model to the PROB array and the simulation to generating the PROB10 season predictions.

(g)

4 marks

Discuss the benefits of comparing predicted results (PROB10) against actual results (PLAYED):
1) Detecting real-world shifts: Significant discrepancies reveal that underlying conditions have changed since previous seasons (e.g., new coaching staff, player injuries, or transfers).
2) Performance evaluation: Identifies specific teams that significantly outperformed or underperformed relative to historical expectations.
3) Model recalibration: Prompts analysts to update the probability matrix for the following season using current-year statistics.
4) Validating model utility: Helps determine whether historical form is a reliable predictor or if the model's assumptions are flawed.

Mentor tip

Discuss both real-world domain insights (team changes) and improvements to the predictive model itself (recalibrating probabilities).

Question 6

17 marks

Town water reservoir simulation, variable bounds, graphical interpretation, and CAD engineering

Modelling and simulation

(a)

1 mark

State the mathematical inequality between reservoir volume V and danger threshold D:
V <= D (or V < D)

V \leq D

Mentor tip

Any equivalent inequality showing V does not exceed D is valid.

(b)

6 marks

Describe the factors considered when establishing simulation parameter ranges for R, T, and E:
1) River and rainfall inflow (R) — Factor: Historical meteorological and hydrological fluctuations across seasons.
2) River and rainfall inflow (R) — Range determination: Analyse multi-year climate records to establish baseline averages, drought minimums, and extreme storm surge outliers.
3) Town consumption outflow (T) — Factor: Temporal variations in municipal water demand.
4) Town consumption outflow (T) — Range determination: Measure town usage across peak vs off-peak hours, weekdays vs weekends, and summer vs winter months.
5) Excess outlet capacity (E) — Factor: Derived net surplus between peak inflow R and minimum town consumption T when V approaches D.
6) Excess outlet capacity (E) — Range determination: Calculate the maximum possible excess volume required to drain water fast enough during worst-case flood scenarios so V never breaches D.

Mentor tip

Address all three variables (R, T, and E) separately, giving both the physical factor and how its numerical bounds are determined.

(c)

6 marks

Discuss how time-series graphs of the reservoir simulation are constructed and interpreted:
1) Graph construction — Governing equation: Compute reservoir volume at each time step using V_current = V_previous + R - T - E.
2) Graph construction — Parameter sweeps: Plot multiple time-series curves of V over time while varying R from drought minimums to flood maximums against different fixed outlet capacities E.
3) Graph construction — Threshold marker: Superimpose a horizontal reference line at V = D to visually demarcate the danger level.
4) Graph interpretation — Identifying failure modes: Locate specific combinations of high inflow R and low consumption T where the plotted curve crosses above the horizontal danger line D.
5) Graph interpretation — Sizing overflow capacity: Determine the minimum required outlet flow rate E that keeps the peak of every simulated curve safely below D.
6) Graph interpretation — Duration analysis: Assess how rapidly V rises during storm spikes to determine how quickly the emergency outlets must open.

V_{new} = V_{old} + R - T - E

Mentor tip

Divide your discussion equally between how the graphs are generated (equations, axes, reference line D) and how engineers read them.

(d)

4 marks

Explain how simulation outputs directly inform the physical design of the reservoir:
1) Sizing emergency outlets: The peak simulated value of E dictates the physical diameter and number of overflow spillways required.
2) Positioning outlets: Time-series rise rates indicate the exact elevation levels along the dam wall where relief outlets should be placed to trigger in time.
3) Material stress specifications: Knowing the maximum sustained water volume and flow velocities guides engineers in selecting appropriate structural materials and wall thicknesses.
4) Future-proofing capacity: Simulating projected town population growth (increased T) ensures the reservoir dimensions remain adequate for decades.

Mentor tip

Connect specific simulation metrics (like maximum E and water level heights) to physical engineering choices (outlet size, placement, and materials).

Question 7

9 marks

2D vs 3D visualization, CAD modelling advantages, and 3D rendering pipelines

Modelling and simulation

(a)

2 marks

Outline a scenario where 2D visualization is preferable to 3D visualization:
1) Scenario: Displaying a navigation route map on a mobile device (or plotting a 2D scientific velocity-time graph).
2) Justification: A 2D map communicates route directions clearly without visual clutter, requires significantly less memory and bandwidth, and renders rapidly on mobile hardware.

Mentor tip

Pair a concrete 2D use case with a technical advantage such as lower computational overhead or clearer readability.

(b)

3 marks

Outline three advantages of using 3D CAD software over hand-drawn sketches when designing a statue:
1) 360-degree viewing: A single 3D wireframe model can be freely rotated, zoomed, and inspected from any angle, whereas sketches only show fixed perspectives.
2) Non-destructive editing: The artist can easily modify proportions, undo changes, or scale dimensions instantly without erasing and redrawing physical paper sketches.
3) Digital durability and sharing: CAD files do not degrade physically and can be easily duplicated to test multiple design variations side by side.

Mentor tip

Focus on specific CAD capabilities like 3D rotation, instant scaling/undoing, and wireframe manipulation.

(c)

4 marks

Outline the rendering steps required to convert the 3D model into a realistic visualization of the statue:
1) Wireframe geometry setup: Start with the 3D polygon mesh or wireframe representation defining the statue's surface geometry.
2) Texture and material mapping: Apply surface textures, colours, and material properties (such as bronze, marble, or stone reflectance) to the mesh.
3) Lighting and ray tracing: Position virtual light sources and execute rendering algorithms (such as ray tracing or shading) to calculate reflections, shadows, and ambient occlusion.
4) Rasterization to display: Project the lit, textured 3D scene onto a 2D pixel viewport so the artist can inspect a photorealistic preview of the finished statue.

Mentor tip

Mention wireframe geometry, applying surface textures, lighting/ray tracing calculations, and producing the final realistic view.

Question 8

20 marks

Natural language processing (NLP), neural network training, fitness functions, and unsupervised learning

Modelling and simulation

(a)

4 marks

Outline two advantages of applying Natural Language Processing (NLP) to 600,000 academic posts:
1) Scale and processing speed: The volume of 600,000 posts is far too vast for human researchers to read and summarize manually within a feasible timeframe; NLP automates ingestion and parsing rapidly.
2) Complex pattern extraction: Academic opinions span numerous overlapping technological themes; NLP algorithms can systematically classify multi-topic relationships and sentiment patterns that human readers would overlook.

Mentor tip

Develop two distinct points: one addressing the sheer volume of data and the second addressing the complexity of multi-group classification.

(b)

2 marks

Identify two data quality challenges when ingesting text posts from diverse online sources:
1) Typographical errors, misspellings, and informal abbreviations/acronyms.
2) Regional dialects, multilingual posts requiring translation, or ambiguous slang.

Mentor tip

List two concrete linguistic or formatting obstacles encountered in raw user-generated text.

(c)

2 marks

Suggest how to develop a fitness function before running the full dataset through the system:
1) Sample analysis: Manually inspect and label a small, representative subset of the academic posts.
2) Metric formulation: Define target evaluation criteria (such as the frequency and co-occurrence of key technology terms) from this sample to construct the mathematical fitness function.

Mentor tip

Explain that a small human-verified sample is analysed first to establish the target criteria for the fitness function.

(d)

6 marks

Explain, with a diagram structure, how the neural network uses the fitness function to analyse the complete dataset:
1) Network architecture diagram: Depict an Input Layer (receiving tokenized text features), at least two Hidden Layers of interconnected nodes, and an Output Layer (producing summarized topic classifications).
2) Weighted connections: Label edges between layers with synaptic weights that scale signals passing from input nodes through hidden nodes.
3) Forward propagation: Input posts are fed through the hidden layers to generate a trial classification output.
4) Fitness evaluation: The output is evaluated against the fitness function (or error/cost metric) to measure how closely it matches desired summary criteria.
5) Backpropagation / weight adjustment: The error signal is propagated backward through the network (or evolved via a genetic algorithm) to adjust the connection weights.
6) Iterative convergence: The cycle repeats across the dataset until the network's output satisfies the fitness function within a predefined tolerance.

Mentor tip

Ensure your diagram clearly labels the Input Layer, at least two Hidden Layers, the Output Layer, weights, and a feedback loop for backpropagation.

(e)

6 marks

Discuss why unsupervised learning may be superior to supervised learning for summarizing future technology opinions:
1) Supervised learning mechanism: Requires pre-labelled training data and a known target outcome, mapping inputs into predefined categories.
2) Supervised limitation here: Because the survey asks about novel future technological advances, researchers may not know all emerging categories in advance, making it nearly impossible to craft an accurate fitness function for 600,000 diverse posts.
3) Unsupervised learning mechanism: Operates without predefined labels or target outputs, autonomously clustering data based on intrinsic similarities and hidden structures.
4) Unsupervised advantage here: Discovers unforeseen connections, emerging research themes, and latent sentiment groupings across the 600,000 posts that human researchers did not anticipate.
5) Scalability: Avoids the bottleneck and bias of manually labelling a training sample before processing the massive corpus.

Mentor tip

Contrast supervised learning (known target labels) with unsupervised learning (autonomous pattern discovery) and tie both to the unpredictability of future technology trends.

Option C: Web science · Option

Questions 9–12 · 65 marks

Question 9

16 marks

X-Sprintz e-commerce layout, HTML forms, server-side PHP/MySQL processing, and client-side scripts

Web science

(a)

4 marks

Outline two reasons for placing a consistent banner across the top of all X-Sprintz web pages:
1) Brand identity and recognition: Displaying the company logo and styling consistently at the top of every page ensures users immediately recognize they are browsing the authentic X-Sprintz store.
2) Intuitive navigation: The top banner conventionally acts as a clickable hotspot linking back to the homepage, appearing first on all device viewports (desktop and mobile) for effortless navigation.

Mentor tip

State both a branding benefit (visual consistency/recognition) and a functional usability benefit (homepage navigation link).

(b)

1 mark

Identify one item typically placed in an e-commerce page footer:
Copyright notice, privacy policy link, terms and conditions, or company contact details.

Mentor tip

Any standard legal or corporate footer element is suitable.

(c)(i)

1 mark

Identify the markup language of the form snippet:
HTML (or XHTML).

Mentor tip

Recognize standard tags such as <html>, <body>, and <form>.

(c)(ii)

3 marks

Describe the exact chronological sequence of user actions to complete the search form:
1) Select product category: The user chooses one option ('Spikes', 'Trail', or 'Endurance') from the drop-down `<select>` menu.
2) Enter shoe size: The user types their shoe size into the text `<input>` field.
3) Submit form: The user clicks the 'Send it' submit button to transmit the form data.

Mentor tip

List the three user interactions in the exact order they appear in the form.

(d)(i)

5 marks

Explain the server-side processing enabled by the HTML form and `process.php`:
1) Parameter reception: Upon submission, the server executes `process.php` and retrieves the `item` and `size` values from the HTTP `$_POST` array.
2) Server-side input validation: The `check_input()` function validates that `size` is numeric (`is_num`) and falls within the inclusive range [34, 46], echoing an error message if validation fails.
3) Database authentication: The script establishes an authenticated connection (`mysqli_connect`) to the local `CatalogDB` database using a username and password.
4) SQL query execution: The server executes a `SELECT` query filtering `CatalogDB` for records matching both the submitted `item` (`it`) and `size` (`sz`).
5) Result rendering and cleanup: If zero rows match, it echoes an out-of-stock message; otherwise, it iterates through the result set (`mysqli_fetch_assoc`) to output the product image, item, size, and price before closing the database connection (`mysqli_close`).

Mentor tip

Trace the full PHP lifecycle: receiving POST variables, validating numeric range [34, 46], connecting to CatalogDB, running the SELECT query, and formatting the output.

(d)(ii)

2 marks

Outline an example of a client-side script that could enhance the form:
1) Script example: A JavaScript function embedded via `<script>` tags that performs immediate client-side validation (verifying the size field is not blank and contains a number between 34 and 46 before submission) or dynamically adjusts UI styling/text magnification.
2) Client-side rationale: Executing validation or UI adjustments directly in the user's browser provides instant feedback and avoids unnecessary round-trip HTTP requests that would waste server bandwidth.

Mentor tip

Provide a concrete client-side JavaScript feature and explain why running it in the browser reduces server load.

Question 10

11 marks

Blogs vs discussion forums, keyword tagging for web crawlers, and search engine ranking

Web science

(a)

4 marks

Distinguish between a blog and a discussion forum with respect to posting and commenting:
1) Posting in a blog: Only the blog owner or authorized author (the scientific journalist) can publish primary articles and initiate new discussion topics.
2) Posting in a forum: Any registered community member can create a new thread and initiate a discussion topic.
3) Commenting in a blog: Readers may submit comments on the journalist's article, but comments are typically moderated by the blog owner to enforce editorial standards and filter out off-topic remarks.
4) Commenting in a forum: Registered participants converse horizontally across threads with minimal pre-moderation, fostering an open peer-to-peer community debate.

Mentor tip

Explicitly contrast both 'posting' (who can start a thread/article) and 'commenting' (editorial moderation vs open peer replies).

(b)

5 marks

Discuss how visible keyword tags on blog articles influence user navigation and web crawler indexing:
1) User navigation: Tags group related articles by theme, allowing readers to click a tag and browse all posts sharing that subject.
2) Internal hyperlink generation: Visible tags function as anchor links connecting pages across the blog's internal architecture.
3) Crawler discovery: When a web crawler visits a blog post, it parses the visible HTML tags and follows their hyperlinks to discover related articles.
4) Preventing orphan pages: This dense internal linking ensures older or deeply nested blog entries are reached and indexed rather than remaining isolated.
5) Semantic classification: The tag keywords provide explicit topical signals that help the search engine categorize the blog's subject matter accurately in its index.

Mentor tip

Connect visible tags to internal hyperlinks, explaining how crawlers follow those links to prevent unindexed orphan pages.

(c)

2 marks

State two reasons why a newly launched discussion forum initially receives a low search engine ranking:
1) Lack of inbound links (backlinks) and low visitor traffic compared to established sites.
2) Unproven domain authority/reputation or absence of syndication feeds (such as RSS) and active user content.

Mentor tip

Focus on PageRank/authority factors such as few external backlinks and low user traffic.

Question 11

18 marks

Assisted living ubiquitous computing, public vs private cloud storage, and healthcare protocols

Web science

(a)(i)

1 mark

Define ubiquitous computing (pervasive computing):
The integration of microprocessors and computational capability into everyday objects and environments so that computing occurs seamlessly anytime and anywhere.

Mentor tip

Emphasize computing embedded invisibly into everyday physical objects.

(a)(ii)

4 marks

Outline two examples of ubiquitous computing in the elderly assisted living project:
1) Wearable biometric smartwatch — Embedded device: A wristwatch equipped with heart-rate and accelerometer sensors. Healthcare use: Continuously monitors vital signs (pulse, blood pressure) and transmits alerts if an arrhythmia is detected.
2) Smart pressure-sensitive flooring — Embedded device: Floor tiles or bathroom mats containing pressure sensors. Healthcare use: Detects if an elderly resident has fallen and remains immobile, automatically alerting emergency medical staff.

Mentor tip

For each example, state both the physical object containing the embedded sensor and its specific healthcare monitoring function.

(b)

5 marks

Discuss the trade-offs between public and private clouds for storing patient healthcare data:
1) Public cloud characteristics: Infrastructure is shared among multiple tenants (multi-tenancy), managed by major commercial providers with high scalability and lower upfront cost.
2) Public cloud security perspective: While commercial providers employ dedicated security experts and advanced patching, shared infrastructure increases exposure if a multi-tenant breach occurs.
3) Private cloud characteristics: Dedicated infrastructure operated exclusively for a single healthcare organization, offering granular control over physical and logical access.
4) Private cloud security perspective: Isolating patient records on dedicated hardware reduces external attack surfaces, though the healthcare provider must fund and maintain its own specialist IT security staff.
5) Conclusion: Because patient medical telemetry is highly confidential and legally protected, a private cloud (combined with end-to-end encryption) is generally preferred to guarantee strict data sovereignty and privacy.

Mentor tip

Address both public and private clouds in detail and finish with a justified conclusion tailored to sensitive medical records.

(c)

2 marks

Outline one risk posed to elderly residents by relying on assisted living technologies:
1) Hardware or network failure: If a wearable sensor malfunctions or the home Wi-Fi drops, a medical emergency (such as a fall) may go unreported while caregivers assume the resident is safe.
2) Alternative: Loss of personal privacy/autonomy (continuous surveillance of daily habits) or forgetting to wear/charge a biometric device.

Mentor tip

Focus on a direct risk to the elderly person (such as undetected emergencies during device failure or loss of domestic privacy) rather than generic cloud storage security.

(d)

1 mark

State a wireless or mobile communication protocol:
Wi-Fi (IEEE 802.11), Bluetooth, 4G/LTE, 3G, GSM, or WiMAX.

Mentor tip

Name any standard wireless or cellular transmission protocol.

(e)

5 marks

Explain the importance of standardized protocols in the assisted living healthcare ecosystem:
1) Interoperability across heterogeneous devices: Assisted living combines wearables, floor sensors, mobile apps, and cloud servers from different manufacturers; open standards ensure these subsystems communicate seamlessly.
2) Secure data transmission: Cryptographic transport protocols (such as TLS/SSL) encrypt vital signs in transit from home sensors to the cloud, preventing eavesdropping or man-in-the-middle tampering.
3) Data integrity and reliability: Packet error-checking protocols guarantee that critical medical alerts arrive intact and uncorrupted so clinicians do not make decisions on faulty telemetry.
4) Future extensibility and backward compatibility: Adhering to industry standards allows the project to integrate newer medical sensors in the future without replacing the entire existing infrastructure.
5) Regulatory compliance: Standardized access control and logging protocols help healthcare providers audit data access and protect patient safety.

Mentor tip

Cover three dimensions: device interoperability/extensibility, encrypted transmission security (TLS), and data integrity for patient safety.

Question 12

20 marks

Semantic web vs traditional web, web graphs, audio search engines, P2P power laws, and collective intelligence

Web science

(a)

3 marks

State three features that distinguish the Semantic Web from the traditional World Wide Web:
1) Machine-readable meaning: The Semantic Web structures data so software agents can automatically interpret meaning and relationships, whereas the traditional web is designed for human reading.
2) Data interchange vs document interchange: The Semantic Web links granular data resources using URIs, RDF, and ontologies (OWL), whereas the traditional web links HTML documents via URLs.
3) Automated querying and inference: Users and agents can execute logical queries (e.g., SPARQL) to extract synthesized knowledge across sources, rather than merely browsing pages via keyword matching.

Mentor tip

Contrast HTML document browsing for humans with RDF/OWL structured data and automated machine reasoning.

(b)

2 marks

Outline the difference between a web graph and a sub-graph with an example:
1) Structural distinction: The web graph is a directed graph modelling the entire World Wide Web (where vertices represent web pages/URLs and directed edges represent hyperlinks between them), whereas a sub-graph is a mathematical subset of those vertices and directed edges.
2) Concrete example: The complete World Wide Web forms the full web graph, while all pages and internal links within the Wikipedia domain (or an university portal) form a sub-graph.

Mentor tip

Mention that edges are directed hyperlinks and provide a clear domain-level example of a sub-graph.

(c)

2 marks

Define the multimedia-web:
1) An evolution of the web specialized in indexing, delivering, and interacting with rich audiovisual media (audio, video, images) rather than plain text.
2) Enables non-textual search queries where users can search using multimedia inputs (such as an audio clip or image) to retrieve matching media.

Mentor tip

Highlight both multimedia content delivery and querying via non-text media inputs.

(d)

4 marks

Explain how an audio-input music search engine identifies a song:
1) Audio capture: The system records an incoming soundwave (or MIDI stream) hummed or played by the user.
2) Feature extraction: Signal processing algorithms convert the raw audio waveform into a digital acoustic fingerprint, extracting physical characteristics such as pitch contour, melody, rhythm, beat, or harmonic frequencies.
3) Database matching: The extracted acoustic signature is compared against a repository of indexed audio fingerprints in a music archive.
4) Ranked retrieval: The engine returns the closest matching tracks or audio excerpts ranked by similarity score.

Mentor tip

Focus on acoustic feature extraction (melody, pitch, rhythm, spectrogram fingerprinting) from soundwaves rather than typing song lyrics.

(e)

5 marks

Discuss how power laws (Zipf's law) explain why identically sized movies download at vastly different speeds on a P2P network:
1) Power-law popularity distribution: According to Zipf's law, movie download frequency is inversely proportional to its popularity rank (a few blockbuster movies account for the vast majority of downloads, while niche films lie in the long tail).
2) Distributed peer caching: In a P2P network, peers cache fragments or complete copies of files they download and serve them to other peers.
3) Fast parallel downloads for popular movies: A top-ranked movie is cached across thousands of active peer nodes simultaneously, allowing a client to download multiple chunks in parallel from nearby peers at high speed.
4) Scarcity of niche movies: A rarely requested movie in the long tail is held by very few (or zero) active peers in the P2P swarm.
5) Bottlenecks for rare files: Locating a rare movie takes longer, and downloading from a single distant peer or fallback repository server creates a bandwidth bottleneck, greatly increasing total download time.

f (r) \propto \frac{1}{r}

Mentor tip

Link Zipf's law (popularity rank) to the number of peers caching file fragments and the ability to download chunks in parallel.

(f)

4 marks

Explain how collective intelligence shapes the online purchasing habits of the general public:
1) Concept definition: Collective intelligence emerges when large groups of individuals share knowledge, ratings, and feedback online, producing aggregated insights superior to individual judgment.
2) Concrete e-commerce example: Online shoppers contributing verified product reviews, star ratings, and folksonomy tags on marketplaces, or backing niche creators via crowdfunding platforms.
3) Influence on consumer trust and trends: Aggregated user reviews and collaborative filtering recommendations guide prospective buyers toward high-rated items and steer them away from poor products.
4) Market and economic impact: Niche ethical or interest communities (such as zero-waste groups or indie gaming communities) can pool demand to create viable markets, drive economies of scale, and pressure vendors to alter pricing or product policies.

Mentor tip

Define collective intelligence, give a specific online shopping example (like crowdsourced reviews or crowdfunding), and explain how it alters buyer decisions and market trends.

Option D: Object-oriented programming · Option

Questions 13–16 · 65 marks

Question 13

10 marks

OOP class vs instantiation, inheritance hierarchies, and software libraries

Object-oriented programming

(a)

3 marks

Distinguish between a class and an instantiation of a class using the company scenario:
1) Class definition: A class is an abstract blueprint or template that defines the attributes (variables) and behaviours (methods) common to a category of objects.
2) Instantiation definition: Instantiation is the runtime creation of a concrete object in memory with its own specific attribute values based on the class blueprint.
3) Scenario example: `Employee` (or `SalesPerson`) is a class defining properties like `id` and `salary`, whereas creating a specific record for employee `'John Smith'` with `id = "101"` is an instantiation of that class.

Mentor tip

Contrast the abstract blueprint (class) with the concrete runtime object in memory (instantiation) and ground both in the company scenario.

(b)

4 marks

Explain two ways inheritance can be incorporated into the company's administration software:
1) GUI module hierarchy: Define a `BaseGUI` superclass containing standard window components (menus, company branding, login bar), and derive specialized subclasses (`SalesGUI`, `PayrollGUI`) that inherit the common layout while adding module-specific controls.
2) Staff hierarchy: Define an `Employee` superclass storing shared attributes (`id`, `name`, `baseSalary`), and create subclasses (`Manager`, `OfficeStaff`, `SalesPerson`) that inherit these fields while adding role-specific attributes like `commissionRate` or `bonus`.

Mentor tip

Provide two distinct superclass/subclass examples from the prompt (one for the GUI modules and one for the staff roles).

(c)

3 marks

Describe how pre-written code libraries facilitate software development:
1) Reduced development time: Developers can import standard, pre-built classes (such as GUI widgets, sorting routines, or `LinkedList` collections) rather than coding complex algorithms from scratch.
2) High reliability: Library routines have already been rigorously tested and debugged by the wider community, reducing software defects.
3) Modularity and team focus: Programmers can focus on the company's core business logic while relying on standardized library interfaces for low-level tasks.

Mentor tip

Highlight time saved, pre-tested reliability (fewer bugs), and reuse of standard functionality.

Question 14

35 marks

SalesPerson and Sales classes: constructors, encapsulation, UML, array algorithms, and polymorphism

Object-oriented programming

(a)

2 marks

Complete the `SalesPerson(String id)` constructor by assigning the ID parameter and instantiating the `salesHistory` array:
```java
public SalesPerson(String id) {
this.id = id;
this.salesHistory = new Sales[100]; // any reasonable array capacity
}
```

Mentor tip

Initialize both `this.id` and the `salesHistory` array (`count` is already initialized to 0 at declaration).

(b)

3 marks

Explain why accessor (getter) methods are required in `SalesPerson`:
1) Private encapsulation: The instance variables (`id`, `salesHistory`, `count`) are declared with the `private` access modifier.
2) Restricted direct access: External classes (such as `Main` or `Payroll`) are prevented from reading or corrupting these fields directly.
3) Controlled read interface: `public` accessor methods (`getId()`, `getCount()`) provide a safe, read-only mechanism for external classes to retrieve attribute values.

Mentor tip

Use exact OOP terminology: `private` instance variables, encapsulation, and `public` read access from outside the class.

(c)(i)

4 marks

Construct the UML class diagram showing the relationship between `SalesPerson` and `Sales`:
1) `SalesPerson` 3-compartment box:
- Attributes: `- id : String`, `- salesHistory : Sales[]`, `- count : int`
- Methods: `+ calcTotalSales() : double`, `+ largestSale() : Sales`
2) `Sales` 3-compartment box:
- Attributes: `- itemId : String`, `- value : double`, `- quantity : int`
- Methods: (accessors/constructors optional per prompt)
3) Relationship arrow: Draw a `has-a` aggregation/composition arrow (or 1-to-many `1..*` association line) pointing from `SalesPerson` to `Sales`.

Mentor tip

Include visibility symbols (`-` for private, `+` for public), data types for all attributes/methods, and a clear 'has-a' relationship link.

(c)(ii)

2 marks

Outline a negative consequence of modifying the `Sales` class design later:
1) Tight coupling / ripple effects: Because `SalesPerson` depends directly on `Sales` methods, altering method signatures or attribute types in `Sales` can break existing code in `SalesPerson`.
2) Concrete example: If `getValue()` or `getQuantity()` in `Sales` is renamed or changed, `calcTotalSales()` and `largestSale()` in `SalesPerson` will fail to compile and must be rewritten.

Mentor tip

Give a specific example showing how changing a method in `Sales` breaks `calcTotalSales()` in `SalesPerson`.

(d)

4 marks

Trace the execution of the `Main` code snippet line by line:
1) `salesPeople[2].getId()` -> Outputs the ID of the third salesperson: `102`
2) `salesPeople[0].getCount()` -> Two `Sales` objects were added to `salesPeople[0]`: `2`
3) `salesPeople[1].getSalesHistory(0).getValue()` -> Unit value of the first sale for `salesPeople[1]`: `2550.40` (or `2550.4`)
4) `salesPeople[0].calcTotalSales()` -> `(300.00 * 10) + (1000.00 * 2) = 3000.00 + 2000.00`: `5000.00` (or `5000.0` / `5000`)

300.00 \times 10 + 1000.00 \times 2 = 5000.00

Mentor tip

Write each of the four printed values clearly on its own line.

(e)

5 marks

Construct the `calcTotalSales()` method inside `SalesPerson`:
```java
public double calcTotalSales() {
double total = 0.0;
for (int i = 0; i < count; i++) {
total = total + (salesHistory[i].getValue() * salesHistory[i].getQuantity());
}
return total;
}
```

Mentor tip

Loop only up to `count` (the number of initialized sales), multiply `getValue()` by `getQuantity()` for each sale, and return the accumulated `total`.

(f)

5 marks

Construct the `highest()` method to return the ID of the salesperson with the maximum total sales across 100 `salesPeople`:
```java
public String highest() {
double maxTotal = -1.0;
int bestIndex = 0;
for (int i = 0; i < 100; i++) {
if (salesPeople[i].calcTotalSales() > maxTotal) {
maxTotal = salesPeople[i].calcTotalSales();
bestIndex = i;
}
}
return salesPeople[bestIndex].getId();
}
```

Mentor tip

Reuse `calcTotalSales()` and `getId()` via dot notation on each element of `salesPeople`.

(g)

4 marks

Construct `addSales(Sales s, String id)` in `Main` to append sale `s` to the matching salesperson:
```java
public void addSales(Sales s, String id) {
for (int i = 0; i < salesPeople.length; i++) {
if (salesPeople[i] != null && salesPeople[i].getId().equals(id)) {
salesPeople[i].setSalesHistory(s);
break;
}
}
}
```

Mentor tip

Search the `salesPeople` array for the matching `getId()`, then invoke `setSalesHistory(s)` on that object.

(h)

3 marks

Suggest code modifications needed so `Payroll` can calculate monthly commissions:
1) Add a `date` (or `month`) instance variable to the `Sales` class and include it in the `Sales` constructor.
2) Add a public `getDate()` (or `getMonth()`) accessor method in the `Sales` class.
3) Overload or update `calcTotalSales(int targetMonth)` in `SalesPerson` to filter and sum only `Sales` objects whose date matches the requested month.

Mentor tip

State the new attribute in `Sales`, its getter/constructor update, and the filtering adjustment in `calcTotalSales()`.

(i)

3 marks

Discuss how polymorphism is demonstrated in the provided classes:
1) Constructor overloading: The `SalesPerson` class defines two constructors sharing the same name: `SalesPerson(String id)` and `SalesPerson(String id, Sales[] s, int c)`.
2) Compile-time resolution: Having multiple methods/constructors with the same identifier but different parameter lists is static polymorphism (overloading).
3) Parameter matching: When a `new SalesPerson(...)` object is instantiated, the compiler selects which constructor to execute based on the number and types of arguments passed.

Mentor tip

Identify the overloaded constructors in `SalesPerson` and explain how the compiler dispatches the call based on argument signatures.

Question 15

7 marks

Replacing arrays with LinkedList<Sales> collections and finding the largest sale

Object-oriented programming

(a)

2 marks

Write the attribute declaration replacing `private Sales[] salesHistory;` with a generic `LinkedList`:
```java
private LinkedList<Sales> salesHistory = new LinkedList<Sales>();
```

Mentor tip

Include both the `LinkedList` class type and the generic `<Sales>` type parameter.

(b)

5 marks

Construct `largestSale()` in `SalesPerson` using `LinkedList` methods to return the `Sales` object with the highest total value:
```java
public Sales largestSale() {
double maxVal = -1.0;
Sales bestSale = null;
for (int i = 0; i < salesHistory.size(); i++) {
Sales current = salesHistory.get(i);
double saleTotal = current.getValue() * current.getQuantity();
if (saleTotal > maxVal) {
maxVal = saleTotal;
bestSale = current;
}
}
return bestSale;
}
```

Mentor tip

Use `salesHistory.size()` and `salesHistory.get(i)` (or an `Iterator<Sales>`) to traverse the list, comparing `getValue() * getQuantity()` and returning the `Sales` object.

Question 16

13 marks

Binary search tree Node references, post-order recursive trace, and finding the minimum value

Object-oriented programming

(a)

3 marks

Describe how object references function as a data type in the `Node` class:
1) Self-referential fields: The variables `left` and `right` are declared of reference type `Node`.
2) Memory address pointers: Rather than holding raw primitive integers, `left` and `right` store the memory addresses (references) pointing to other instantiated `Node` objects (the left and right children).
3) Null termination: When a node has no child on a given branch, its reference variable holds `null`.

Mentor tip

Identify `left` and `right` as reference variables that store memory pointers to other `Node` instances or `null`.

(b)

5 marks

Trace the post-order recursive execution of `print(root)` on the tree (`node1` = 33, `node2` = 12, `node3` = 14, `node4` = 50):
1) `print(node1 [33])` calls `print(node1.left)` -> enters `print(node2 [12])`.
2) `print(node2 [12])` calls `print(node2.left)` -> `null` (base case returns immediately).
3) `print(node2 [12])` calls `print(node2.right)` -> enters `print(node3 [14])`.
4) `print(node3 [14])` calls `print(node3.left)` -> `null` (returns) and `print(node3.right)` -> `null` (returns), then outputs `14`.
5) Control returns to `node2 [12]`, which completes its right call and outputs `12`.
6) Control returns to `node1 [33]`, which calls `print(node1.right)` -> enters `print(node4 [50])`.
7) `print(node4 [50])` calls `print(node4.left)` -> `null` (returns) and `print(node4.right)` -> `null` (returns), then outputs `50`.
8) Control returns to `node1 [33]`, which outputs `33`.
Final output sequence: `14 12 50 33`

Mentor tip

Clearly show each recursive call level, the `null` base case returns, and the exact post-order output sequence: 14, 12, 50, 33.

(c)

5 marks

Construct the `smallest(Node node)` method to return the minimum integer in a non-empty Binary Search Tree (located at the leftmost node):
Iterative solution:
```java
public int smallest(Node node) {
Node current = node;
while (current.left != null) {
current = current.left;
}
return current.data;
}
```
Alternative recursive solution:
```java
public int smallest(Node node) {
if (node.left == null) {
return node.data;
}
return smallest(node.left);
}
```

Mentor tip

In a binary search tree, smaller values are always placed in the left subtree, so traverse `.left` references until `left == null`.

Our analysis

Where the marks are

Each part sorted by the topic it mostly tests. Databases carries the most marks on this paper.

Databases65 marks · 25%
Q1, Q2, Q3, Q4
Modelling and simulation65 marks · 25%
Q5, Q6, Q7, Q8
Web science65 marks · 25%
Q9, Q10, Q11, Q12
Object-oriented programming65 marks · 25%
Q13, Q14, Q15, Q16

Mentor insights

Traps to watch for

01
Databases: Telecom SMS data structures, normalization to 3NF, SQL querying, · Q3
Point out that relational tables cannot dynamically add columns as a subscriber sends more text messages.
02
Modelling and simulation: Natural language processing (NLP), neural network training, fitne · Q8
Develop two distinct points: one addressing the sheer volume of data and the second addressing the complexity of multi-group classification.
03
Web science: Semantic web vs traditional web, web graphs, audio search engines · Q12
Contrast HTML document browsing for humans with RDF/OWL structured data and automated machine reasoning.
04
Object-oriented programming: SalesPerson and Sales classes: constructors, encapsulation, UML, · Q14
Initialize both `this.id` and the `salesHistory` array (`count` is already initialized to 0 at declaration).

Solutions and tips are GeorgeTheTutor's own; the questions are not reproduced here. Not affiliated with or endorsed by the International Baccalaureate Organization.

Want someone to go through your paper with you?

Our maths mentors took these exams recently. One-to-one, they'll find where your marks are going and fix it before May.

Find your maths mentor More past papers