Loading…

Worked solutions

IB Computer Science HL · Paper 2

Computer Science, Higher Level

Session: November 2014
Marks: 260
Duration: 1 hour 20 min
Calculator: Not allowed
Questions: 17

Q1 Q2 Q3 Q4 Q5 Q6 Q7 Q8 Q9 Q10 Q11 Q12 Q13 Q14 Q15 Q16 Q17

Keep your copy of the paper open alongside. We don't reproduce the questions: each part below is labelled to match the November 2014 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

16 marks

Banking database transactions, rollback recovery, validation vs verification, and ACID isolation

Databases

(a)(i)

3 marks

An operation is a single, elementary action performed on a database (such as a single read or write command).
A transaction is a logical unit of work comprising a sequence of multiple operations that must all succeed together to preserve database consistency.
Example: A cash withdrawal transaction combines an initial query operation to check the current balance followed by an update operation to deduct the withdrawn sum.

Mentor tip

Contrast a single atomic database action (operation) with a multi-step logical unit of work (transaction) using a banking example.

(a)(ii)

1 mark

A transaction must be rolled back if any constituent operation fails before completion (e.g., network interruption, hardware failure, or insufficient account funds).

Mentor tip

Any condition preventing the full sequence of operations from completing successfully triggers a rollback.

(a)(iii)

1 mark

Restores the database to its exact consistent state prior to the start of the transaction, undoing any partial modifications.

Mentor tip

State clearly that the database is returned to its pre-transaction state.

(b)

4 marks

1) Built-in transaction and concurrency management: A database engine enforces ACID guarantees and record locking out of the box, ensuring deposits and withdrawals do not corrupt balances.
2) Standardized querying and reduced redundancy: Centralizing customer and account records eliminates duplicate data entry and provides fast, standardized retrieval and sorting across multiple banking workflows.

Mentor tip

Provide two distinct reasons (e.g., transaction integrity and efficient querying/minimal redundancy) with an elaboration for each.

(c)

4 marks

Data Validation:
- Definition: Automated check ensuring entered input conforms to expected data type, range, and format rules.
- Banking Example: Checking that the deposit amount field contains a positive numeric currency value rather than alphabetic characters or a negative number.
Data Verification:
- Definition: Confirming that the data entered into the system accurately matches the original source value intended by the user.
- Banking Example: Prompting the teller or customer to enter the deposit amount twice (double entry) or displaying a confirmation screen to verify the figure before committing.

Mentor tip

Remember: validation checks 'is it sensible/valid format?' while verification checks 'was it copied/typed accurately as intended?'

(d)

3 marks

Isolation ensures that multiple transactions executing concurrently leave the database in the exact same state as if they had been executed sequentially one after another.
Without isolation, interleaved read/write steps can cause a lost update anomaly.
Banking Example: If a $10 d e p os i t an d a$ 50 withdrawal occur simultaneously on an account, isolation locks the balance so both updates apply cleanly in sequence, guaranteeing a net change of -$40 rather than one transaction overwriting the other.

Mentor tip

Define isolation in terms of sequential equivalence and illustrate with simultaneous deposit/withdrawal amounts.

Question 2

15 marks

Online hat store DBMS: attribute data types, core DBMS tools, security, data dictionary, and modelling

Databases

(a)

3 marks

1) Hat size: Integer (or Float/Numeric, or single Character/String for S/M/L).
2) Price: Float / Real / Decimal / Currency.
3) Maker's name: String / Text / Varchar.

Mentor tip

Assign a distinct, appropriate primitive/SQL data type to each of the three attributes.

(b)

3 marks

1) Database creation / Data Definition (defining tables, schemas, and constraints).
2) Database manipulation (inserting, updating, and deleting records).
3) Database interrogation / querying (retrieving and filtering data via queries and reports).

Mentor tip

Think of DDL (creation), DML (manipulation), and DQL (interrogation/querying).

(c)

4 marks

1) Role-Based Access Control (Access Rights): Restricts privileges by user role so online shoppers can only view hat catalogs and their own orders, while only authenticated store administrators can alter prices or inventory.
2) Record Locking / Encryption and Audit Logging: Locks records during concurrent checkout updates to prevent stock corruption, and encrypts stored customer payment/personal details against unauthorized breach.

Mentor tip

Name two concrete DBMS security mechanisms (such as access rights/permissions and locking/encryption) and explain how each protects the online shop.

(d)

2 marks

A data dictionary is a centralized metadata repository that defines the structure and properties of all data elements in the database.
It specifies field names, data types, format constraints, semantic meanings, and inter-table relationships.

Mentor tip

Define the data dictionary as 'data about data' (metadata) detailing formats, meanings, and relationships.

(e)

3 marks

Provides a formal conceptual blueprint prior to physical implementation.
Ensures all required business entities, attributes, and relationships are accurately captured without omissions.
Offers a clear visual representation that developers and store owners can review and validate together before coding.

Mentor tip

Highlight its role as a design blueprint, completeness check, and communication bridge with stakeholders.

Question 3

14 marks

Bookstore database normalization, redundancy issues, ERD construction, and reader privacy

Databases

(a)

2 marks

A normalized database organizes attributes into separate related tables to minimize or eliminate data redundancy and undesirable functional dependencies.
Consequently, updating or modifying a specific fact only requires changing a single record in one table.

Mentor tip

Mention both eliminating redundancy/dependencies and single-point updates.

(b)

2 marks

1) Wastes storage capacity by repeating identical attribute values across multiple rows.
2) Introduces update anomalies and data inconsistency if a repeated value is updated in one location but overlooked in another.

Mentor tip

Focus on wasted disk space, extra maintenance overhead, and the risk of inconsistent records.

(c)

3 marks

Define separate entity boxes reflecting the four specification points:
1) BOOK (Book_Title, Author, Publication_Date, List_Price, Number_of_Copies)
2) BOOKCASE (Bookcase_Number, Aisle_Number, Shelf*)
3) SHELF (Shelf_ID, Book_Title*)
4) SALES_LOG (Book_Title, Customer_Name, Price_Paid, Sale_Date)
Draw relationship links with cardinalities: BOOKCASE (1) ----< contains >---- (0..N) SHELF, SHELF (1) ----< holds >---- (0..N) BOOK, and BOOK (1) ----< recorded in >---- (0..N) SALES_LOG.

Mentor tip

Include the four entities with their listed attributes and connect them with clear 1-to-many relationship lines.

(d)(i)

1 mark

The tables lack explicit unique Primary Keys, or the 'Shelf' attribute in Bookcase / 'Book' attribute in Shelf contains non-atomic repeating groups of values.

Mentor tip

Point out either missing primary keys or non-atomic multi-valued columns (e.g., multiple books listed inside a single shelf record).

(d)(ii)

3 marks

1) Assign atomic Primary Keys to each entity (e.g., ISBN/Book_ID for Book, Customer_ID for a separate Customer table, Transaction_ID for Sales_Log).
2) Decompose repeating shelving groups into a dedicated associative/junction table (e.g., SHELVING with Book_ID, Shelf_Number, Bookcase_Number) so every field is atomic.
3) Extract Customer details (Customer_ID, Customer_Name) into a separate CUSTOMER table so repeat buyers are not duplicated across Sales_Log rows.

Mentor tip

Explain how to make fields atomic via a junction table and eliminate transitive/partial dependencies using separate tables with primary/foreign keys.

(e)

3 marks

Issue Identification: Disclosure of personal reading and purchasing histories in the Sales_Log without individual consent or a specific warrant constitutes a breach of civil privacy.
Discussion: Law enforcement might profile citizens based on controversial or political book titles they purchased, yet purchasing a book does not reliably indicate malicious intent (e.g., academic research or gifts), potentially leading to unjust suspicion or chilling effects on free inquiry.

Mentor tip

Identify the privacy violation clearly, then discuss why profiling reading habits can be both intrusive and misleading.

Question 4

20 marks

Global retail data warehouse: ETL pipelines, association vs classification, deviation detection, and targeted marketing ethics

Databases

(a)

2 marks

Consolidates historical data from global stores, factories, and logistics into a single repository to enable longitudinal trend analysis.
Example: Analysts can compare multi-year seasonal sales across regions to optimize inventory forecasting.

Mentor tip

State the core advantage (integrated historical trend analysis) and give a concrete retail example.

(b)

3 marks

1) Extract: Pull raw operational data from disparate heterogeneous sources (global stores, Chinese factories, logistics partners).
2) Transform: Cleanse, validate, standardize formats/currencies, and deduplicate the extracted data to match the warehouse schema.
3) Load: Write and time-stamp the transformed data into the target data warehouse fact and dimension tables.

Mentor tip

Define all three stages of ETL—Extract, Transform, and Load—in sequence.

(c)

4 marks

1) Manufacturing Data Source: Chinese plants record production labor costs in Chinese Yuan (CNY) and dates as YYYY/MM/DD.
2) Transport Provider Source: Logistics carriers record shipping invoices in US Dollars (USD) and weights in pounds (lbs).
3) Warehouse Standard: The central data warehouse requires all financial figures in Euros (EUR) and weights in kilograms (kg).
4) Transformation Action: During ETL, the transformation rules convert CNY and USD values into Euros using historical exchange rates and convert pounds to kilograms before loading.

Mentor tip

Provide a concrete standardization example (such as currency conversion or unit/date formatting) linking manufacturing data, shipping data, and the unified warehouse format.

(d)

4 marks

Association Rule Mining:
- Discovers co-occurrence relationships ('if-then' rules) between items within the same transaction.
- Example: Discovering that customers who purchase baby formula also frequently purchase diapers in the same basket.
Classification:
- Assigns records into predefined categorical classes based on historical attribute profiles.
- Example: Categorizing a customer into a predefined loyalty tier ('High-Value', 'At-Risk', 'Budget') based on annual spend and visit frequency.

Mentor tip

Contrast item-to-item co-occurrence rules (Association) against sorting records into predefined groups (Classification) with a retail example of each.

(e)

2 marks

Deviation detection (outlier/anomaly analysis) identifies records that diverge significantly from established norms or expected statistical distributions.
Retail Example: Flagging a sudden fraudulent bulk credit-card purchase or a shipping container whose transit time took three times longer than average.

Mentor tip

Define deviation detection as finding statistical outliers/anomalies and give a clear retail or logistics example.

(f)

2 marks

1) Association rule mining (or Cluster analysis).
2) Classification (or Sequential pattern mining).

Mentor tip

Any two standard data mining techniques—Association, Classification, or Cluster Analysis—earn full marks.

(g)

3 marks

Ethical Issue: Inferring sensitive personal health conditions (pregnancy) from routine shopping habits without consent violates medical and personal privacy.
Contextual Example: Sending baby-related coupons to a household mailbox can inadvertently disclose a pregnancy to family members, roommates, or partners before the customer chooses to share the news.
Broader Discussion: Even if individual product purchases are benign, predictive analytics creates sensitive health profiles outside healthcare privacy laws, risking emotional distress if the prediction is wrong or premature.

Mentor tip

Discuss how predictive profiling bypasses medical privacy norms and can inadvertently reveal intimate life events to other household members.

Option B: Modelling and simulation · Option

Questions 5–8 · 65 marks

Question 5

15 marks

Patient antibiotic dosage mathematical model, scheduling algorithm, and bloodstream retention simulation

Modelling and simulation

(a)

1 mark

A 20-year-old patient falls into the 18–40 age bracket (1.50 mg per kg). Multiply weight by rate:
Dose = 50 kg * 1.50 mg/kg = 75 mg.

Dose = 50 \times 1.50 = 75 mg

Mentor tip

Match age 20 to the 18–40 row (1.50 mg/kg) and multiply by 50 kg.

(b)

2 marks

Variables: patient age (integer), patient weight (real/float), rate per kg (real/float), and total dose amount (real/float).
Calculation: Determine rate_per_kg from age bracket, then compute Dose = weight * rate_per_kg.

Dose = weight \times rate (age)

Mentor tip

List the input/output variables clearly and state the multiplication formula.

(c)

4 marks

Input patient weight and age, then compute the single-dose amount using the age-bracket lookup.
Use an outer loop running day from 1 to 10, and an inner loop running dose index from 0 to 2 (representing 8:00, 16:00, and 24:00/0:00):
Pseudocode:
input WEIGHT, AGE
if AGE <= 40 then RATE = 1.50
else if AGE <= 60 then RATE = 1.75
else RATE = 2.00
end if
DOSE = WEIGHT * RATE
loop DAY from 1 to 10
loop PERIOD from 0 to 2
TIME = 8 + PERIOD * 8
output "day" + DAY, TIME + ":00", DOSE
end loop
end loop

Mentor tip

Include input/dose calculation, an outer loop for 10 days, and an inner loop generating the three 8-hour intervals (08:00, 16:00, 24:00).

(d)(i)

4 marks

1) In a spreadsheet, create columns for Dose Number (1 to 30, or Day/Time), Dose Administered (Column B, constant D), and Bloodstream Medicine Level (Column C).
2) In Row 2 (first dose at day1 08:00), set the initial bloodstream amount equal to the administered dose: C2 = B2.
3) In Row 3 (second dose), enter the recurrence formula adding the new dose to 20% of the previous bloodstream level: C3 = B3 + 0.20 * C2.
4) Replicate (fill down) the formula in Column C down to Dose 30 to model accumulation over the 10-day course.

BloodLevel_{k} = Dose + 0.20 \times BloodLevel_{k - 1}

Mentor tip

Show both the spreadsheet column setup and the recurrence formula (New Level = Dose + 0.2 * Previous Level).

(d)(ii)

1 mark

Mistyping or mismeasuring the patient's weight or age (e.g., entering an out-of-range age or wrong unit) causes the model to compute an incorrect dosage (underdosing or overdosing the patient).

Mentor tip

Link an input error in age or weight directly to an incorrect calculated dosage.

(d)(iii)

3 marks

Take clinical blood samples from patients immediately after each scheduled dose to measure actual antibiotic concentrations in vivo.
Compare these empirical blood test measurements against the values predicted by the spreadsheet model.
Large discrepancies reveal either input errors or oversimplified absorption assumptions (such as the fixed 20% retention rate).

Mentor tip

Validate a physiological simulation by comparing predicted values against real-world empirical blood measurements.

Question 6

15 marks

Town population growth simulation: demographic factors, model accuracy, and urban planning consequences

Modelling and simulation

(a)

1 mark

Initialize population P_0 = 25,000, and for each year t from 1 to 10, multiply the previous year's population by 1.07 (P_t = P_{t-1} * 1.07).

P_{t} = P_{t - 1} \times 1.07

Mentor tip

State the initial population input and the annual 1.07 multiplication factor.

(b)

3 marks

1) Birth rate (annual live births per capita).
2) Death rate (annual mortality per capita).
3) Net migration: Immigration (new residents moving in) and Emigration (residents moving away).

Mentor tip

The classic demographic variables are birth rate, death rate, immigration, and emigration.

(c)

3 marks

Replace the fixed 7% multiplier with a composite annual rate derived from the three demographic factors:
1) Estimate dynamic annual birth and death rates using historical civil registry trends.
2) Incorporate projected immigration and emigration figures based on local housing developments, new factory openings, or business closures.
3) Compute each year's population as: P_t = P_{t-1} + Births_t - Deaths_t + Immigration_t - Emigration_t.

P_{t} = P_{t - 1} + B_{t} - D_{t} + I_{t} - E_{t}

Mentor tip

Explain how each of the three factors is estimated and combined into the annual population update equation.

(d)

4 marks

1) Omitted macroeconomic variables: The simulation ignores external drivers such as national economic recessions, interest rates, or employment shifts that heavily influence migration.
2) Non-linear 'new town' dynamics: Extrapolating a 10-year horizon from a single year's 7% surge is unreliable because newly established towns often experience an initial rapid influx that quickly plateaus once housing capacity fills.

Mentor tip

Discuss two distinct sources of inaccuracy: unmodelled external factors and the danger of extrapolating short-term new-town surges over 10 years.

(e)

4 marks

1) Consequence of Under-prediction (Actual > Predicted): Infrastructure such as schools, hospitals, water supply, and road networks will be under-built, resulting in overcrowded classrooms and strained healthcare services.
2) Consequence of Over-prediction (Actual < Predicted): The municipality over-invests public funds into unneeded facilities and housing, leaving the town with vacant buildings and severe budget deficits.

Mentor tip

Contrast the social impact of under-estimating (shortage of schools/hospitals) vs over-estimating (wasted tax revenue and empty infrastructure).

Question 7

15 marks

Automotive CAD wireframe representation, geometric manipulation, and 3D rendering techniques

Modelling and simulation

(a)

2 marks

A wireframe model represents the 3D structural skeleton of the car as a collection of vertices (3D coordinate points) connected by edges (lines/vectors) and curves.
These mathematical coordinates are stored in memory and projected onto the 2D screen without solid surfaces or shading.

Mentor tip

Mention vertices (points) connected by lines/vectors representing the 3D skeletal structure.

(b)(i)

1 mark

Altering the structural body profile, such as extending the chassis length, lowering the roofline, or converting a 3-door hatchback into a 5-door model.

Mentor tip

Choose a structural/geometric shape modification rather than surface color or texture.

(b)(ii)

2 marks

The stored (x, y, z) coordinates of the relevant vertices are translated or scaled.
The connecting vectors automatically recalculate their lengths and angles, reshaping the wireframe geometry.

Mentor tip

Explain that modifying vertex coordinates updates the connecting vectors/lines.

(b)(iii)

2 marks

The designer selects and drags vertices/control points directly on screen using a mouse, stylus/graphics tablet, or touch display.
Alternatively, the designer types exact numerical coordinate dimensions or scaling factors into the CAD interface.

Mentor tip

Mention both direct graphical manipulation (mouse/stylus drag) and numeric coordinate entry.

(c)

2 marks

1) Exterior: Testing different metallic paint colors, surface finishes, or headlight/grille styling.
2) Interior: Changing seat upholstery fabric/leather textures or dashboard trim colors.

Mentor tip

Give one aesthetic/surface change for the exterior and one for the interior.

(d)

6 marks

1) Polygon Mesh / Surface Skinning: The open spaces between wireframe edges are filled with polygon surfaces to form a solid 3D shell, hiding occluded back-facing lines (hidden surface removal).
2) Texture Mapping and Colouring: Surface colours and material textures (metallic paint, glass, fabric) are mapped onto the polygon faces.
3) Lighting and Shading / Ray Tracing: Virtual light sources are positioned, and algorithms like ray tracing or Phong shading calculate reflections, specular highlights, and shadows across curved body panels.
4) Continuous Re-rendering: Whenever the designer rotates or zooms the camera view, the GPU recomputes the 3D-to-2D raster projection in real time.

Mentor tip

Cover surface filling, colour/texture mapping, ray tracing/lighting calculations, and re-rendering upon rotation.

Question 8

20 marks

Chatbot machine learning paradigms and cognitive care robots for the elderly

Modelling and simulation

(a)(i)

2 marks

A chatbot is an interactive computer program designed to simulate human conversation.
It parses natural language input (text or speech) from a user and generates contextually appropriate text or spoken responses.

Mentor tip

Mention both receiving human language input and generating conversational language output.

(a)(ii)

1 mark

Supervised learning trains a model using labeled data consisting of known input queries paired with verified target responses.

Mentor tip

Emphasize training on labeled input-output pairs where the desired response is known in advance.

(a)(iii)

3 marks

Unsupervised learning analyzes unlabeled data to discover hidden statistical patterns and clusters without predefined target outputs.
Comparison: Unlike supervised learning—which maps inputs to a curated set of known answers—unsupervised learning self-organizes raw data without human-labeled ground truth.
Chatbot Training Application: A supervised chatbot learns from human-annotated Q&A transcripts, whereas an unsupervised chatbot clusters raw conversation logs to discover new user intents or conversational patterns autonomously.

Mentor tip

Define unsupervised learning, contrast it with supervised learning's labeled pairs, and apply the difference to chatbot training.

(b)(i)

2 marks

Cognitive learning is an AI approach where a machine acquires knowledge, reasons, and refines skills through experience and observation.
It mimics human cognitive processes such as perception, memory retention, and adaptive problem-solving.

Mentor tip

Connect acquiring skills/knowledge with mimicking human thought and experience.

(b)(ii)

2 marks

The robot can dynamically adapt to changing home environments and evolving daily routines without needing manual reprogramming.
Example: It learns to navigate around rearranged living-room furniture, closed doors, or adjusts to changes in the resident's mobility and medication schedule.

Mentor tip

State the core benefit (adapting to novel situations) and give a concrete domestic example like moved furniture.

(c)

4 marks

1) Communication Interface: A touchscreen tablet with large, high-contrast icons (or a natural voice-command microphone) connected wirelessly (Wi-Fi/Bluetooth) to the robot.
2) Concrete Task Example: The elderly resident taps 'Prepare Meal' and selects 'Soup' on the tablet (or speaks the command).
3) Signal Transmission: The tablet transmits the command code wirelessly to the robot's onboard controller.
4) Robotic Execution: The robot navigates to the kitchen and executes its pre-programmed sequence of actuator movements to heat and serve the meal.

Mentor tip

Describe the hardware link (wireless tablet or voice interface), a specific domestic request (e.g., cooking dinner), and how the robot executes the stored routine.

(d)

6 marks

Positive Social Implications:
- Promotes independent living and safety: Robots assist with heavy chores, monitor for falls, and issue medication reminders, allowing elderly individuals to remain in their own homes longer rather than moving into institutional care.
Negative Social Implications:
- Social isolation and emotional neglect: Relatives and care services may reduce personal visits, leaving the elderly person starved of genuine human warmth and companionship.
Evaluative Synthesis:
- While assistive robots provide invaluable physical support and safety monitoring, they should complement rather than replace human caregivers, because machines cannot replicate shared human empathy and emotional connection.

Mentor tip

Structure 6-mark discussions into three 2-mark parts: positive impacts, negative impacts, and a weighed evaluative conclusion.

Option C: Web science · Option

Questions 9–13 · 65 marks

Question 9

15 marks

Internet vs World Wide Web, URL structure, Common Gateway Interface (CGI), and IPv6 addressing

Web science

(a)

3 marks

The Internet is the global physical and logical infrastructure of interconnected hardware networks (routers, undersea fiber-optic cables, ISP links) that routes data packets between South Africa and the United Kingdom.
The World Wide Web (WWW) is the software service layer of hypertext documents, images, and multimedia resources stored on the South African newspaper's web server and accessed via HTTP/HTTPS.
In short, the Internet is the underlying transport network, whereas the Web is the collection of linked content travelling over it.

Mentor tip

Tie your distinction directly to the prompt's scenario (physical network link between SA and UK vs the newspaper's web pages/media).

(b)

2 marks

1) RSS feeds / push news alerts for automated story syndication.
2) Podcasts / embedded video broadcasts / interactive blogs and social media channels.

Mentor tip

List any two modern web media channels such as RSS feeds, podcasts, video clips, blogs, or social network feeds.

(c)

3 marks

1) 'http://' — Identifies the application-layer protocol (scheme) used to fetch the resource.
2) 'www.southafricantimes.com' — Identifies the domain name (host server) hosting the website.
3) '/football/mon/rt' — Specifies the hierarchical directory path and target file/resource on that server.

Mentor tip

Break the exact URL from the question into its three standard parts: protocol, domain/host, and directory/file path.

(d)

4 marks

1) Role of CGI: The Common Gateway Interface (CGI) is a standard specification enabling a web server to pass user data to an external server-side script or program.
2) Form Submission: When the user selects a radio button and submits the HTML form, the browser sends the selected parameter to the CGI script URL on the server.
3) Backend Execution: The server executes the CGI script, which uses the radio button value to query the backend database for matching content.
4) Dynamic Response: The CGI script formats the retrieved database results into a newly generated HTML page, which the web server returns to the user's browser.

Mentor tip

Trace the full CGI pipeline: form submission -> CGI script execution -> database lookup -> dynamic HTML response.

(e)

3 marks

Protocol Updated: The Internet Protocol (IP), specifically transitioning from IPv4 to IPv6.
Reason 1: Every web server and networked device requires a unique IP address, and IPv4's 32-bit address space (~4.3 billion addresses) has been exhausted.
Reason 2: IPv6 expands the address length to 128 bits, exponentially increasing the number of available unique addresses ( $2^{128}$ ) to accommodate future web growth.

IPv4 : 2^{32} \approx 4.3 \times 1 0^{9} ⟶ IPv6 : 2^{128} \approx 3.4 \times 1 0^{38}

Mentor tip

Name IP (IPv4 to IPv6) and explain that increasing the bit length (32-bit to 128-bit) exponentially expands the address space.

Question 10

15 marks

HITS algorithm ranking, lossless vs lossy compression, personalized search, and distributed systems

Web science

(a)

4 marks

1) Dual Ranking Metrics: The HITS algorithm evaluates pages matching the content query based on two scores: Authority value and Hub value.
2) Authority Calculation: A magazine's Authority score depends on the number and quality of incoming links (in-links) pointing to its articles.
3) Hub Calculation: Its Hub score depends on the quality of outgoing links (out-links) it provides to other reputable resources.
4) Contextual Differentiation: If prestigious established journals (such as Nature or Science) link frequently to Magazine A's articles while rarely linking to Magazine B, Magazine A will achieve a much higher Authority rank.

Mentor tip

Explain both Hub (out-links) and Authority (in-links) scores and include a concrete example comparing the two scientific magazines.

(b)

4 marks

1) Lossless Compression for Text: Scientific articles and text files require lossless compression (e.g., ZIP/GZIP) because losing or altering even a single character or numerical digit would corrupt the meaning and accuracy of the text.
2) Lossy Compression for Images: Photographs and illustrations can use lossy compression (e.g., JPEG) to achieve much smaller file sizes and faster downloads, as minor reductions in pixel detail are imperceptible to the human eye.

Mentor tip

Match lossless compression to text (where exact accuracy is mandatory) and lossy compression to images (where minor quality loss is tolerable for smaller size).

(c)

3 marks

Search engines track individual user context—including past browsing history, cookies, and geographic location—to personalize rankings.
Ambiguous Query Example: If two users type the search query 'Jaguar' (or 'Paris Hilton' / 'Python'), a user whose browsing history consists of automotive websites will receive car manufacturer links at the top.
Conversely, a user who frequently reads wildlife biology or programming tutorials will be prioritized links to the animal or the Python programming language.

Mentor tip

Identify user profiling (browsing history/cookies) and provide a concrete ambiguous search term that yields different top results for two user profiles.

(d)

4 marks

1) Grid Computing: Pools idle processing power across many networked computers to solve massive computational problems—for example, analyzing radio telescope data (SETI@home) or protein folding.
2) Peer-to-Peer (P2P) Network: Decentralizes file storage and bandwidth across equal participant nodes without a central server—for example, distributing large open-source Linux ISO files or media via BitTorrent.

Mentor tip

Name two distinct distributed architectures (e.g., Grid computing and Peer-to-Peer networks) and pair each with a clear real-world application.

Question 11

15 marks

Cloud-only laptops, decentralized web governance, and intellectual property protection measures

Web science

(a)(i)

2 marks

1) Software applications run remotely on cloud servers (accessed via a web browser) rather than being installed locally.
2) User files and data are stored on remote cloud storage rather than on a large local hard drive.

Mentor tip

Highlight the two core cloud shifts: remote web-based applications and remote online file storage.

(a)(ii)

4 marks

1) Connectivity Dependence: Because applications and files reside online, the user cannot access their work or use most software features when Wi-Fi/internet connectivity is unavailable or slow.
2) Software Restriction and Data Privacy: Users are locked into the cloud provider's web ecosystem (unable to install arbitrary local desktop software) and must entrust sensitive personal documents to third-party servers.

Mentor tip

Outline two distinct user drawbacks (such as offline uselessness, bandwidth latency, software lock-in, or cloud security) with an explanation for each.

(b)

5 marks

Positive Aspect (Arguments against central control):
- Promotes freedom of expression and open innovation: Information flows freely across borders without a single government or authority censoring viewpoints, and small businesses can publish globally without gatekeepers.
Negative Aspect (Arguments for control):
- Proliferation of harmful or illegal content: Absence of central oversight allows cybercrime, hate speech, terrorist propaganda (e.g., bomb-making guides), and unverified misinformation to spread rapidly.
Balanced Conclusion:
- While lack of central control requires coordinated international laws to combat criminal abuse, preserving decentralized openness remains vital to democracy and global knowledge sharing.

Mentor tip

Include a developed positive point, a developed negative point, and a reasoned concluding synthesis for full marks.

(c)

4 marks

1) Intellectual Property Example: Academic essays/theses (or digital music recordings).
2) Protection Measure: Automated plagiarism detection databases like Turnitin (or ISP monitoring and Digital Rights Management / P2P tracking for music).
3) Evaluation of Effectiveness: Turnitin cross-checks student submissions against billions of web pages and academic papers, generating a statistical similarity report that strongly deters and detects copy-paste plagiarism.
4) Limitation: However, it struggles to detect translated plagiarism from other languages, ghostwritten original essays, or AI-paraphrased text (and in music, P2P piracy persists via VPNs despite legal streaming growth).

Mentor tip

Name a specific IP type, describe a concrete technical/legal protection measure, and evaluate both its successes and its limitations.

Question 12

10 marks

Bowtie web graph structure, strongly connected core, diameter, and power-law inlink distribution

Web science

(a)(i)

1 mark

The Strongly Connected Core (SCC) is the central subset of web pages where any page can be reached from any other page via a directed path of hyperlinks.

Mentor tip

Emphasize mutual reachability: every node in the SCC can navigate to every other node in the SCC.

(a)(ii)

1 mark

The diameter of a graph is the length of the longest shortest-path (maximum geodesic distance) between any pair of connected nodes (or the average shortest path length).

Mentor tip

State 'maximum shortest path between any two nodes' (or average shortest path).

(b)

2 marks

Tubes are directed hyperlink pathways that connect pages in the IN component directly to pages in the OUT component.
Crucially, traffic travelling through a tube bypasses the Strongly Connected Core (SCC) entirely.

Mentor tip

Specify both endpoints (from IN pages to OUT pages) and that the path bypasses the SCC.

(c)(i)

1 mark

The IN component of the bowtie structure.

Mentor tip

Newly created resource lists with outgoing links belong to the IN component.

(c)(ii)

2 marks

Because the course resource lists were only recently published, external websites have not yet discovered or linked back to them (zero or very few in-links).
However, the resource lists themselves contain many outgoing hyperlinks pointing to established reference pages inside the SCC.

Mentor tip

Combine both facts: they are too new to have incoming links, but they point outward to established core sites.

(d)

3 marks

The distribution follows a scale-free power law of the form $y = a x^{- c}$ (where frequency is inversely proportional to a power of the number of inlinks).
Preferential Attachment ('rich-get-richer'): A tiny minority of well-known authority hubs attract millions of incoming links, while the vast majority of web pages ('the long tail') receive very few links.
As the number of inlinks increases, the number of pages possessing that many links drops off exponentially faster.

P (k) \propto k^{- γ}

Mentor tip

Mention the inverse power-law relationship ( $y = a x^{- c}$ ), the 'long tail' of pages with few links, and preferential attachment.

Question 13

10 marks

Social bookmarking, folksonomies vs search engine indexing, and ontologies for climate change research

Web science

(a)

2 marks

Social bookmarking services allow users to save, organize, and share links to web resources online.
When users apply their own free-form keyword tags to categorize those bookmarks, the aggregated crowd-sourced tag collection forms a folksonomy.

Mentor tip

Explain that user tagging within social bookmarking platforms generates a folksonomy.

(b)

4 marks

Human Relevance vs Algorithmic Errors:
- Folksonomy tags are chosen by human readers who understand the nuanced meaning, tone, and context of a resource, whereas automated search engine crawlers rely on keyword frequency algorithms that can misclassify pages.
Vocabulary Inconsistency vs Algorithmic Consistency:
- However, folksonomies suffer from human inconsistency—such as typos, synonyms, plurals, and personal bias—whereas automated search engine indexing applies deterministic, uniform rules across all pages.

Mentor tip

Compare both sides: human contextual accuracy vs crawler mislabelling, and human tag inconsistency vs algorithmic uniformity.

(c)

4 marks

1) Standardized Vocabulary and Relationships: Ontologies formally define domain concepts, data types, and hierarchical relationships (e.g., linking 'CO2 emissions', 'sea surface temperature', and 'glacial melt').
2) Machine Readability: By annotating climate datasets with shared ontological metadata (RDF/OWL), disparate databases become machine-readable and semantically interoperable.
3) Automated Global Aggregation: Automated software agents can seamlessly aggregate and cross-reference climate records from meteorological agencies, satellites, and universities worldwide.
4) Enhanced Scientific Insight: Researchers can run complex cross-disciplinary queries and climate models on unified data, improving predictive accuracy and policy decisions.

Mentor tip

Trace the chain: formal shared vocabulary -> machine-readable interoperability -> automated multi-source data integration -> deeper climate insights.

Option D: Object-oriented programming · Option

Questions 14–17 · 65 marks

Question 14

15 marks

Train and RollingStock OOP hierarchy: constructors, polymorphism, UML, and array stack operations

Object-oriented programming

(a)

2 marks

A constructor is a special method invoked automatically when a new instance of a class is instantiated.
Its primary function is to allocate memory for the new object and initialize its instance variables to valid starting values.

Mentor tip

Mention both object instantiation (creation) and initializing member variables.

(b)

3 marks

Advantage: Polymorphism allows external code to invoke the same method signature (such as getWeight()) across a collection of different RollingStock subclasses without needing to know their exact runtime subtype.
Contextual Example: Both Engine and Wagon inherit from RollingStock and implement getWeight().
Elaboration: Calling getWeight() on an Engine returns its fixed 120,000 kg base weight via super.getWeight(), whereas calling getWeight() on a Wagon overrides the method to sum the empty wagon's 32,000 kg weight plus the dynamic weights of all loaded Parcel objects.

Mentor tip

Connect the general benefit of polymorphism (unified interface with specialized behavior) directly to Engine.getWeight() vs Wagon.getWeight().

(c)

3 marks

Draw a 3-compartment UML class box for Train:
1) Class Name: Train
2) Private Attributes (-):
- mEngines : Engine[]
- mWagons : Wagon[]
- mEngineCount : int
- mWagonCount : int
- mTrainNumber : int
- mWeight : double
3) Public Methods (+):
+ Train(number : int)
+ addEngine(newEngine : Engine) : void
+ removeEngine() : Engine
+ addWagon(newWagon : Wagon) : void
+ removeWagon() : Wagon
+ getWeight() : double

Mentor tip

List all six private fields with '-' and all six public methods/constructors with '+' inside a 3-section UML box.

(d)

2 marks

Construct a public accessor method returning mWagonCount:
public int getNumberOfWagons() {
return mWagonCount;
}

Mentor tip

Ensure the return type is int and the method returns the mWagonCount counter.

(e)

5 marks

Check that mWagonCount > 0 before decrementing and returning the top Wagon element; return null if the train has no wagons:
public Wagon removeWagon() {
if (mWagonCount > 0) {
mWagonCount--;
return mWagons[mWagonCount];
} else {
return null;
}
}

Mentor tip

Mirror the logic of removeEngine(), adding an if (mWagonCount > 0) guard to prevent an ArrayIndexOutOfBoundsException.

Question 15

15 marks

Standard library collections, team software development, and refactoring Parcel with Address classes

Object-oriented programming

(a)

2 marks

Saves development time through reliable code reuse:
Developers can immediately use pre-built, thoroughly tested, and optimized data structures (like ArrayList or LinkedList) without having to write and debug them from scratch.

Mentor tip

Emphasize time saved and reliability gained from reusing pre-tested library code.

(b)

4 marks

1) Role Specialization: In a team, developers can focus on their specific areas of expertise (e.g., UI design, backend algorithms, or quality assurance testing), whereas a solo programmer must handle every aspect alone.
2) Parallel Development via Information Hiding: By agreeing on public class interfaces and encapsulating internal details, multiple team members can code different classes simultaneously, though teams require formal communication and version control.

Mentor tip

State two clear differences (such as specialist roles and parallel modular development) and explain each.

(c)(i)

1 mark

String

Mentor tip

Street names contain alphabetic characters and spaces.

(c)(ii)

1 mark

int (or String if alphanumeric building numbers like '12B' are considered).

Mentor tip

int is the standard primitive type expected for a building number.

(c)(iii)

1 mark

boolean

Mentor tip

A true/false flag is always represented by a boolean.

(d)

3 marks

1) Change the data types of destinationAddress and originAddress from String to Address (and encapsulate them as private fields).
2) Update the Parcel constructor parameters (or initialization logic) to receive Address objects.
3) Add public getter and setter methods for destinationAddress and originAddress referencing the Address type.

Mentor tip

Mention updating the field declarations, the constructor, and the accessor/mutator methods.

(e)

3 marks

Use inheritance by defining OriginAddress and DestinationAddress as subclasses that extend the base Address superclass.
Both subclasses automatically inherit shared attributes (street name, building number, city, country, isBusiness) without rewriting them.
DestinationAddress adds only its specific deliveryInstructions attribute, while OriginAddress adds only its collectionLocation indicator.

Mentor tip

Explain that extending the base Address class via inheritance shares common fields while isolating unique attributes in each subclass.

Question 16

15 marks

Tracing array operations and implementing getWeight() across Wagon and Train classes

Object-oriented programming

(a)(i)

2 marks

Engine B (ID 7) is added at index 0, followed by new Engine(9) at index 1:
Index [0]: Engine ID 7
Index [1]: Engine ID 9
Indices [2..5]: null / empty

Mentor tip

Show the array indices [0] and [1] containing Engine 7 and Engine 9 in exact insertion order.

(a)(ii)

1 mark

Mentor tip

Two engines were added and none were removed, so mEngineCount equals 2.

(a)(iii)

2 marks

Initially Wagons 23, 66, and 71 are placed at indices [0], [1], and [2].
Two removeWagon() calls decrement mWagonCount from 3 to 1 (removing 71 and 66), and addWagon(new Wagon(214)) places Wagon 214 at index [1]:
Index [0]: Wagon ID 23
Index [1]: Wagon ID 214

Mentor tip

Trace LIFO stack behavior: adding 23, 66, 71, popping twice leaves 23 at [0], and pushing 214 places it at [1].

(b)

4 marks

Start with the empty wagon's base weight via super.getWeight() (or 32000), loop from 0 to mParcelCount - 1 to add each parcel's weight, and return the sum:
public double getWeight() {
double totalWeight = super.getWeight();
for (int i = 0; i < mParcelCount; i++) {
totalWeight += mParcels[i].getWeight();
}
return totalWeight;
}

Mentor tip

Don't forget to include the empty wagon's own weight (super.getWeight()) before summing the mParcels array up to mParcelCount.

(c)

4 marks

Initialize totalWeight to 0, sum the weights of all active engines (0 to mEngineCount - 1), sum the weights of all active wagons (0 to mWagonCount - 1, where Wagon.getWeight() already includes its parcels), and return the total:
public double getWeight() {
double totalWeight = 0;
for (int i = 0; i < mEngineCount; i++) {
totalWeight += mEngines[i].getWeight();
}
for (int i = 0; i < mWagonCount; i++) {
totalWeight += mWagons[i].getWeight();
}
return totalWeight;
}

Mentor tip

Calling mWagons[i].getWeight() automatically returns the combined weight of the wagon and its parcels.

(d)

2 marks

The two getWeight() methods are defined within separate class scopes (Train vs Wagon).
The compiler resolves which getWeight() method to invoke based on the specific class type of the object reference on which the method is called.

Mentor tip

Explain that each class has its own namespace/scope, so the compiler binds the method call to the object's class type.

Question 17

20 marks

Refactoring Train to use a custom linked list of Wagons, LIFO Stack ADT for parcels, and coding conventions

Object-oriented programming

(a)

3 marks

1) Modify the Wagon class to include a self-referential pointer/reference attribute (e.g., public Wagon nextWagon;) pointing to the next wagon node.
2) Replace the mWagons[] array in the Train class with a head reference variable (e.g., private Wagon firstWagon;) initialized to null.
3) Rewrite the Train methods (addWagon, removeWagon, getWeight) to traverse and manipulate node pointers rather than array indices.

Mentor tip

Identify the three structural changes: adding a 'next' pointer in Wagon, replacing the array with a 'head' pointer in Train, and updating the list-handling methods.

(b)

3 marks

Point the new wagon's nextWagon reference to the current head (firstWagon), then update firstWagon to reference newWagon:
public void addWagon(Wagon newWagon) {
newWagon.nextWagon = firstWagon;
firstWagon = newWagon;
mWagonCount++;
}

Mentor tip

Always assign newWagon.nextWagon = firstWagon BEFORE overwriting firstWagon = newWagon so you don't lose the rest of the list.

(c)

5 marks

1) Method Parameter: Take the target wagon's unique ID number (int targetID) as a parameter.
2) Head Initialization: Start traversal pointers (current and previous) at the head of the list (firstWagon).
3) Search Loop: Traverse the linked list node-by-node while current != null, comparing current.getWagonID() against targetID.
4) Unlinking Node: If a match is found, bypass the node by setting previous.nextWagon = current.nextWagon (or updating firstWagon = current.nextWagon if removing the head node) and return the removed Wagon.
5) Not Found / Empty Case: If traversal reaches the end of the list (current == null) without a match, return null.

Mentor tip

Cover all 5 steps: taking the ID parameter, starting at firstWagon, looping and checking IDs, relinking pointers to bypass the target node, and handling the not-found/empty case.

(d)

1 mark

Stack (Last-In, First-Out / LIFO dynamic data structure).

Mentor tip

First-in, last-out (single-door loading/unloading) is the exact definition of a Stack.

(e)(i)

2 marks

Push the new Parcel object onto the stack instance named 'model':
public void addParcel(Parcel newParcel) {
model.push(newParcel);
}

Mentor tip

Use the standard stack push() operation on the 'model' object.

(e)(ii)

2 marks

Pop and return the top Parcel object from the stack instance named 'model':
public Parcel getParcel() {
return (Parcel) model.pop();
}

Mentor tip

Use the standard stack pop() operation to remove and return the most recently loaded parcel.

(f)

4 marks

1) Code Style Conventions (Indentation, Whitespace, and Comments):
- Consistent indentation and explanatory comments clearly expose control-flow nesting and algorithmic intent, making code faster to read, debug, and maintain.
2) Naming Conventions (Meaningful Identifiers and Prefixes):
- Descriptive variable and method names (such as mWagonCount or getWeight() rather than x or f1) make code self-documenting, reducing cognitive load and preventing misunderstandings across collaborative programming teams.

Mentor tip

Address both style conventions (indentation/comments) and naming conventions (meaningful identifiers), explaining how each reduces maintenance cost in teams.

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, Q13
Object-oriented programming65 marks · 25%
Q14, Q15, Q16, Q17

Mentor insights

Traps to watch for

01
Databases: Global retail data warehouse: ETL pipelines, association vs class · Q4
State the core advantage (integrated historical trend analysis) and give a concrete retail example.
02
Modelling and simulation: Chatbot machine learning paradigms and cognitive care robots for · Q8
Mention both receiving human language input and generating conversational language output.
03
Web science: Internet vs World Wide Web, URL structure, Common Gateway Interfa · Q9
Tie your distinction directly to the prompt's scenario (physical network link between SA and UK vs the newspaper's web pages/media).
04
Object-oriented programming: Refactoring Train to use a custom linked list of Wagons, LIFO Sta · Q17
Identify the three structural changes: adding a 'next' pointer in Wagon, replacing the array with a 'head' pointer in Train, and updating the list-handling methods.

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: November 2014
Marks: 260
Duration: 1 hour 20 min
Calculator: Not allowed
Questions: 17

Q1 Q2 Q3 Q4 Q5 Q6 Q7 Q8 Q9 Q10 Q11 Q12 Q13 Q14 Q15 Q16 Q17

Keep your copy of the paper open alongside. We don't reproduce the questions: each part below is labelled to match the November 2014 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

16 marks

Banking database transactions, rollback recovery, validation vs verification, and ACID isolation

Databases

(a)(i)

3 marks

An operation is a single, elementary action performed on a database (such as a single read or write command).
A transaction is a logical unit of work comprising a sequence of multiple operations that must all succeed together to preserve database consistency.
Example: A cash withdrawal transaction combines an initial query operation to check the current balance followed by an update operation to deduct the withdrawn sum.

Mentor tip

Contrast a single atomic database action (operation) with a multi-step logical unit of work (transaction) using a banking example.

(a)(ii)

1 mark

A transaction must be rolled back if any constituent operation fails before completion (e.g., network interruption, hardware failure, or insufficient account funds).

Mentor tip

Any condition preventing the full sequence of operations from completing successfully triggers a rollback.

(a)(iii)

1 mark

Restores the database to its exact consistent state prior to the start of the transaction, undoing any partial modifications.

Mentor tip

State clearly that the database is returned to its pre-transaction state.

(b)

4 marks

1) Built-in transaction and concurrency management: A database engine enforces ACID guarantees and record locking out of the box, ensuring deposits and withdrawals do not corrupt balances.
2) Standardized querying and reduced redundancy: Centralizing customer and account records eliminates duplicate data entry and provides fast, standardized retrieval and sorting across multiple banking workflows.

Mentor tip

Provide two distinct reasons (e.g., transaction integrity and efficient querying/minimal redundancy) with an elaboration for each.

(c)

4 marks

Data Validation:
- Definition: Automated check ensuring entered input conforms to expected data type, range, and format rules.
- Banking Example: Checking that the deposit amount field contains a positive numeric currency value rather than alphabetic characters or a negative number.
Data Verification:
- Definition: Confirming that the data entered into the system accurately matches the original source value intended by the user.
- Banking Example: Prompting the teller or customer to enter the deposit amount twice (double entry) or displaying a confirmation screen to verify the figure before committing.

Mentor tip

Remember: validation checks 'is it sensible/valid format?' while verification checks 'was it copied/typed accurately as intended?'

(d)

3 marks

Isolation ensures that multiple transactions executing concurrently leave the database in the exact same state as if they had been executed sequentially one after another.
Without isolation, interleaved read/write steps can cause a lost update anomaly.
Banking Example: If a $10 d e p os i t an d a$ 50 withdrawal occur simultaneously on an account, isolation locks the balance so both updates apply cleanly in sequence, guaranteeing a net change of -$40 rather than one transaction overwriting the other.

Mentor tip

Define isolation in terms of sequential equivalence and illustrate with simultaneous deposit/withdrawal amounts.

Question 2

15 marks

Online hat store DBMS: attribute data types, core DBMS tools, security, data dictionary, and modelling

Databases

(a)

3 marks

1) Hat size: Integer (or Float/Numeric, or single Character/String for S/M/L).
2) Price: Float / Real / Decimal / Currency.
3) Maker's name: String / Text / Varchar.

Mentor tip

Assign a distinct, appropriate primitive/SQL data type to each of the three attributes.

(b)

3 marks

1) Database creation / Data Definition (defining tables, schemas, and constraints).
2) Database manipulation (inserting, updating, and deleting records).
3) Database interrogation / querying (retrieving and filtering data via queries and reports).

Mentor tip

Think of DDL (creation), DML (manipulation), and DQL (interrogation/querying).

(c)

4 marks

1) Role-Based Access Control (Access Rights): Restricts privileges by user role so online shoppers can only view hat catalogs and their own orders, while only authenticated store administrators can alter prices or inventory.
2) Record Locking / Encryption and Audit Logging: Locks records during concurrent checkout updates to prevent stock corruption, and encrypts stored customer payment/personal details against unauthorized breach.

Mentor tip

Name two concrete DBMS security mechanisms (such as access rights/permissions and locking/encryption) and explain how each protects the online shop.

(d)

2 marks

A data dictionary is a centralized metadata repository that defines the structure and properties of all data elements in the database.
It specifies field names, data types, format constraints, semantic meanings, and inter-table relationships.

Mentor tip

Define the data dictionary as 'data about data' (metadata) detailing formats, meanings, and relationships.

(e)

3 marks

Provides a formal conceptual blueprint prior to physical implementation.
Ensures all required business entities, attributes, and relationships are accurately captured without omissions.
Offers a clear visual representation that developers and store owners can review and validate together before coding.

Mentor tip

Highlight its role as a design blueprint, completeness check, and communication bridge with stakeholders.

Question 3

14 marks

Bookstore database normalization, redundancy issues, ERD construction, and reader privacy

Databases

(a)

2 marks

A normalized database organizes attributes into separate related tables to minimize or eliminate data redundancy and undesirable functional dependencies.
Consequently, updating or modifying a specific fact only requires changing a single record in one table.

Mentor tip

Mention both eliminating redundancy/dependencies and single-point updates.

(b)

2 marks

1) Wastes storage capacity by repeating identical attribute values across multiple rows.
2) Introduces update anomalies and data inconsistency if a repeated value is updated in one location but overlooked in another.

Mentor tip

Focus on wasted disk space, extra maintenance overhead, and the risk of inconsistent records.

(c)

3 marks

Define separate entity boxes reflecting the four specification points:
1) BOOK (Book_Title, Author, Publication_Date, List_Price, Number_of_Copies)
2) BOOKCASE (Bookcase_Number, Aisle_Number, Shelf*)
3) SHELF (Shelf_ID, Book_Title*)
4) SALES_LOG (Book_Title, Customer_Name, Price_Paid, Sale_Date)
Draw relationship links with cardinalities: BOOKCASE (1) ----< contains >---- (0..N) SHELF, SHELF (1) ----< holds >---- (0..N) BOOK, and BOOK (1) ----< recorded in >---- (0..N) SALES_LOG.

Mentor tip

Include the four entities with their listed attributes and connect them with clear 1-to-many relationship lines.

(d)(i)

1 mark

The tables lack explicit unique Primary Keys, or the 'Shelf' attribute in Bookcase / 'Book' attribute in Shelf contains non-atomic repeating groups of values.

Mentor tip

Point out either missing primary keys or non-atomic multi-valued columns (e.g., multiple books listed inside a single shelf record).

(d)(ii)

3 marks

1) Assign atomic Primary Keys to each entity (e.g., ISBN/Book_ID for Book, Customer_ID for a separate Customer table, Transaction_ID for Sales_Log).
2) Decompose repeating shelving groups into a dedicated associative/junction table (e.g., SHELVING with Book_ID, Shelf_Number, Bookcase_Number) so every field is atomic.
3) Extract Customer details (Customer_ID, Customer_Name) into a separate CUSTOMER table so repeat buyers are not duplicated across Sales_Log rows.

Mentor tip

Explain how to make fields atomic via a junction table and eliminate transitive/partial dependencies using separate tables with primary/foreign keys.

(e)

3 marks

Issue Identification: Disclosure of personal reading and purchasing histories in the Sales_Log without individual consent or a specific warrant constitutes a breach of civil privacy.
Discussion: Law enforcement might profile citizens based on controversial or political book titles they purchased, yet purchasing a book does not reliably indicate malicious intent (e.g., academic research or gifts), potentially leading to unjust suspicion or chilling effects on free inquiry.

Mentor tip

Identify the privacy violation clearly, then discuss why profiling reading habits can be both intrusive and misleading.

Question 4

20 marks

Global retail data warehouse: ETL pipelines, association vs classification, deviation detection, and targeted marketing ethics

Databases

(a)

2 marks

Consolidates historical data from global stores, factories, and logistics into a single repository to enable longitudinal trend analysis.
Example: Analysts can compare multi-year seasonal sales across regions to optimize inventory forecasting.

Mentor tip

State the core advantage (integrated historical trend analysis) and give a concrete retail example.

(b)

3 marks

1) Extract: Pull raw operational data from disparate heterogeneous sources (global stores, Chinese factories, logistics partners).
2) Transform: Cleanse, validate, standardize formats/currencies, and deduplicate the extracted data to match the warehouse schema.
3) Load: Write and time-stamp the transformed data into the target data warehouse fact and dimension tables.

Mentor tip

Define all three stages of ETL—Extract, Transform, and Load—in sequence.

(c)

4 marks

1) Manufacturing Data Source: Chinese plants record production labor costs in Chinese Yuan (CNY) and dates as YYYY/MM/DD.
2) Transport Provider Source: Logistics carriers record shipping invoices in US Dollars (USD) and weights in pounds (lbs).
3) Warehouse Standard: The central data warehouse requires all financial figures in Euros (EUR) and weights in kilograms (kg).
4) Transformation Action: During ETL, the transformation rules convert CNY and USD values into Euros using historical exchange rates and convert pounds to kilograms before loading.

Mentor tip

Provide a concrete standardization example (such as currency conversion or unit/date formatting) linking manufacturing data, shipping data, and the unified warehouse format.

(d)

4 marks

Association Rule Mining:
- Discovers co-occurrence relationships ('if-then' rules) between items within the same transaction.
- Example: Discovering that customers who purchase baby formula also frequently purchase diapers in the same basket.
Classification:
- Assigns records into predefined categorical classes based on historical attribute profiles.
- Example: Categorizing a customer into a predefined loyalty tier ('High-Value', 'At-Risk', 'Budget') based on annual spend and visit frequency.

Mentor tip

Contrast item-to-item co-occurrence rules (Association) against sorting records into predefined groups (Classification) with a retail example of each.

(e)

2 marks

Deviation detection (outlier/anomaly analysis) identifies records that diverge significantly from established norms or expected statistical distributions.
Retail Example: Flagging a sudden fraudulent bulk credit-card purchase or a shipping container whose transit time took three times longer than average.

Mentor tip

Define deviation detection as finding statistical outliers/anomalies and give a clear retail or logistics example.

(f)

2 marks

1) Association rule mining (or Cluster analysis).
2) Classification (or Sequential pattern mining).

Mentor tip

Any two standard data mining techniques—Association, Classification, or Cluster Analysis—earn full marks.

(g)

3 marks

Ethical Issue: Inferring sensitive personal health conditions (pregnancy) from routine shopping habits without consent violates medical and personal privacy.
Contextual Example: Sending baby-related coupons to a household mailbox can inadvertently disclose a pregnancy to family members, roommates, or partners before the customer chooses to share the news.
Broader Discussion: Even if individual product purchases are benign, predictive analytics creates sensitive health profiles outside healthcare privacy laws, risking emotional distress if the prediction is wrong or premature.

Mentor tip

Discuss how predictive profiling bypasses medical privacy norms and can inadvertently reveal intimate life events to other household members.

Option B: Modelling and simulation · Option

Questions 5–8 · 65 marks

Question 5

15 marks

Patient antibiotic dosage mathematical model, scheduling algorithm, and bloodstream retention simulation

Modelling and simulation

(a)

1 mark

A 20-year-old patient falls into the 18–40 age bracket (1.50 mg per kg). Multiply weight by rate:
Dose = 50 kg * 1.50 mg/kg = 75 mg.

Dose = 50 \times 1.50 = 75 mg

Mentor tip

Match age 20 to the 18–40 row (1.50 mg/kg) and multiply by 50 kg.

(b)

2 marks

Variables: patient age (integer), patient weight (real/float), rate per kg (real/float), and total dose amount (real/float).
Calculation: Determine rate_per_kg from age bracket, then compute Dose = weight * rate_per_kg.

Dose = weight \times rate (age)

Mentor tip

List the input/output variables clearly and state the multiplication formula.

(c)

4 marks

Input patient weight and age, then compute the single-dose amount using the age-bracket lookup.
Use an outer loop running day from 1 to 10, and an inner loop running dose index from 0 to 2 (representing 8:00, 16:00, and 24:00/0:00):
Pseudocode:
input WEIGHT, AGE
if AGE <= 40 then RATE = 1.50
else if AGE <= 60 then RATE = 1.75
else RATE = 2.00
end if
DOSE = WEIGHT * RATE
loop DAY from 1 to 10
loop PERIOD from 0 to 2
TIME = 8 + PERIOD * 8
output "day" + DAY, TIME + ":00", DOSE
end loop
end loop

Mentor tip

Include input/dose calculation, an outer loop for 10 days, and an inner loop generating the three 8-hour intervals (08:00, 16:00, 24:00).

(d)(i)

4 marks

1) In a spreadsheet, create columns for Dose Number (1 to 30, or Day/Time), Dose Administered (Column B, constant D), and Bloodstream Medicine Level (Column C).
2) In Row 2 (first dose at day1 08:00), set the initial bloodstream amount equal to the administered dose: C2 = B2.
3) In Row 3 (second dose), enter the recurrence formula adding the new dose to 20% of the previous bloodstream level: C3 = B3 + 0.20 * C2.
4) Replicate (fill down) the formula in Column C down to Dose 30 to model accumulation over the 10-day course.

BloodLevel_{k} = Dose + 0.20 \times BloodLevel_{k - 1}

Mentor tip

Show both the spreadsheet column setup and the recurrence formula (New Level = Dose + 0.2 * Previous Level).

(d)(ii)

1 mark

Mistyping or mismeasuring the patient's weight or age (e.g., entering an out-of-range age or wrong unit) causes the model to compute an incorrect dosage (underdosing or overdosing the patient).

Mentor tip

Link an input error in age or weight directly to an incorrect calculated dosage.

(d)(iii)

3 marks

Take clinical blood samples from patients immediately after each scheduled dose to measure actual antibiotic concentrations in vivo.
Compare these empirical blood test measurements against the values predicted by the spreadsheet model.
Large discrepancies reveal either input errors or oversimplified absorption assumptions (such as the fixed 20% retention rate).

Mentor tip

Validate a physiological simulation by comparing predicted values against real-world empirical blood measurements.

Question 6

15 marks

Town population growth simulation: demographic factors, model accuracy, and urban planning consequences

Modelling and simulation

(a)

1 mark

Initialize population P_0 = 25,000, and for each year t from 1 to 10, multiply the previous year's population by 1.07 (P_t = P_{t-1} * 1.07).

P_{t} = P_{t - 1} \times 1.07

Mentor tip

State the initial population input and the annual 1.07 multiplication factor.

(b)

3 marks

1) Birth rate (annual live births per capita).
2) Death rate (annual mortality per capita).
3) Net migration: Immigration (new residents moving in) and Emigration (residents moving away).

Mentor tip

The classic demographic variables are birth rate, death rate, immigration, and emigration.

(c)

3 marks

Replace the fixed 7% multiplier with a composite annual rate derived from the three demographic factors:
1) Estimate dynamic annual birth and death rates using historical civil registry trends.
2) Incorporate projected immigration and emigration figures based on local housing developments, new factory openings, or business closures.
3) Compute each year's population as: P_t = P_{t-1} + Births_t - Deaths_t + Immigration_t - Emigration_t.

P_{t} = P_{t - 1} + B_{t} - D_{t} + I_{t} - E_{t}

Mentor tip

Explain how each of the three factors is estimated and combined into the annual population update equation.

(d)

4 marks

1) Omitted macroeconomic variables: The simulation ignores external drivers such as national economic recessions, interest rates, or employment shifts that heavily influence migration.
2) Non-linear 'new town' dynamics: Extrapolating a 10-year horizon from a single year's 7% surge is unreliable because newly established towns often experience an initial rapid influx that quickly plateaus once housing capacity fills.

Mentor tip

Discuss two distinct sources of inaccuracy: unmodelled external factors and the danger of extrapolating short-term new-town surges over 10 years.

(e)

4 marks

1) Consequence of Under-prediction (Actual > Predicted): Infrastructure such as schools, hospitals, water supply, and road networks will be under-built, resulting in overcrowded classrooms and strained healthcare services.
2) Consequence of Over-prediction (Actual < Predicted): The municipality over-invests public funds into unneeded facilities and housing, leaving the town with vacant buildings and severe budget deficits.

Mentor tip

Contrast the social impact of under-estimating (shortage of schools/hospitals) vs over-estimating (wasted tax revenue and empty infrastructure).

Question 7

15 marks

Automotive CAD wireframe representation, geometric manipulation, and 3D rendering techniques

Modelling and simulation

(a)

2 marks

A wireframe model represents the 3D structural skeleton of the car as a collection of vertices (3D coordinate points) connected by edges (lines/vectors) and curves.
These mathematical coordinates are stored in memory and projected onto the 2D screen without solid surfaces or shading.

Mentor tip

Mention vertices (points) connected by lines/vectors representing the 3D skeletal structure.

(b)(i)

1 mark

Altering the structural body profile, such as extending the chassis length, lowering the roofline, or converting a 3-door hatchback into a 5-door model.

Mentor tip

Choose a structural/geometric shape modification rather than surface color or texture.

(b)(ii)

2 marks

The stored (x, y, z) coordinates of the relevant vertices are translated or scaled.
The connecting vectors automatically recalculate their lengths and angles, reshaping the wireframe geometry.

Mentor tip

Explain that modifying vertex coordinates updates the connecting vectors/lines.

(b)(iii)

2 marks

The designer selects and drags vertices/control points directly on screen using a mouse, stylus/graphics tablet, or touch display.
Alternatively, the designer types exact numerical coordinate dimensions or scaling factors into the CAD interface.

Mentor tip

Mention both direct graphical manipulation (mouse/stylus drag) and numeric coordinate entry.

(c)

2 marks

1) Exterior: Testing different metallic paint colors, surface finishes, or headlight/grille styling.
2) Interior: Changing seat upholstery fabric/leather textures or dashboard trim colors.

Mentor tip

Give one aesthetic/surface change for the exterior and one for the interior.

(d)

6 marks

1) Polygon Mesh / Surface Skinning: The open spaces between wireframe edges are filled with polygon surfaces to form a solid 3D shell, hiding occluded back-facing lines (hidden surface removal).
2) Texture Mapping and Colouring: Surface colours and material textures (metallic paint, glass, fabric) are mapped onto the polygon faces.
3) Lighting and Shading / Ray Tracing: Virtual light sources are positioned, and algorithms like ray tracing or Phong shading calculate reflections, specular highlights, and shadows across curved body panels.
4) Continuous Re-rendering: Whenever the designer rotates or zooms the camera view, the GPU recomputes the 3D-to-2D raster projection in real time.

Mentor tip

Cover surface filling, colour/texture mapping, ray tracing/lighting calculations, and re-rendering upon rotation.

Question 8

20 marks

Chatbot machine learning paradigms and cognitive care robots for the elderly

Modelling and simulation

(a)(i)

2 marks

A chatbot is an interactive computer program designed to simulate human conversation.
It parses natural language input (text or speech) from a user and generates contextually appropriate text or spoken responses.

Mentor tip

Mention both receiving human language input and generating conversational language output.

(a)(ii)

1 mark

Supervised learning trains a model using labeled data consisting of known input queries paired with verified target responses.

Mentor tip

Emphasize training on labeled input-output pairs where the desired response is known in advance.

(a)(iii)

3 marks

Unsupervised learning analyzes unlabeled data to discover hidden statistical patterns and clusters without predefined target outputs.
Comparison: Unlike supervised learning—which maps inputs to a curated set of known answers—unsupervised learning self-organizes raw data without human-labeled ground truth.
Chatbot Training Application: A supervised chatbot learns from human-annotated Q&A transcripts, whereas an unsupervised chatbot clusters raw conversation logs to discover new user intents or conversational patterns autonomously.

Mentor tip

Define unsupervised learning, contrast it with supervised learning's labeled pairs, and apply the difference to chatbot training.

(b)(i)

2 marks

Cognitive learning is an AI approach where a machine acquires knowledge, reasons, and refines skills through experience and observation.
It mimics human cognitive processes such as perception, memory retention, and adaptive problem-solving.

Mentor tip

Connect acquiring skills/knowledge with mimicking human thought and experience.

(b)(ii)

2 marks

The robot can dynamically adapt to changing home environments and evolving daily routines without needing manual reprogramming.
Example: It learns to navigate around rearranged living-room furniture, closed doors, or adjusts to changes in the resident's mobility and medication schedule.

Mentor tip

State the core benefit (adapting to novel situations) and give a concrete domestic example like moved furniture.

(c)

4 marks

1) Communication Interface: A touchscreen tablet with large, high-contrast icons (or a natural voice-command microphone) connected wirelessly (Wi-Fi/Bluetooth) to the robot.
2) Concrete Task Example: The elderly resident taps 'Prepare Meal' and selects 'Soup' on the tablet (or speaks the command).
3) Signal Transmission: The tablet transmits the command code wirelessly to the robot's onboard controller.
4) Robotic Execution: The robot navigates to the kitchen and executes its pre-programmed sequence of actuator movements to heat and serve the meal.

Mentor tip

Describe the hardware link (wireless tablet or voice interface), a specific domestic request (e.g., cooking dinner), and how the robot executes the stored routine.

(d)

6 marks

Positive Social Implications:
- Promotes independent living and safety: Robots assist with heavy chores, monitor for falls, and issue medication reminders, allowing elderly individuals to remain in their own homes longer rather than moving into institutional care.
Negative Social Implications:
- Social isolation and emotional neglect: Relatives and care services may reduce personal visits, leaving the elderly person starved of genuine human warmth and companionship.
Evaluative Synthesis:
- While assistive robots provide invaluable physical support and safety monitoring, they should complement rather than replace human caregivers, because machines cannot replicate shared human empathy and emotional connection.

Mentor tip

Structure 6-mark discussions into three 2-mark parts: positive impacts, negative impacts, and a weighed evaluative conclusion.

Option C: Web science · Option

Questions 9–13 · 65 marks

Question 9

15 marks

Internet vs World Wide Web, URL structure, Common Gateway Interface (CGI), and IPv6 addressing

Web science

(a)

3 marks

The Internet is the global physical and logical infrastructure of interconnected hardware networks (routers, undersea fiber-optic cables, ISP links) that routes data packets between South Africa and the United Kingdom.
The World Wide Web (WWW) is the software service layer of hypertext documents, images, and multimedia resources stored on the South African newspaper's web server and accessed via HTTP/HTTPS.
In short, the Internet is the underlying transport network, whereas the Web is the collection of linked content travelling over it.

Mentor tip

Tie your distinction directly to the prompt's scenario (physical network link between SA and UK vs the newspaper's web pages/media).

(b)

2 marks

1) RSS feeds / push news alerts for automated story syndication.
2) Podcasts / embedded video broadcasts / interactive blogs and social media channels.

Mentor tip

List any two modern web media channels such as RSS feeds, podcasts, video clips, blogs, or social network feeds.

(c)

3 marks

1) 'http://' — Identifies the application-layer protocol (scheme) used to fetch the resource.
2) 'www.southafricantimes.com' — Identifies the domain name (host server) hosting the website.
3) '/football/mon/rt' — Specifies the hierarchical directory path and target file/resource on that server.

Mentor tip

Break the exact URL from the question into its three standard parts: protocol, domain/host, and directory/file path.

(d)

4 marks

1) Role of CGI: The Common Gateway Interface (CGI) is a standard specification enabling a web server to pass user data to an external server-side script or program.
2) Form Submission: When the user selects a radio button and submits the HTML form, the browser sends the selected parameter to the CGI script URL on the server.
3) Backend Execution: The server executes the CGI script, which uses the radio button value to query the backend database for matching content.
4) Dynamic Response: The CGI script formats the retrieved database results into a newly generated HTML page, which the web server returns to the user's browser.

Mentor tip

Trace the full CGI pipeline: form submission -> CGI script execution -> database lookup -> dynamic HTML response.

(e)

3 marks

Protocol Updated: The Internet Protocol (IP), specifically transitioning from IPv4 to IPv6.
Reason 1: Every web server and networked device requires a unique IP address, and IPv4's 32-bit address space (~4.3 billion addresses) has been exhausted.
Reason 2: IPv6 expands the address length to 128 bits, exponentially increasing the number of available unique addresses ( $2^{128}$ ) to accommodate future web growth.

IPv4 : 2^{32} \approx 4.3 \times 1 0^{9} ⟶ IPv6 : 2^{128} \approx 3.4 \times 1 0^{38}

Mentor tip

Name IP (IPv4 to IPv6) and explain that increasing the bit length (32-bit to 128-bit) exponentially expands the address space.

Question 10

15 marks

HITS algorithm ranking, lossless vs lossy compression, personalized search, and distributed systems

Web science

(a)

4 marks

1) Dual Ranking Metrics: The HITS algorithm evaluates pages matching the content query based on two scores: Authority value and Hub value.
2) Authority Calculation: A magazine's Authority score depends on the number and quality of incoming links (in-links) pointing to its articles.
3) Hub Calculation: Its Hub score depends on the quality of outgoing links (out-links) it provides to other reputable resources.
4) Contextual Differentiation: If prestigious established journals (such as Nature or Science) link frequently to Magazine A's articles while rarely linking to Magazine B, Magazine A will achieve a much higher Authority rank.

Mentor tip

Explain both Hub (out-links) and Authority (in-links) scores and include a concrete example comparing the two scientific magazines.

(b)

4 marks

1) Lossless Compression for Text: Scientific articles and text files require lossless compression (e.g., ZIP/GZIP) because losing or altering even a single character or numerical digit would corrupt the meaning and accuracy of the text.
2) Lossy Compression for Images: Photographs and illustrations can use lossy compression (e.g., JPEG) to achieve much smaller file sizes and faster downloads, as minor reductions in pixel detail are imperceptible to the human eye.

Mentor tip

Match lossless compression to text (where exact accuracy is mandatory) and lossy compression to images (where minor quality loss is tolerable for smaller size).

(c)

3 marks

Search engines track individual user context—including past browsing history, cookies, and geographic location—to personalize rankings.
Ambiguous Query Example: If two users type the search query 'Jaguar' (or 'Paris Hilton' / 'Python'), a user whose browsing history consists of automotive websites will receive car manufacturer links at the top.
Conversely, a user who frequently reads wildlife biology or programming tutorials will be prioritized links to the animal or the Python programming language.

Mentor tip

Identify user profiling (browsing history/cookies) and provide a concrete ambiguous search term that yields different top results for two user profiles.

(d)

4 marks

1) Grid Computing: Pools idle processing power across many networked computers to solve massive computational problems—for example, analyzing radio telescope data (SETI@home) or protein folding.
2) Peer-to-Peer (P2P) Network: Decentralizes file storage and bandwidth across equal participant nodes without a central server—for example, distributing large open-source Linux ISO files or media via BitTorrent.

Mentor tip

Name two distinct distributed architectures (e.g., Grid computing and Peer-to-Peer networks) and pair each with a clear real-world application.

Question 11

15 marks

Cloud-only laptops, decentralized web governance, and intellectual property protection measures

Web science

(a)(i)

2 marks

1) Software applications run remotely on cloud servers (accessed via a web browser) rather than being installed locally.
2) User files and data are stored on remote cloud storage rather than on a large local hard drive.

Mentor tip

Highlight the two core cloud shifts: remote web-based applications and remote online file storage.

(a)(ii)

4 marks

1) Connectivity Dependence: Because applications and files reside online, the user cannot access their work or use most software features when Wi-Fi/internet connectivity is unavailable or slow.
2) Software Restriction and Data Privacy: Users are locked into the cloud provider's web ecosystem (unable to install arbitrary local desktop software) and must entrust sensitive personal documents to third-party servers.

Mentor tip

Outline two distinct user drawbacks (such as offline uselessness, bandwidth latency, software lock-in, or cloud security) with an explanation for each.

(b)

5 marks

Positive Aspect (Arguments against central control):
- Promotes freedom of expression and open innovation: Information flows freely across borders without a single government or authority censoring viewpoints, and small businesses can publish globally without gatekeepers.
Negative Aspect (Arguments for control):
- Proliferation of harmful or illegal content: Absence of central oversight allows cybercrime, hate speech, terrorist propaganda (e.g., bomb-making guides), and unverified misinformation to spread rapidly.
Balanced Conclusion:
- While lack of central control requires coordinated international laws to combat criminal abuse, preserving decentralized openness remains vital to democracy and global knowledge sharing.

Mentor tip

Include a developed positive point, a developed negative point, and a reasoned concluding synthesis for full marks.

(c)

4 marks

1) Intellectual Property Example: Academic essays/theses (or digital music recordings).
2) Protection Measure: Automated plagiarism detection databases like Turnitin (or ISP monitoring and Digital Rights Management / P2P tracking for music).
3) Evaluation of Effectiveness: Turnitin cross-checks student submissions against billions of web pages and academic papers, generating a statistical similarity report that strongly deters and detects copy-paste plagiarism.
4) Limitation: However, it struggles to detect translated plagiarism from other languages, ghostwritten original essays, or AI-paraphrased text (and in music, P2P piracy persists via VPNs despite legal streaming growth).

Mentor tip

Name a specific IP type, describe a concrete technical/legal protection measure, and evaluate both its successes and its limitations.

Question 12

10 marks

Bowtie web graph structure, strongly connected core, diameter, and power-law inlink distribution

Web science

(a)(i)

1 mark

The Strongly Connected Core (SCC) is the central subset of web pages where any page can be reached from any other page via a directed path of hyperlinks.

Mentor tip

Emphasize mutual reachability: every node in the SCC can navigate to every other node in the SCC.

(a)(ii)

1 mark

The diameter of a graph is the length of the longest shortest-path (maximum geodesic distance) between any pair of connected nodes (or the average shortest path length).

Mentor tip

State 'maximum shortest path between any two nodes' (or average shortest path).

(b)

2 marks

Tubes are directed hyperlink pathways that connect pages in the IN component directly to pages in the OUT component.
Crucially, traffic travelling through a tube bypasses the Strongly Connected Core (SCC) entirely.

Mentor tip

Specify both endpoints (from IN pages to OUT pages) and that the path bypasses the SCC.

(c)(i)

1 mark

The IN component of the bowtie structure.

Mentor tip

Newly created resource lists with outgoing links belong to the IN component.

(c)(ii)

2 marks

Because the course resource lists were only recently published, external websites have not yet discovered or linked back to them (zero or very few in-links).
However, the resource lists themselves contain many outgoing hyperlinks pointing to established reference pages inside the SCC.

Mentor tip

Combine both facts: they are too new to have incoming links, but they point outward to established core sites.

(d)

3 marks

The distribution follows a scale-free power law of the form $y = a x^{- c}$ (where frequency is inversely proportional to a power of the number of inlinks).
Preferential Attachment ('rich-get-richer'): A tiny minority of well-known authority hubs attract millions of incoming links, while the vast majority of web pages ('the long tail') receive very few links.
As the number of inlinks increases, the number of pages possessing that many links drops off exponentially faster.

P (k) \propto k^{- γ}

Mentor tip

Mention the inverse power-law relationship ( $y = a x^{- c}$ ), the 'long tail' of pages with few links, and preferential attachment.

Question 13

10 marks

Social bookmarking, folksonomies vs search engine indexing, and ontologies for climate change research

Web science

(a)

2 marks

Social bookmarking services allow users to save, organize, and share links to web resources online.
When users apply their own free-form keyword tags to categorize those bookmarks, the aggregated crowd-sourced tag collection forms a folksonomy.

Mentor tip

Explain that user tagging within social bookmarking platforms generates a folksonomy.

(b)

4 marks

Human Relevance vs Algorithmic Errors:
- Folksonomy tags are chosen by human readers who understand the nuanced meaning, tone, and context of a resource, whereas automated search engine crawlers rely on keyword frequency algorithms that can misclassify pages.
Vocabulary Inconsistency vs Algorithmic Consistency:
- However, folksonomies suffer from human inconsistency—such as typos, synonyms, plurals, and personal bias—whereas automated search engine indexing applies deterministic, uniform rules across all pages.

Mentor tip

Compare both sides: human contextual accuracy vs crawler mislabelling, and human tag inconsistency vs algorithmic uniformity.

(c)

4 marks

1) Standardized Vocabulary and Relationships: Ontologies formally define domain concepts, data types, and hierarchical relationships (e.g., linking 'CO2 emissions', 'sea surface temperature', and 'glacial melt').
2) Machine Readability: By annotating climate datasets with shared ontological metadata (RDF/OWL), disparate databases become machine-readable and semantically interoperable.
3) Automated Global Aggregation: Automated software agents can seamlessly aggregate and cross-reference climate records from meteorological agencies, satellites, and universities worldwide.
4) Enhanced Scientific Insight: Researchers can run complex cross-disciplinary queries and climate models on unified data, improving predictive accuracy and policy decisions.

Mentor tip

Trace the chain: formal shared vocabulary -> machine-readable interoperability -> automated multi-source data integration -> deeper climate insights.

Option D: Object-oriented programming · Option

Questions 14–17 · 65 marks

Question 14

15 marks

Train and RollingStock OOP hierarchy: constructors, polymorphism, UML, and array stack operations

Object-oriented programming

(a)

2 marks

A constructor is a special method invoked automatically when a new instance of a class is instantiated.
Its primary function is to allocate memory for the new object and initialize its instance variables to valid starting values.

Mentor tip

Mention both object instantiation (creation) and initializing member variables.

(b)

3 marks

Advantage: Polymorphism allows external code to invoke the same method signature (such as getWeight()) across a collection of different RollingStock subclasses without needing to know their exact runtime subtype.
Contextual Example: Both Engine and Wagon inherit from RollingStock and implement getWeight().
Elaboration: Calling getWeight() on an Engine returns its fixed 120,000 kg base weight via super.getWeight(), whereas calling getWeight() on a Wagon overrides the method to sum the empty wagon's 32,000 kg weight plus the dynamic weights of all loaded Parcel objects.

Mentor tip

Connect the general benefit of polymorphism (unified interface with specialized behavior) directly to Engine.getWeight() vs Wagon.getWeight().

(c)

3 marks

Draw a 3-compartment UML class box for Train:
1) Class Name: Train
2) Private Attributes (-):
- mEngines : Engine[]
- mWagons : Wagon[]
- mEngineCount : int
- mWagonCount : int
- mTrainNumber : int
- mWeight : double
3) Public Methods (+):
+ Train(number : int)
+ addEngine(newEngine : Engine) : void
+ removeEngine() : Engine
+ addWagon(newWagon : Wagon) : void
+ removeWagon() : Wagon
+ getWeight() : double

Mentor tip

List all six private fields with '-' and all six public methods/constructors with '+' inside a 3-section UML box.

(d)

2 marks

Construct a public accessor method returning mWagonCount:
public int getNumberOfWagons() {
return mWagonCount;
}

Mentor tip

Ensure the return type is int and the method returns the mWagonCount counter.

(e)

5 marks

Check that mWagonCount > 0 before decrementing and returning the top Wagon element; return null if the train has no wagons:
public Wagon removeWagon() {
if (mWagonCount > 0) {
mWagonCount--;
return mWagons[mWagonCount];
} else {
return null;
}
}

Mentor tip

Mirror the logic of removeEngine(), adding an if (mWagonCount > 0) guard to prevent an ArrayIndexOutOfBoundsException.

Question 15

15 marks

Standard library collections, team software development, and refactoring Parcel with Address classes

Object-oriented programming

(a)

2 marks

Saves development time through reliable code reuse:
Developers can immediately use pre-built, thoroughly tested, and optimized data structures (like ArrayList or LinkedList) without having to write and debug them from scratch.

Mentor tip

Emphasize time saved and reliability gained from reusing pre-tested library code.

(b)

4 marks

1) Role Specialization: In a team, developers can focus on their specific areas of expertise (e.g., UI design, backend algorithms, or quality assurance testing), whereas a solo programmer must handle every aspect alone.
2) Parallel Development via Information Hiding: By agreeing on public class interfaces and encapsulating internal details, multiple team members can code different classes simultaneously, though teams require formal communication and version control.

Mentor tip

State two clear differences (such as specialist roles and parallel modular development) and explain each.

(c)(i)

1 mark

String

Mentor tip

Street names contain alphabetic characters and spaces.

(c)(ii)

1 mark

int (or String if alphanumeric building numbers like '12B' are considered).

Mentor tip

int is the standard primitive type expected for a building number.

(c)(iii)

1 mark

boolean

Mentor tip

A true/false flag is always represented by a boolean.

(d)

3 marks

1) Change the data types of destinationAddress and originAddress from String to Address (and encapsulate them as private fields).
2) Update the Parcel constructor parameters (or initialization logic) to receive Address objects.
3) Add public getter and setter methods for destinationAddress and originAddress referencing the Address type.

Mentor tip

Mention updating the field declarations, the constructor, and the accessor/mutator methods.

(e)

3 marks

Use inheritance by defining OriginAddress and DestinationAddress as subclasses that extend the base Address superclass.
Both subclasses automatically inherit shared attributes (street name, building number, city, country, isBusiness) without rewriting them.
DestinationAddress adds only its specific deliveryInstructions attribute, while OriginAddress adds only its collectionLocation indicator.

Mentor tip

Explain that extending the base Address class via inheritance shares common fields while isolating unique attributes in each subclass.

Question 16

15 marks

Tracing array operations and implementing getWeight() across Wagon and Train classes

Object-oriented programming

(a)(i)

2 marks

Engine B (ID 7) is added at index 0, followed by new Engine(9) at index 1:
Index [0]: Engine ID 7
Index [1]: Engine ID 9
Indices [2..5]: null / empty

Mentor tip

Show the array indices [0] and [1] containing Engine 7 and Engine 9 in exact insertion order.

(a)(ii)

1 mark

Mentor tip

Two engines were added and none were removed, so mEngineCount equals 2.

(a)(iii)

2 marks

Initially Wagons 23, 66, and 71 are placed at indices [0], [1], and [2].
Two removeWagon() calls decrement mWagonCount from 3 to 1 (removing 71 and 66), and addWagon(new Wagon(214)) places Wagon 214 at index [1]:
Index [0]: Wagon ID 23
Index [1]: Wagon ID 214

Mentor tip

Trace LIFO stack behavior: adding 23, 66, 71, popping twice leaves 23 at [0], and pushing 214 places it at [1].

(b)

4 marks

Start with the empty wagon's base weight via super.getWeight() (or 32000), loop from 0 to mParcelCount - 1 to add each parcel's weight, and return the sum:
public double getWeight() {
double totalWeight = super.getWeight();
for (int i = 0; i < mParcelCount; i++) {
totalWeight += mParcels[i].getWeight();
}
return totalWeight;
}

Mentor tip

Don't forget to include the empty wagon's own weight (super.getWeight()) before summing the mParcels array up to mParcelCount.

(c)

4 marks

Initialize totalWeight to 0, sum the weights of all active engines (0 to mEngineCount - 1), sum the weights of all active wagons (0 to mWagonCount - 1, where Wagon.getWeight() already includes its parcels), and return the total:
public double getWeight() {
double totalWeight = 0;
for (int i = 0; i < mEngineCount; i++) {
totalWeight += mEngines[i].getWeight();
}
for (int i = 0; i < mWagonCount; i++) {
totalWeight += mWagons[i].getWeight();
}
return totalWeight;
}

Mentor tip

Calling mWagons[i].getWeight() automatically returns the combined weight of the wagon and its parcels.

(d)

2 marks

The two getWeight() methods are defined within separate class scopes (Train vs Wagon).
The compiler resolves which getWeight() method to invoke based on the specific class type of the object reference on which the method is called.

Mentor tip

Explain that each class has its own namespace/scope, so the compiler binds the method call to the object's class type.

Question 17

20 marks

Refactoring Train to use a custom linked list of Wagons, LIFO Stack ADT for parcels, and coding conventions

Object-oriented programming

(a)

3 marks

1) Modify the Wagon class to include a self-referential pointer/reference attribute (e.g., public Wagon nextWagon;) pointing to the next wagon node.
2) Replace the mWagons[] array in the Train class with a head reference variable (e.g., private Wagon firstWagon;) initialized to null.
3) Rewrite the Train methods (addWagon, removeWagon, getWeight) to traverse and manipulate node pointers rather than array indices.

Mentor tip

Identify the three structural changes: adding a 'next' pointer in Wagon, replacing the array with a 'head' pointer in Train, and updating the list-handling methods.

(b)

3 marks

Point the new wagon's nextWagon reference to the current head (firstWagon), then update firstWagon to reference newWagon:
public void addWagon(Wagon newWagon) {
newWagon.nextWagon = firstWagon;
firstWagon = newWagon;
mWagonCount++;
}

Mentor tip

Always assign newWagon.nextWagon = firstWagon BEFORE overwriting firstWagon = newWagon so you don't lose the rest of the list.

(c)

5 marks

1) Method Parameter: Take the target wagon's unique ID number (int targetID) as a parameter.
2) Head Initialization: Start traversal pointers (current and previous) at the head of the list (firstWagon).
3) Search Loop: Traverse the linked list node-by-node while current != null, comparing current.getWagonID() against targetID.
4) Unlinking Node: If a match is found, bypass the node by setting previous.nextWagon = current.nextWagon (or updating firstWagon = current.nextWagon if removing the head node) and return the removed Wagon.
5) Not Found / Empty Case: If traversal reaches the end of the list (current == null) without a match, return null.

Mentor tip

Cover all 5 steps: taking the ID parameter, starting at firstWagon, looping and checking IDs, relinking pointers to bypass the target node, and handling the not-found/empty case.

(d)

1 mark

Stack (Last-In, First-Out / LIFO dynamic data structure).

Mentor tip

First-in, last-out (single-door loading/unloading) is the exact definition of a Stack.

(e)(i)

2 marks

Push the new Parcel object onto the stack instance named 'model':
public void addParcel(Parcel newParcel) {
model.push(newParcel);
}

Mentor tip

Use the standard stack push() operation on the 'model' object.

(e)(ii)

2 marks

Pop and return the top Parcel object from the stack instance named 'model':
public Parcel getParcel() {
return (Parcel) model.pop();
}

Mentor tip

Use the standard stack pop() operation to remove and return the most recently loaded parcel.

(f)

4 marks

1) Code Style Conventions (Indentation, Whitespace, and Comments):
- Consistent indentation and explanatory comments clearly expose control-flow nesting and algorithmic intent, making code faster to read, debug, and maintain.
2) Naming Conventions (Meaningful Identifiers and Prefixes):
- Descriptive variable and method names (such as mWagonCount or getWeight() rather than x or f1) make code self-documenting, reducing cognitive load and preventing misunderstandings across collaborative programming teams.

Mentor tip

Address both style conventions (indentation/comments) and naming conventions (meaningful identifiers), explaining how each reduces maintenance cost in teams.

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, Q13
Object-oriented programming65 marks · 25%
Q14, Q15, Q16, Q17

Mentor insights

Traps to watch for

01
Databases: Global retail data warehouse: ETL pipelines, association vs class · Q4
State the core advantage (integrated historical trend analysis) and give a concrete retail example.
02
Modelling and simulation: Chatbot machine learning paradigms and cognitive care robots for · Q8
Mention both receiving human language input and generating conversational language output.
03
Web science: Internet vs World Wide Web, URL structure, Common Gateway Interfa · Q9
Tie your distinction directly to the prompt's scenario (physical network link between SA and UK vs the newspaper's web pages/media).
04
Object-oriented programming: Refactoring Train to use a custom linked list of Wagons, LIFO Sta · Q17
Identify the three structural changes: adding a 'next' pointer in Wagon, replacing the array with a 'head' pointer in Train, and updating the list-handling methods.

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