Loading…

Worked solutions

IB Computer Science HL · Paper 2

Computer Science, Higher Level

Session: November 2019
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 2019 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

14 marks

One-to-7 educator forum registration, ACID properties, DBA tasks, and composite keys

Databases

(a)

1 mark

Splitting the name into First Name and Last Name ensures attribute atomicity (1st Normal Form) and enables fast, accurate sorting or searching by surname.

Mentor tip

State either attribute atomicity or the ability to sort/search directly by Last Name.

(b)

2 marks

The form demands excessive sensitive personal details (such as Date of Birth, Place of Birth, and Country of Birth) that are unnecessary simply to participate in an educator forum.
Teachers may refuse to register due to privacy concerns, fear of identity theft if the database is breached or shared with third parties, or frustration with the time required to complete the form.

Mentor tip

Connect excessive personal data collection to privacy/identity-theft risks or lower user registration rates.

(c)

2 marks

Transaction atomicity ensures that all database operations triggered when submitting the registration form either execute completely as a single indivisible unit or do not occur at all.
This prevents partial registration records from being saved if a network drop or system error interrupts submission, preserving database integrity.

Mentor tip

Explain 'all-or-nothing' execution and how it prevents incomplete educator profiles from corrupting the table.

(d)

3 marks

Transaction isolation ensures that when multiple teachers submit forum replies simultaneously, concurrent transactions execute independently and produce the same result as sequential processing.
When a teacher's post transaction updates a discussion thread, the DBMS places a lock on the relevant table rows so another concurrent post cannot overwrite or displace it before the first transaction commits.
A transaction log is also maintained prior to commit so that if an error occurs mid-post, the transaction rolls back cleanly to the previous valid state.

Mentor tip

Link row/record locking, independent sequential-equivalent execution, and transaction log rollback to preventing lost or garbled forum posts.

(e)

2 marks

Task 1: Configuring user access permissions, passwords, and security policies.
Task 2: Performing scheduled database backups, disaster recovery planning, data cleansing, or applying system patches/upgrades.

Mentor tip

List any two distinct operational DBA responsibilities such as managing access control/security and performing regular backups/recovery.

(f)

4 marks

With 250,000 educators registering, any composite primary key must combine enough attributes to guarantee 100% uniqueness across all records.
The DBA must consider demographic overlaps: for instance, combining First_Name + Last_Name + DOB might still produce collisions if many teachers come from the same country or age group.
Adding attributes that are invariant for a person (or incorporating a system-generated unique identifier/phone number alongside name and birth details) helps prevent a single educator from registering multiple accounts under different email addresses.
However, using too many fields in a composite key increases index overhead and makes foreign-key joins more cumbersome.

Mentor tip

Discuss uniqueness across 250,000 users, which specific form fields could be combined, and the risk of demographic collisions (shared names/birthdates).

Question 2

16 marks

Linnaean bird taxonomy ERD, validation vs verification, schemas, and referential integrity

Databases

(a)

2 marks

Identify the three hierarchical taxonomic entities: CLASS, ORDER, and FAMILY.
Establish the two 1-to-many (1:M) relationships: one CLASS contains many ORDERS, and one ORDER contains many FAMILIES.
Construct the top-down ERD:
CLASS (1) ----< (contains / has) >---- (M) ORDER (1) ----< (contains / holds) >---- (M) FAMILY

Mentor tip

Show both 1:M relationships clearly labelled in hierarchy from CLASS to ORDER and from ORDER to FAMILY.

(b)(i)

2 marks

Apply a range check or data type check on the beak length field.
For example, enforce a numeric (float/integer) data type and a range check between 5 cm and 60 cm so negative numbers, text strings, or implausible measurements are automatically rejected.

Mentor tip

Name a specific validation rule (such as range check or type check) and give a concrete numerical example for spoonbill beak length.

(b)(ii)

2 marks

Use double entry or visual proofreading against the original field research notes.
For example, requiring the researcher to type the spoonbill beak measurement twice (and flagging any mismatch) ensures that transcription typos are caught during entry.

Mentor tip

Remember: validation checks if data is reasonable/properly formatted via rules, whereas verification (double entry, proofreading) checks if data was copied accurately from the source.

(c)

2 marks

Data consists of raw, uncontextualized figures entered into a cell - for example, storing the raw number '18' in the database.
Information is produced when that stored figure is queried and interpreted within the database context and field metadata - for example, understanding that '18' represents 'a spoonbill beak length of 18 cm'.

Mentor tip

Contrast the raw numerical value stored in the table ('18') with the contextualized meaning retrieved from the database ('beak length of 18 cm').

(d)

2 marks

Characteristic 1: Provides a high-level, implementation-independent overview of the entire database architecture.
Characteristic 2: Defines the core business entities/concepts and the primary relationships between them (often represented via an ERD or Data Structure Diagram).

Mentor tip

Focus on high-level abstraction of entities and relationships independent of physical storage details.

(e)

3 marks

Data modelling creates a clear visual blueprint of entities, attributes, and relationships so scientists and database developers share a common understanding before implementation.
It translates conceptual and logical schemas into a sound physical database structure while establishing consistent naming conventions.
Without thorough modelling, the taxonomic database could suffer from structural flaws, data redundancy, and broken integrity constraints.

Mentor tip

Link stakeholder alignment and schema design (conceptual -> logical -> physical) to preventing redundancy and integrity errors.

(f)

3 marks

Referential integrity enforces the constraint that every foreign key value in a child table must correspond to an existing primary key value in the parent table.
In the bird taxonomy database, this prevents a researcher from inserting the 'Spoonbill' family linked to the order 'Ciconiiformes' unless 'Ciconiiformes' already exists as a valid primary key in the ORDER table.
It also ensures that updating or deleting a taxonomic order automatically cascades or blocks orphaned family records, preventing update and deletion anomalies.

Mentor tip

Explicitly reference the Spoonbill / Ciconiiformes relationship to earn all 3 marks.

Question 3

15 marks

Trucking On haulage database normalization to 3NF and SQL join queries

Databases

(a)

2 marks

Characteristic 1: Cells contain non-atomic, multi-valued repeating groups of data within a single row (such as multiple Trailer IDs, spaces, and coupling dates for one truck).
Characteristic 2: The table lacks a single primary key that uniquely identifies each individual coupling record without duplication.

Mentor tip

Point out repeating multi-valued attributes in cells and the absence of atomic tuples / unique row keys.

(b)

2 marks

Data redundancy means identical facts (such as Trailer TR9 having a space of 30 cubic metres, or Truck make 'Fodel' using 'Diesel') are duplicated across multiple rows.
If an attribute is modified later, failing to update every duplicate instance causes update anomalies and inconsistent operational records.

Mentor tip

Cite duplicated values in the LORRY table and explain how partial updates lead to data inconsistency.

(c)

4 marks

Step 1: Select the two required tables: LORRY and TRAILER.
Step 2: Join the tables on their shared identifier: LORRY.TruckID = TRAILER.TruckID.
Step 3: Apply the filter condition WHERE Trailerspace > 60.
Step 4: Project the Driver and Tel fields (SQL: SELECT Driver, Tel FROM LORRY INNER JOIN TRAILER ON LORRY.TruckID = TRAILER.TruckID WHERE Trailerspace > 60;).

Mentor tip

Include all four components: the two tables (LORRY, TRAILER), the join condition on TruckID, the filter (Trailerspace > 60), and the selected attributes (Driver, Tel).

(d)

7 marks

Separate repeating trailer assignments into a junction table (JOURNEY or COUPLING), isolate trailer capacity into a TRAILER table, and separate driver phone numbers into a DRIVER table to remove transitive dependencies:
DRIVER (Driver [PK], Tel)
TRUCK (TruckID [PK], Truckmake, Energysource, Driver [FK])
TRAILER (TrailerID [PK], Trailerspace)
JOURNEY (TruckID [PK, FK], TrailerID [PK, FK], Coupled_from [PK], Coupled_to) - or with a surrogate primary key: JOURNEY (JourneyID [PK], TruckID [FK], TrailerID [FK], Coupled_from, Coupled_to).

Mentor tip

Ensure all 4 relations (DRIVER, TRUCK, TRAILER, JOURNEY) are defined with underlined/marked Primary Keys and appropriate Foreign Keys.

Question 4

20 marks

Relational vs object-oriented databases, election data warehousing, and data mining ethics

Databases

(a)

4 marks

Advantage 1 - Mature industry standards andwidespread tooling: Relational databases (RDBMS) have decades of standardization (SQL), making it easier to recruit experienced staff, maintain stability, and use built-in reporting, security, and concurrency tools.
Advantage 2 - Simpler tabular visualization and modelling: Representing data as normalized two-dimensional tables with explicit relationships is easier to conceptualize and verify for data integrity than complex object-oriented class hierarchies.

Mentor tip

Present two clear 2-mark arguments: industry standardization/tooling/staffing and simpler tabular modelling compared to OOP complexity.

(b)

2 marks

Characteristic 1: Stores historical, time-variant data integrated from multiple heterogeneous sources.
Characteristic 2: Optimized for read-heavy analytical querying (OLAP), reporting, and data mining rather than transactional updates.

Mentor tip

Mention time-variant/historical multi-source storage and OLAP/analytical query optimization.

(c)

2 marks

Data gathered from disparate external sources (polling agencies, voter registries, social media) arrives in incompatible formats and conventions (for example, dates formatted as DD/MM/YY vs MM/DD/YYYY).
Transformation cleans, validates, and standardizes these records into a unified format so cross-source analytical queries execute accurately.

Mentor tip

Give a concrete format mismatch example (like date formats) and explain the need for standardization before loading.

(d)

2 marks

Voter sentiment during an election campaign is highly time-sensitive and shifts rapidly in response to political events.
Timestamping opinion polls allows analysts to track longitudinal shifts in voter preferences - for example, comparing support levels before and after a televised debate.

Mentor tip

Explicitly tie timestamping to the US presidential election scenario (such as comparing polls before and after a debate).

(e)

2 marks

Analytics firms use link analysis to uncover hidden associations and relationships between disparate entities or datasets.
In an election campaign, link analysis can connect voter preferences to lifestyle variables such as social media usage patterns, household size, or regional community memberships.

Mentor tip

Define link analysis as discovering associations across entities and give an election-specific example.

(f)

2 marks

Deviation detection identifies anomalous patterns or statistical outliers that depart from established historical baselines.
In the election context, it alerts campaign strategists to sudden, unexpected swings in pre-election polling or unusual spikes in negative voter sentiment within a specific swing state.

Mentor tip

Connect anomaly/outlier detection directly to spotting sudden shifts in state-level voter polling or sentiment.

(g)

6 marks

Advantages for campaign strategy: Data mining techniques like cluster analysis, classification, and association rules uncover subtle demographic patterns - such as grouping voters by income or identifying which policy issues resonate with specific demographics - allowing campaigns to tailor messaging and allocate advertising budgets efficiently.
Disadvantages and ethical risks: Election data mining relies on harvesting vast quantities of personal, behavioural, and social media data from citizens, often without their explicit informed consent.
Aggregating these datasets can strip away personal anonymity, expose sensitive private traits, and enable manipulative micro-targeting or political polarization.
Balanced conclusion: While data mining offers powerful analytical efficiency for campaign strategists, the societal and ethical risks outweigh the benefits unless strict data-protection laws, transparency, and anonymization safeguards are enforced.

Mentor tip

Include three components linked to the election scenario: specific data mining techniques/benefits (clustering/classification), privacy/anonymity drawbacks, and a reasoned conclusion.

Option B: Modelling and simulation · Option

Questions 5–8 · 65 marks

Question 5

17 marks

Compound interest investment planning, spreadsheet modelling, and progressive tax algorithms

Modelling and simulation

(a)

2 marks

Substitute P = 30,000, r = 0.10, and n = 2 into the compound interest formula T = P * (1 + r)^n:
T = 30,000 * (1 + 0.10)^2 = 30,000 * 1.21 = $36,300.

T = 30000 \times (1.10)^{2} = $36 300

Mentor tip

State the total investment value ( $36, 300), n o t j u s tt h e in t er es t e a r n e d ($ 6,300).

(b)

5 marks

Set up a 12-row iterative table or spreadsheet model with columns: Month (A), Start Balance (B), Monthly Interest (C), Monthly Deposit (D), and End Balance (E).
Initialize Month 1 Start Balance (B2) = 30000.
Calculate Month 1 Interest (C2) = B2 * 0.005 (since 0.5% = 0.005) and set Monthly Deposit (D2) = 1000.
Calculate Month 1 End Balance (E2) = B2 + C2 + D2 (or B2 * 1.005 + 1000).
For Months 2 through 12, set each month's Start Balance equal to the previous month's End Balance (e.g., B3 = E2) and copy the formulas down to row 13 (Month 12).

Mentor tip

Clearly show the 5 key elements in a spreadsheet layout: initial $30, 000 ba l an ce, 0.005 m o n t h l y in t er es t m u l t i pl i c a t i o n,$ 1,000 monthly addition, end-of-month sum, and carrying the balance forward for 12 months.

(c)

6 marks

Initialize PRINCIPAL = 30000, RATE = 1.005, and running TAX = 0.
Loop 12 times (for each month): compute the new balance with interest, determine the monthly PROFIT (interest earned), add the $1,000 deposit, and apply the conditional tax rate (25% if total <= 40,000, else 40%).
After the 12-month loop finishes, subtract the accumulated TAX from PRINCIPAL and output the final value:
PRINCIPAL = 30000 RATE = 1.005 TAX = 0 loop MONTH from 1 to 12 OLD_PRINCIPAL = PRINCIPAL INVESTMENT = OLD_PRINCIPAL * RATE PROFIT = INVESTMENT - OLD_PRINCIPAL PRINCIPAL = INVESTMENT + 1000 if PRINCIPAL <= 40000 then TAX = TAX + (PROFIT * 0.25) else TAX = TAX + (PROFIT * 0.40) end if output "End of month ", MONTH, " fund value: ", PRINCIPAL end loop FINAL_VALUE = PRINCIPAL - TAX output "Final Investment Value after tax: ", FINAL_VALUE

Mentor tip

Remember that tax is calculated each month on that month's profit and accumulated in a running total, but only deducted from the fund total at the end of the 12 months.

(d)

4 marks

Mathematical modelling enables investment companies to run 'what-if' scenarios by adjusting key variables before launching a financial product.
Analysts can track month-by-month fund growth and test how varying the interest rate impacts company profitability and client payouts.
They can also simulate how changes in customer monthly contributions or government tax brackets affect final investment returns.

Mentor tip

Mention 'what-if' scenario testing and give specific examples of varying interest rates, monthly deposits, and tax rates to observe their impact on profits.

Question 6

16 marks

Supermarket regional distribution centre location simulation

Modelling and simulation

(a)

4 marks

Reason 1 - Avoiding prohibitive capital cost: Constructing a physical distribution warehouse is extremely expensive, so it is financially impossible to build all four centres just to test which site performs best.
Reason 2 - Time compression and scenario testing: A computer simulation can model years of regional delivery traffic, roadworks, and worst-case congestion scenarios across all four sites in hours rather than waiting years for physical construction.

Mentor tip

Provide two distinct reasons (cost savings over physical trial-and-error, and speed/flexibility to test 'what-if' traffic scenarios) with a 1-mark expansion for each.

(b)(i)

2 marks

Input 1: Geographical coordinates/distances from each prospective site to the regional supermarkets and supply ports/rail hubs (or road network capacities and travel times at different hours).
Input 2: Land purchase prices and local labour wage rates at each of the four locations.

Mentor tip

List concrete raw data inputs such as road distances/travel times to supermarkets, land costs, or local wage rates.

(b)(ii)

2 marks

Criterion 1: Lowest total capital construction cost and ongoing operational/shipping cost.
Criterion 2: Shortest average delivery transit time to distribute goods to the supermarkets (or shortest construction completion time).

Mentor tip

Distinguish between raw data inputs in (b)(i) and decision evaluation criteria in (b)(ii) such as total construction/running cost and delivery turnaround time.

(b)(iii)

6 marks

Evaluating capital construction cost and build time: The simulation takes inputs such as land price, local labour wages, worker headcount, and material delivery schedules for each of the four sites, running scenarios (such as supply delays or wage changes) to project total build cost and completion time.
Evaluating ongoing distribution running costs: The model combines distances to supermarkets, lorry fuel economy, fuel prices, and shipment frequencies, simulating scenarios like fuel price spikes or increased order volumes to calculate annual operating expenditure.
Evaluating delivery transit times: Using road network maps, loading/unloading durations, and time-of-day traffic data, the simulation tests worst-case traffic congestion or roadworks to rank the four locations against the cost and speed criteria.

Mentor tip

Explicitly connect the inputs from (b)(i) to the criteria in (b)(ii) by explaining how the simulation runs varied scenarios (fuel spikes, congestion, construction delays) to score and compare the four sites.

(c)

2 marks

Developing a highly detailed simulation can be expensive and time-consuming, potentially outweighing the benefits if one site is already obviously superior.
Furthermore, real-world logistics involve too many unpredictable confounding variables (such as sudden political changes, extreme weather, or unmodelled human factors), meaning an oversimplified simulation may produce misleading recommendations.

Mentor tip

Mention both the cost/complexity of building the simulation and the risk of inaccurate results due to omitted or unpredictable real-world variables.

Question 7

12 marks

Tibetan antelope migration satellite tracking and real-time 2D visualization

Modelling and simulation

(a)

4 marks

Item 1: Real-time GPS coordinates of tagged antelopes transmitted continuously from satellite trackers.
Item 2: Current movement speed of the herd and the time required for the herd to cross a checkpoint.
Item 3: Historical migration route coordinates and fixed map geometry of the QT Railway and QT Highway.
Item 4: Calculated predicted path and estimated time of intersection with the highway/railway.

Mentor tip

Include four distinct data items: live GPS positions, herd speed/crossing duration, historical/road map coordinates, and predicted intersection timing.

(b)

4 marks

Reason 1 - Sufficiency of 2D spatial mapping: Antelopes travel across the ground surface, so their proximity to the railway and highway is fully represented on a 2D planar map without needing vertical 3D elevation rendering.
Reason 2 - Real-time rendering speed and lower hardware demands: A 2D visualization is computationally simpler and updates rapidly in real time on standard hardware, whereas complex 3D rendering requires powerful GPUs and could introduce processing delays when issuing urgent traffic warnings.

Mentor tip

Give two explained points: ground movement only needs 2D planar representation, and 2D updates much faster in real time without heavy GPU overhead.

(c)

4 marks

Challenge of manual observation: The Hoh Xil Nature Reserve is a vast, remote, and inhospitable mountainous region spanning over 300 km with harsh weather and poor visibility, making it impossible to station enough human observers along the entire railway and highway.
Necessity of satellite-fed visualization: Automated satellite telemetry displayed on a real-time visualization model allows authorities to monitor herd movements remotely 24/7 and proactively halt road or rail traffic before collisions occur.

Mentor tip

Spend 2 marks explaining the physical difficulties of human monitoring in remote/mountainous Tibet, and 2 marks on how satellite-driven visualization solves this.

Question 8

20 marks

Optical character recognition (OCR) and neural networks for handwriting and predictive text

Modelling and simulation

(a)

2 marks

Printed documents may suffer from degraded print quality (faint ink, smudges, or skewed scans) or use ornate/unusual fonts.
As a result, visually similar glyphs - such as the digit '5' and letter 'S', or '1' and 'I' in sans-serif typefaces - can be misclassified by the OCR scanner.

Mentor tip

State a cause (poor print quality or decorative/sans-serif fonts) and give a specific character confusion example (such as '5' vs 'S' or '1' vs 'I').

(b)

2 marks

Unlike rigid template matching that requires an exact pixel-for-pixel match, an Artificial Neural Network (ANN) can be trained on thousands of varied handwriting samples to generalize across different stroke styles.
The ANN extracts local sub-features of a character and calculates probabilistic similarity, allowing it to recognize a skewed or imperfect handwritten 'X' without requiring the entire grid to match Figure 4 identically.

Mentor tip

Emphasize the ANN's ability to learn from training examples and recognize partial/generalized stroke features rather than requiring an exact pixel match.

(c)

4 marks

Step 1 - Feature filtering / Convolution: The ANN breaks the 9x9 pixel grid into smaller local patches (such as 3x3 pixel filters representing diagonal lines and crossing intersections) and slides these filters across the image.
Step 2 - Pixel multiplication and scoring: At each position, the network multiplies image pixel values (+1 or -1) by the filter weights, sums them, and divides by the patch size to compute a feature match score (where 1.0 indicates a perfect local match).
Step 3 - Activation and Pooling: A Rectified Linear Unit (ReLU) layer zeroes out negative values to simplify computation, and pooling layers downsample the feature maps to make detection invariant to slight shifts in position.
Step 4 - Classification output: The stacked feature maps are fed into fully connected output layers and compared against stored character classes to identify the pattern as the letter 'X'.

Mentor tip

Walk through the convolutional neural network stages: sliding small pixel filters across the grid, computing local match scores, applying ReLU/pooling, and matching the combined features to 'X'.

(d)

2 marks

Feature 1: A stored lexicon / dictionary of valid words and grammatical syntax rules (nouns, verbs, adjectives).
Feature 2: Recurrent memory / feedback loops that retain the preceding words in the sentence as context inputs.

Mentor tip

Mention a vocabulary dictionary/grammar model and memory of previous words in the current sentence.

(e)

6 marks

Step 1 - Dual context inputs: The neural network receives two sets of inputs into its input layer: the sequence of preceding words ('The child is feeling') and the newly typed character prefix ('hu').
Step 2 - Weighted hidden layers: These inputs are combined and passed through multiple hidden layers where synaptic weights reflect learned statistical co-occurrence frequencies from training text.
Step 3 - Non-linear activation: Hidden nodes apply non-linear activation functions (such as a sigmoid or softmax function) to compute the conditional probability of prospective words.
Step 4 - Recurrent feedback and backpropagation: When the user types 'hu', the network merges the grammatical context (expecting an adjective after 'is feeling') with the prefix constraint ('hu*'), ranking 'hungry' as the highest-probability output while using feedback loops to refine future predictions.

Mentor tip

Trace the data flow through the ANN: input of prior words + input of new letters ('hu') -> weighted hidden layers -> sigmoid/probability calculation -> merging context and prefix to output 'hungry'.

(f)

4 marks

Problem 1 - Slang, text-speak, and poor training corpora: If users type informal abbreviations, slang, or 'text-speak' not present in the training dictionary (or if unsupervised learning absorbs misspelled user inputs), the ANN will fail to suggest appropriate words.
Problem 2 - Vanishing or exploding gradients during training: When backpropagating error signals across many recurrent steps or deep layers, repeated multiplication by the weight matrix can cause the gradient signal to shrink toward zero (vanishing gradient) or grow uncontrollably (exploding gradient), stalling or destabilizing learning.

Mentor tip

State two distinct training challenges: vocabulary/text-speak quality issues and the mathematical vanishing/exploding gradient problem.

Option C: Web science · Option

Questions 9–13 · 65 marks

Question 9

16 marks

OrderYourFoods.com web protocols, server-side PHP email validation, and black hat SEO

Web science

(a)(i)

1 mark

HTTPS (Hypertext Transfer Protocol Secure).

Mentor tip

Commercial ordering websites handling customer addresses and payments use HTTPS.

(a)(ii)

4 marks

Function 1 - Standardized communication and packet switching: Protocols define strict rules for formatting, addressing, and routing data packets across heterogeneous networks.
Function 2 - Data integrity and error correction: Protocols use checksums and acknowledgements to detect corrupted or dropped packets and trigger retransmission over unreliable channels.
Function 3 - Flow control and congestion management: Protocols regulate transmission rates so a fast server does not overwhelm a slower client device, and prevent network deadlock.
Function 4 - Security and encryption: Protocols like HTTPS/TLS authenticate servers and encrypt data in transit.

Mentor tip

Explain four distinct roles of network protocols: error checking/correction, flow control, congestion/deadlock management, and encryption/security.

(b)(i)

1 mark

Server-side scripting.

Mentor tip

PHP code executes on the web server rather than in the client's browser.

(b)(ii)

1 mark

State either built-in PHP function call with its arguments:
filter_var( $_{P} O S T [^{'} e mai l^{'}], F I L T E R_{V} A L I D A T E_{E} M A I L) - or mai l ($ to, $s u bj ec t,$ body, $h e a d er s, " - f ".$ from).

Mentor tip

Include both the function name (filter_var or mail) and the parameters inside the parentheses.

(b)(iii)

3 marks

Step 1: The PHP script retrieves the customer's submitted email address from the form via the $_POST['email'] associative array and assigns recipient/header variables.
Step 2: It validates the format of the submitted email address using filter_var($_POST['email'], FILTER_VALIDATE_EMAIL); if the email format is invalid, no further action or output occurs.
Step 3: If the email is valid, it calls the mail() function to send the subscription notification to info@orderyourfoods.com and outputs a confirmation message if mail() succeeds, or outputs an error message if sending fails.

Mentor tip

Be sure to note what happens in all branches: valid + mail sent -> confirmation echoed; valid + mail fails -> error echoed; invalid format -> nothing happens.

(c)

1 mark

Live online chat widget (or embedded social media links such as Facebook/Twitter, or a customer discussion forum).

Mentor tip

Any interactive web communication feature beyond static email/phone (such as live chat or social media integration) is valid.

(d)

5 marks

Short-term advantages: Black hat SEO techniques (such as keyword stuffing, hidden text, link farms, or doorway pages) can artificially inflate OrderYourFoods' ranking for 'food delivery' quickly, generating a short-term spike in website traffic and orders.
Long-term disadvantages: Black hat methods violate search engine guidelines and are considered deceptive and unethical.
Modern search engine crawlers actively detect manipulative patterns and will penalize or permanently blacklist (de-index) www.orderyourfoods.com from search results altogether.
Overall evaluation: While black hat SEO might yield brief initial visibility, the severe long-term risk of being banned from search engines and damaging the company's reputation makes it an unwise strategy compared to legitimate white hat SEO.

Mentor tip

Structure your evaluation with short-term visibility gains, long-term blacklisting/reputational risks, and a clear concluding judgement.

Question 10

15 marks

Search engines, PageRank vs HITS algorithms, cloud vs client-server, and image compression

Web science

(a)

1 mark

A search engine is a software system that queries an indexed database of web pages based on user-supplied keywords and returns a ranked list of relevant resources.

Mentor tip

Mention querying an indexed database of web pages to retrieve matching results.

(b)

4 marks

PageRank algorithm (Google): Evaluates the importance of a web page by recursively counting both the quantity and the quality (authority weight) of incoming links (in-links) pointing to it, under the assumption that authoritative sites receive more links from other important sites.
HITS algorithm (Ask - Hyperlink-Induced Topic Search): Evaluates pages dynamically for a query by calculating two distinct scores for every page: an Authority score (high number of in-links from hubs, indicating valuable content on the topic) and a Hub score (high number of out-links pointing to reputable authorities).

Mentor tip

Devote 2 marks to PageRank (count and quality of in-links) and 2 marks to HITS (distinguishing between Hubs with many out-links and Authorities with many in-links).

(c)

3 marks

Local client-server architecture: Hosted on dedicated servers maintained locally within the school/organization, providing low-latency local data transmission and direct physical security control for internal clients.
Cloud computing: Hosts data and applications on distributed remote server infrastructure accessed over the internet as an on-demand service.
Key difference: Cloud platforms can elastically scale storage and computing capacity up or down and allow ubiquitous remote access from anywhere, whereas local client-server networks have fixed hardware capacity and are managed at the organizational level.

Mentor tip

Define local client-server, define cloud computing, and state a clear operational contrast (such as elastic scalability, remote accessibility, or local latency/security).

(d)(i)

1 mark

Lossless compression.

Mentor tip

Lossless is the exact counterpart to lossy compression.

(d)(ii)

4 marks

Advantages: Lossy JPG compression drastically reduces image file sizes, allowing students to upload and download coursework documents much faster on the cloud platform and enabling smoother scrolling through multi-page essays.
Disadvantages: Because lossy compression permanently discards pixel data, high compression ratios introduce visible artifacts and blurriness, degrading visual quality when coursework is printed or when fine diagram details are examined.
Conclusion: Using compressed images is beneficial for general illustrative photos to keep coursework file sizes manageable, provided the compression ratio does not render essential academic diagrams illegible.

Mentor tip

Cover upload/download speed and document responsiveness (pros), reduced visual/print resolution (cons), and a contextual conclusion.

(e)

2 marks

Interoperability is the ability of different software applications, browsers, and operating systems to exchange and render information seamlessly.
Open standards are publicly available, royalty-free technical specifications (like the PNG format) that make interoperability across diverse platforms possible.

Mentor tip

Clearly distinguish the capability of systems working together (interoperability) from the publicly shared specifications that enable it (open standards).

Question 11

14 marks

World Wide Web network growth, browser and DNS functions, and intellectual property

Web science

(a)

3 marks

The World Wide Web is a scale-free distributed network where the total number of connected nodes (such as IB World Schools) grows exponentially.
However, due to small-world network topology and highly connected hubs, the average path length (number of routing hops between any two nodes) grows only logarithmically/linearly.
Consequently, schools worldwide can communicate and share resources with minimal latency regardless of physical distance, acting as a catalyst for global collaboration.

Mentor tip

Relate exponential node growth vs small hop counts / negligible latency to why global collaboration scales so effectively.

(b)

3 marks

Function - Fetching and rendering web pages (or bridging URLs to DNS):
The web browser takes the URL entered by the user, contacts a DNS server to resolve the domain into an IP address, and sends an HTTP/HTTPS request to the target web server.
It then receives the server's response files (HTML, CSS, JavaScript) and parses/interprets the markup code to render the visual, interactive web page on the user's screen.

Mentor tip

Trace the browser's end-to-end role: resolving the URL via DNS, fetching HTML/CSS/JS over HTTP(S), and parsing/rendering the page visually.

(c)

3 marks

Step 1: When a user enters a domain name into the browser, the request is sent to a Domain Name Server (DNS), which searches its local database for the matching IP address.
Step 2: If the domain is found, the DNS returns the IP address so the request can be routed to the destination web server.
Step 3: If the domain is not in its local database, the DNS queries other upstream DNS servers in the hierarchy until the IP address is resolved, or returns a 'site not found' error if no match exists anywhere.

Mentor tip

Include local database lookup, forwarding to other DNS servers on a cache miss, and returning either the resolved IP address or an error.

(d)

5 marks

Copyright issues: Because the World Wide Web makes digital text, images, and media effortlessly accessible across billions of pages, users can copy, redistribute, or publicly display works instantaneously - often infringing copyright laws unintentionally or due to ignorance of complex licensing terms.
Intellectual property (IP) issues: Web content frequently incorporates protected trademarks, patents, and proprietary designs; downloading, modifying, or remixing resources without the creator's authorization violates IP rights.
Jurisdictional complexity: Because the web is globally distributed across countries with conflicting copyright and IP statutes, an action that is legal in one jurisdiction may constitute infringement in another, making enforcement difficult.

Mentor tip

Address both copyright infringement (unauthorized copying/republishing due to ease of access) and broader Intellectual Property issues (modifying content, trademarks, and differing international laws).

Question 12

14 marks

OBI European offices directed graph, web connectivity, multimedia search, and collective intelligence

Web science

(a)

2 marks

Represent the three offices as vertices (nodes) A, B, and C.
Draw a bidirectional arrow between A and B (A <---> B) to represent their mutual project collaboration.
Draw unidirectional arrows from A to C (A ---> C) and from B to C (B ---> C) to represent reporting to the head office.

Mentor tip

Double-check arrow directions: A <---> B (two-way collaboration) and A ---> C, B ---> C (one-way reporting).

(b)

3 marks

Graph theory models the World Wide Web by representing individual web pages as vertices (nodes) and hyperlinks between pages as directed edges (arrows indicating 1-way or 2-way links).
In graph theory, connectivity measures whether a directed path exists to traverse from any given vertex to another vertex.
Applying graph theory to the web allows researchers to analyse bow-tie structure, identify isolated components vs the strongly connected core, and measure how many hyperlink hops separate web pages.

Mentor tip

Explicitly map web pages to vertices/nodes, hyperlinks to directed edges, and path traversal between vertices to web connectivity.

(c)

4 marks

Issue 1 - Reliance on accurate manual metadata/tags: Search engines typically index multimedia (images, audio, video) using surrounding text or assigned meta-tags; if contributors omit or mislabel tags, the non-text files will not be found by keyword queries.
Issue 2 - Computational difficulty of content-based similarity search: Searching directly by visual or acoustic content (such as reverse image search) requires complex feature-extraction algorithms because there are no explicit text tokens to match, and fewer search engines support efficient multimedia indexing.

Mentor tip

Explain two distinct challenges: missing/inaccurate meta-tags on multimedia files, and the algorithmic difficulty of content-based image/audio similarity matching.

(d)

5 marks

Advantages: Utilizing collective intelligence allows OBI to tap into a massive global pool of contributors working simultaneously, completing large-scale data gathering far more rapidly and at a much lower cost per unit than hiring full-time specialists.
Disadvantages: Because crowdsourced environments are decentralized and loosely regulated, it is difficult to enforce rigorous quality control, verify the accuracy or originality of submissions, or evaluate thousands of competing ideas without clear hierarchy.
Conclusion: Collective intelligence is highly effective for OBI when rapidly gathering broad, non-safety-critical information, provided Office A implements verification workflows to filter out low-quality contributions.

Mentor tip

Weigh speed, scale, and low cost (pros) against lack of quality control, unverified originality, and coordination overhead (cons), finishing with a contextual conclusion.

Question 13

6 marks

Southend Council ambient intelligence and IoT humanoid robot (Pepper) evaluation

Web science

6 marks

Environmental complexity and sub-optimal decisions: Guiding elderly residents with memory loss from care homes to local shops exposes Pepper to unpredictable outdoor environments (traffic, roadworks, weather) that may exceed its ambient intelligence training, leading to navigation errors with serious safety consequences.
IoT connectivity black spots: Pepper relies on continuous wireless communication with Internet of Things (IoT) sensors and cloud servers; network dead zones or latency spikes along the route could leave the robot and resident stranded.
Over-reliance vs distrust by vulnerable residents: Residents suffering from dementia might either distrust the robot and become distressed, or over-trust Pepper and delegate critical road-crossing judgements that the robot is not equipped to handle.
Privacy and data security concerns: Families and care home regulators may object to Pepper's cameras and sensors continuously recording and transmitting sensitive location and health data over the IoT.
Legal accountability and liability: If an elderly resident is injured in an accident while following Pepper's directions, legal liability between Southend Council, the care home, and the robot manufacturer remains ambiguous.

Mentor tip

Cover technical limitations (unpredictable outdoor environments, IoT black spots), human factors (over-trust by dementia patients), privacy of sensor data, and legal accountability in case of accidents.

Option D: Object-oriented programming · Option

Questions 14–17 · 65 marks

Question 14

16 marks

Supermarket point-of-sale (POS) simulation: POSsystem, POSline, and Cart array manipulation

Object-oriented programming

(a)

1 mark

Aggregation (a 1-to-many 'has-a' relationship: one POSsystem aggregates multiple POSline objects).

Mentor tip

State 'aggregation' (or 'has-a' / 1-to-many relationship).

(b)

2 marks

Draw a hierarchy or UML aggregation diagram showing one POSsystem connected to multiple POSline objects, and each POSline connected to multiple Cart objects:
POSsystem (1) ◇---- (many) POSline (1) ◇---- (many) Cart

Mentor tip

Clearly show the two levels of 1-to-many relationships: POSsystem -> multiple POSlines, and each POSline -> multiple Carts.

(c)

1 mark

An identifier is the unique symbolic name assigned by a programmer to a program element such as a variable, method, or class.

Mentor tip

Must mention that it is the 'name' of a variable, method, or class.

(d)

3 marks

A class is the abstract template or blueprint that defines the attributes (data fields) and methods (behaviours) of an entity.
An instantiation is a concrete object created from that class blueprint and allocated memory at runtime with specific attribute values.
In the provided UML, POSline is the class blueprint defining id, active, and line, whereas an individual checkout counter in the supermarket is an instantiation of POSline storing instantiated Cart objects inside its line array.

Mentor tip

Define class, define instantiation, and explicitly reference POSline and Cart from the UML diagram.

(e)

2 marks

Use the new keyword to allocate an array of 20 Cart references:
private Cart[] line = new Cart[20]; (or line = new Cart[20];)

Mentor tip

Include both the new keyword and the size specification Cart[20].

(f)

3 marks

Start at index i = 0 and advance through the line array until the first null (empty) slot is found, then assign newCart to line[i]:
public void joinLine(Cart newCart) { int i = 0; while (this.line[i] != null) { i++; } this.line[i] = newCart; }

Mentor tip

Searching for the first null entry (this.line[i] != null) locates the end of the current queue in the array.

(g)

4 marks

Save the Cart at index n in a temporary variable result to return at the end.
Loop from i = n up to 18, shifting each subsequent element one position to the left (line[i] = line[i + 1]).
Set the final slot line[19] = null to clear the duplicate trailing reference, and return result:
public Cart leaveLine(int n) { Cart result = this.line[n]; for (int i = n; i < 19; i++) { this.line[i] = this.line[i + 1]; } this.line[19] = null; return result; }

Mentor tip

Don't forget to set this.line[19] = null after shifting elements down, and loop up to i < 19 so line[i + 1] does not cause an ArrayIndexOutOfBoundsException.

Question 15

15 marks

POS simulation inheritance, class libraries, model refactoring, and distributed teams

Object-oriented programming

(a)

2 marks

Inheritance is the OOP mechanism whereby a new class (subclass/child) is derived from an existing class (superclass/parent), automatically acquiring all of the parent's attributes and methods.

Mentor tip

Mention both deriving a subclass from a superclass and inheriting its attributes and methods.

(b)

3 marks

Inheritance promotes code reuse and extensibility.
Because the parent class POSline already defines common attributes (id, active, line) and queue operations (joinLine, leaveLine), developers can easily derive specialized subclasses such as CashOnlyPOSline or ExpressTenItemPOSline without rewriting shared queue logic, reducing development and testing time.

Mentor tip

State the advantage (code reuse / extensibility / reduced maintenance), explain how the parent holds shared code, and tie it directly to POSline and specialized checkout counters.

(c)

4 marks

Advantage 1 - Faster development and code reuse: Pre-written library classes provide ready-made data structures and utilities so programmers do not have to reinvent standard algorithms from scratch.
Advantage 2 - High reliability and abstraction: Library classes have been thoroughly tested and debugged by thousands of developers, and can be used as black-box abstractions without needing to know their internal implementation details.

Mentor tip

Give two distinct advantages (such as saving development time, pre-tested reliability, or black-box abstraction) and elaborate on each.

(d)

2 marks

In the original model, POSline tightly couples a single checkout counter with its own Cart[] line array.
To allow one shared queue feeding multiple counters, the model must decouple the waiting line from the checkout counter - for example, by creating a separate Counter class (or an array of active counters inside POSline/POSsystem) that pulls the next Cart from a single shared Cart[] line.

Mentor tip

Explain that the checkout counter entity and the waiting line of carts must be separated so multiple counters can draw from one shared line.

(e)

4 marks

Point 1: Teams working across different international time zones experience communication delays and cannot easily hold synchronous face-to-face meetings, slowing down debugging and integration.
Point 2: Language barriers and cultural differences between distributed teams can cause misunderstandings of system specifications.
Point 3: Different teams may follow inconsistent coding conventions, naming styles, or data formats (such as date/currency representations), causing interface mismatches when modules are integrated.
Point 4: Coordinating version control and project management across remote sites is more complex.

Mentor tip

Make four clear points covering time zones, language/communication barriers, differing coding/format conventions, and coordination delays.

Question 16

14 marks

Boolean checkout flags, splitting a POSline queue, and open-source software ethics

Object-oriented programming

(a)

2 marks

A checkout counter has only two mutually exclusive states: open (true) or closed (false).
A boolean data type models binary true/false states directly while using the minimum possible memory (1 bit).

Mentor tip

Mention both the two-state nature of an open/closed counter and memory efficiency.

(b)

8 marks

Instantiate a new POSline object newLine to hold the transferred carts.
First pass - Copy every second cart (odd indices 1, 3, 5, ...) from oldLine into newLine using oldLine.getLine(i) and newLine.joinLine(...).
Second pass - Remove those copied carts from oldLine using oldLine.leaveLine(i): note that since leaveLine(i) shifts all subsequent carts down by 1 index, either iterate in reverse from the end of the line down to 0, or increment i by 1 on each removal in a forward pass!
Alternatively, combine copying and removal in a single loop using a boolean toggle flag:
public POSline split(POSline oldLine) { POSline newLine = new POSline(); int i = 0; while (i < 20 && oldLine.getLine(i) != null) { if (i % 2 == 1) { newLine.joinLine(oldLine.getLine(i)); } i++; } i--; // start from last valid index and remove odd indices in reverse while (i > 0) { if (i % 2 == 1) { oldLine.leaveLine(i); } i--; } return newLine; }

Mentor tip

Watch out for index shifting when calling oldLine.leaveLine(i)! Removing odd indices in reverse order (while i > 0) guarantees that shifting later elements never alters the indices of earlier elements waiting to be removed.

(c)

4 marks

Issue 1 - Intellectual property attribution and licensing compliance: Even though open-source code is publicly available, it is still intellectual property governed by licenses (such as GPL or MIT); the supermarket company has an ethical obligation to credit the original authors and respect license terms rather than passing off community work as proprietary.
Issue 2 - Customer security and reliability accountability: Because open-source code can be inspected by anyone (or may have been modified by unvetted contributors), failing to audit and test adapted modules thoroughly could expose supermarket customers to malware or payment data breaches.

Mentor tip

Describe two ethical issues with elaboration: proper attribution/licensing of open-source creators' work, and the duty of care to test open-source code so customer payment data isn't compromised.

Question 17

20 marks

Singly linked list of CartNode objects (POSlist), pointer manipulation, and coding conventions

Object-oriented programming

(a)

2 marks

An object reference is a variable that holds a pointer to the memory address where an instantiated object is stored.

Mentor tip

State both 'pointer/memory address' and 'where the object is stored in memory' for full marks.

(b)

4 marks

Declare Cart result = null; and check if the list is non-empty (head != null).
Extract the Cart object from the first node via head.getCart(), advance head = head.getNext() to unhook the first node, and return result:
public Cart removeFirst() { Cart result = null; if (head != null) { result = head.getCart(); head = head.getNext(); } return result; }

Mentor tip

Always extract head.getCart() before advancing head = head.getNext(), and return null safely when head == null.

(c)

2 marks

Trace the operations in sequence:
1) addLast(cart2), addLast(cart1), addLast(cart4) -> [cart2 -> cart1 -> cart4]
2) removeFirst() removes cart2 -> [cart1 -> cart4]
3) addLast(cart5), addLast(cart3) -> [cart1 -> cart4 -> cart5 -> cart3]
4) removeFirst() removes cart1 -> final list contains 3 nodes in order:
head ---> [ cart4 | • ] ---> [ cart5 | • ] ---> [ cart3 | null ]

Mentor tip

Include the head (root) pointer at the start, the three remaining nodes in exact FIFO order (cart4 -> cart5 -> cart3), and the terminating null pointer.

(d)

2 marks

A standard queue is a strict First-In, First-Out (FIFO) abstract data structure that only permits insertions at the rear (enqueue) and removals from the front (dequeue).
This is unsuitable for a real supermarket checkout line because impatient customers in the middle of the line may leave or switch to a newly opened counter at any position.

Mentor tip

Connect the strict FIFO restriction of a queue (only removing from the front) to the need for customers in the middle of a line to leave or switch lanes.

(e)

6 marks

Handle the special case where n == 1 (removing the first node) by calling removeFirst().
Otherwise, traverse the list using curr and prev pointers from i = 1 up to n so curr points to the nth CartNode and prev points to node n - 1.
Retrieve curr.getCart(), bypass the nth node via prev.setNext(curr.getNext()), and return the removed Cart:
public Cart leaveList(int n) { if (n == 1) { return removeFirst(); } CartNode prev = null; CartNode curr = head; int i = 1; while (i < n) { prev = curr; curr = curr.getNext(); i++; } Cart result = curr.getCart(); prev.setNext(curr.getNext()); return result; }

Mentor tip

Remember to handle n == 1 separately (using removeFirst() or updating head), then use prev.setNext(curr.getNext()) to unhook node n.

(f)

4 marks

Elements of style and naming conventions: Using meaningful variable/method identifiers, consistent capitalization (such as CamelCase for classes and lowerCamelCase for methods/variables), proper code indentation/whitespace, and inline comments.
Why they are important: They make source code self-documenting and easier to read, simplify debugging and future maintenance extensions, and allow multiple developers on a team to collaborate seamlessly without confusion.

Mentor tip

Include at least two concrete style elements (meaningful identifiers, indentation, capitalization, comments) and two benefits (readability, easier maintenance/debugging, team collaboration).

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: Relational vs object-oriented databases, election data warehousin · Q4
Present two clear 2-mark arguments: industry standardization/tooling/staffing and simpler tabular modelling compared to OOP complexity.
02
Modelling and simulation: Optical character recognition (OCR) and neural networks for handw · Q8
State a cause (poor print quality or decorative/sans-serif fonts) and give a specific character confusion example (such as '5' vs 'S' or '1' vs 'I').
03
Web science: OrderYourFoods.com web protocols, server-side PHP email validatio · Q9
Commercial ordering websites handling customer addresses and payments use HTTPS.
04
Object-oriented programming: Singly linked list of CartNode objects (POSlist), pointer manipul · Q17
State both 'pointer/memory address' and 'where the object is stored in memory' for full marks.

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 2019
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 2019 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

14 marks

One-to-7 educator forum registration, ACID properties, DBA tasks, and composite keys

Databases

(a)

1 mark

Splitting the name into First Name and Last Name ensures attribute atomicity (1st Normal Form) and enables fast, accurate sorting or searching by surname.

Mentor tip

State either attribute atomicity or the ability to sort/search directly by Last Name.

(b)

2 marks

The form demands excessive sensitive personal details (such as Date of Birth, Place of Birth, and Country of Birth) that are unnecessary simply to participate in an educator forum.
Teachers may refuse to register due to privacy concerns, fear of identity theft if the database is breached or shared with third parties, or frustration with the time required to complete the form.

Mentor tip

Connect excessive personal data collection to privacy/identity-theft risks or lower user registration rates.

(c)

2 marks

Transaction atomicity ensures that all database operations triggered when submitting the registration form either execute completely as a single indivisible unit or do not occur at all.
This prevents partial registration records from being saved if a network drop or system error interrupts submission, preserving database integrity.

Mentor tip

Explain 'all-or-nothing' execution and how it prevents incomplete educator profiles from corrupting the table.

(d)

3 marks

Transaction isolation ensures that when multiple teachers submit forum replies simultaneously, concurrent transactions execute independently and produce the same result as sequential processing.
When a teacher's post transaction updates a discussion thread, the DBMS places a lock on the relevant table rows so another concurrent post cannot overwrite or displace it before the first transaction commits.
A transaction log is also maintained prior to commit so that if an error occurs mid-post, the transaction rolls back cleanly to the previous valid state.

Mentor tip

Link row/record locking, independent sequential-equivalent execution, and transaction log rollback to preventing lost or garbled forum posts.

(e)

2 marks

Task 1: Configuring user access permissions, passwords, and security policies.
Task 2: Performing scheduled database backups, disaster recovery planning, data cleansing, or applying system patches/upgrades.

Mentor tip

List any two distinct operational DBA responsibilities such as managing access control/security and performing regular backups/recovery.

(f)

4 marks

With 250,000 educators registering, any composite primary key must combine enough attributes to guarantee 100% uniqueness across all records.
The DBA must consider demographic overlaps: for instance, combining First_Name + Last_Name + DOB might still produce collisions if many teachers come from the same country or age group.
Adding attributes that are invariant for a person (or incorporating a system-generated unique identifier/phone number alongside name and birth details) helps prevent a single educator from registering multiple accounts under different email addresses.
However, using too many fields in a composite key increases index overhead and makes foreign-key joins more cumbersome.

Mentor tip

Discuss uniqueness across 250,000 users, which specific form fields could be combined, and the risk of demographic collisions (shared names/birthdates).

Question 2

16 marks

Linnaean bird taxonomy ERD, validation vs verification, schemas, and referential integrity

Databases

(a)

2 marks

Identify the three hierarchical taxonomic entities: CLASS, ORDER, and FAMILY.
Establish the two 1-to-many (1:M) relationships: one CLASS contains many ORDERS, and one ORDER contains many FAMILIES.
Construct the top-down ERD:
CLASS (1) ----< (contains / has) >---- (M) ORDER (1) ----< (contains / holds) >---- (M) FAMILY

Mentor tip

Show both 1:M relationships clearly labelled in hierarchy from CLASS to ORDER and from ORDER to FAMILY.

(b)(i)

2 marks

Apply a range check or data type check on the beak length field.
For example, enforce a numeric (float/integer) data type and a range check between 5 cm and 60 cm so negative numbers, text strings, or implausible measurements are automatically rejected.

Mentor tip

Name a specific validation rule (such as range check or type check) and give a concrete numerical example for spoonbill beak length.

(b)(ii)

2 marks

Use double entry or visual proofreading against the original field research notes.
For example, requiring the researcher to type the spoonbill beak measurement twice (and flagging any mismatch) ensures that transcription typos are caught during entry.

Mentor tip

Remember: validation checks if data is reasonable/properly formatted via rules, whereas verification (double entry, proofreading) checks if data was copied accurately from the source.

(c)

2 marks

Data consists of raw, uncontextualized figures entered into a cell - for example, storing the raw number '18' in the database.
Information is produced when that stored figure is queried and interpreted within the database context and field metadata - for example, understanding that '18' represents 'a spoonbill beak length of 18 cm'.

Mentor tip

Contrast the raw numerical value stored in the table ('18') with the contextualized meaning retrieved from the database ('beak length of 18 cm').

(d)

2 marks

Characteristic 1: Provides a high-level, implementation-independent overview of the entire database architecture.
Characteristic 2: Defines the core business entities/concepts and the primary relationships between them (often represented via an ERD or Data Structure Diagram).

Mentor tip

Focus on high-level abstraction of entities and relationships independent of physical storage details.

(e)

3 marks

Data modelling creates a clear visual blueprint of entities, attributes, and relationships so scientists and database developers share a common understanding before implementation.
It translates conceptual and logical schemas into a sound physical database structure while establishing consistent naming conventions.
Without thorough modelling, the taxonomic database could suffer from structural flaws, data redundancy, and broken integrity constraints.

Mentor tip

Link stakeholder alignment and schema design (conceptual -> logical -> physical) to preventing redundancy and integrity errors.

(f)

3 marks

Referential integrity enforces the constraint that every foreign key value in a child table must correspond to an existing primary key value in the parent table.
In the bird taxonomy database, this prevents a researcher from inserting the 'Spoonbill' family linked to the order 'Ciconiiformes' unless 'Ciconiiformes' already exists as a valid primary key in the ORDER table.
It also ensures that updating or deleting a taxonomic order automatically cascades or blocks orphaned family records, preventing update and deletion anomalies.

Mentor tip

Explicitly reference the Spoonbill / Ciconiiformes relationship to earn all 3 marks.

Question 3

15 marks

Trucking On haulage database normalization to 3NF and SQL join queries

Databases

(a)

2 marks

Characteristic 1: Cells contain non-atomic, multi-valued repeating groups of data within a single row (such as multiple Trailer IDs, spaces, and coupling dates for one truck).
Characteristic 2: The table lacks a single primary key that uniquely identifies each individual coupling record without duplication.

Mentor tip

Point out repeating multi-valued attributes in cells and the absence of atomic tuples / unique row keys.

(b)

2 marks

Data redundancy means identical facts (such as Trailer TR9 having a space of 30 cubic metres, or Truck make 'Fodel' using 'Diesel') are duplicated across multiple rows.
If an attribute is modified later, failing to update every duplicate instance causes update anomalies and inconsistent operational records.

Mentor tip

Cite duplicated values in the LORRY table and explain how partial updates lead to data inconsistency.

(c)

4 marks

Step 1: Select the two required tables: LORRY and TRAILER.
Step 2: Join the tables on their shared identifier: LORRY.TruckID = TRAILER.TruckID.
Step 3: Apply the filter condition WHERE Trailerspace > 60.
Step 4: Project the Driver and Tel fields (SQL: SELECT Driver, Tel FROM LORRY INNER JOIN TRAILER ON LORRY.TruckID = TRAILER.TruckID WHERE Trailerspace > 60;).

Mentor tip

Include all four components: the two tables (LORRY, TRAILER), the join condition on TruckID, the filter (Trailerspace > 60), and the selected attributes (Driver, Tel).

(d)

7 marks

Separate repeating trailer assignments into a junction table (JOURNEY or COUPLING), isolate trailer capacity into a TRAILER table, and separate driver phone numbers into a DRIVER table to remove transitive dependencies:
DRIVER (Driver [PK], Tel)
TRUCK (TruckID [PK], Truckmake, Energysource, Driver [FK])
TRAILER (TrailerID [PK], Trailerspace)
JOURNEY (TruckID [PK, FK], TrailerID [PK, FK], Coupled_from [PK], Coupled_to) - or with a surrogate primary key: JOURNEY (JourneyID [PK], TruckID [FK], TrailerID [FK], Coupled_from, Coupled_to).

Mentor tip

Ensure all 4 relations (DRIVER, TRUCK, TRAILER, JOURNEY) are defined with underlined/marked Primary Keys and appropriate Foreign Keys.

Question 4

20 marks

Relational vs object-oriented databases, election data warehousing, and data mining ethics

Databases

(a)

4 marks

Advantage 1 - Mature industry standards andwidespread tooling: Relational databases (RDBMS) have decades of standardization (SQL), making it easier to recruit experienced staff, maintain stability, and use built-in reporting, security, and concurrency tools.
Advantage 2 - Simpler tabular visualization and modelling: Representing data as normalized two-dimensional tables with explicit relationships is easier to conceptualize and verify for data integrity than complex object-oriented class hierarchies.

Mentor tip

Present two clear 2-mark arguments: industry standardization/tooling/staffing and simpler tabular modelling compared to OOP complexity.

(b)

2 marks

Characteristic 1: Stores historical, time-variant data integrated from multiple heterogeneous sources.
Characteristic 2: Optimized for read-heavy analytical querying (OLAP), reporting, and data mining rather than transactional updates.

Mentor tip

Mention time-variant/historical multi-source storage and OLAP/analytical query optimization.

(c)

2 marks

Data gathered from disparate external sources (polling agencies, voter registries, social media) arrives in incompatible formats and conventions (for example, dates formatted as DD/MM/YY vs MM/DD/YYYY).
Transformation cleans, validates, and standardizes these records into a unified format so cross-source analytical queries execute accurately.

Mentor tip

Give a concrete format mismatch example (like date formats) and explain the need for standardization before loading.

(d)

2 marks

Voter sentiment during an election campaign is highly time-sensitive and shifts rapidly in response to political events.
Timestamping opinion polls allows analysts to track longitudinal shifts in voter preferences - for example, comparing support levels before and after a televised debate.

Mentor tip

Explicitly tie timestamping to the US presidential election scenario (such as comparing polls before and after a debate).

(e)

2 marks

Analytics firms use link analysis to uncover hidden associations and relationships between disparate entities or datasets.
In an election campaign, link analysis can connect voter preferences to lifestyle variables such as social media usage patterns, household size, or regional community memberships.

Mentor tip

Define link analysis as discovering associations across entities and give an election-specific example.

(f)

2 marks

Deviation detection identifies anomalous patterns or statistical outliers that depart from established historical baselines.
In the election context, it alerts campaign strategists to sudden, unexpected swings in pre-election polling or unusual spikes in negative voter sentiment within a specific swing state.

Mentor tip

Connect anomaly/outlier detection directly to spotting sudden shifts in state-level voter polling or sentiment.

(g)

6 marks

Advantages for campaign strategy: Data mining techniques like cluster analysis, classification, and association rules uncover subtle demographic patterns - such as grouping voters by income or identifying which policy issues resonate with specific demographics - allowing campaigns to tailor messaging and allocate advertising budgets efficiently.
Disadvantages and ethical risks: Election data mining relies on harvesting vast quantities of personal, behavioural, and social media data from citizens, often without their explicit informed consent.
Aggregating these datasets can strip away personal anonymity, expose sensitive private traits, and enable manipulative micro-targeting or political polarization.
Balanced conclusion: While data mining offers powerful analytical efficiency for campaign strategists, the societal and ethical risks outweigh the benefits unless strict data-protection laws, transparency, and anonymization safeguards are enforced.

Mentor tip

Include three components linked to the election scenario: specific data mining techniques/benefits (clustering/classification), privacy/anonymity drawbacks, and a reasoned conclusion.

Option B: Modelling and simulation · Option

Questions 5–8 · 65 marks

Question 5

17 marks

Compound interest investment planning, spreadsheet modelling, and progressive tax algorithms

Modelling and simulation

(a)

2 marks

Substitute P = 30,000, r = 0.10, and n = 2 into the compound interest formula T = P * (1 + r)^n:
T = 30,000 * (1 + 0.10)^2 = 30,000 * 1.21 = $36,300.

T = 30000 \times (1.10)^{2} = $36 300

Mentor tip

State the total investment value ( $36, 300), n o t j u s tt h e in t er es t e a r n e d ($ 6,300).

(b)

5 marks

Set up a 12-row iterative table or spreadsheet model with columns: Month (A), Start Balance (B), Monthly Interest (C), Monthly Deposit (D), and End Balance (E).
Initialize Month 1 Start Balance (B2) = 30000.
Calculate Month 1 Interest (C2) = B2 * 0.005 (since 0.5% = 0.005) and set Monthly Deposit (D2) = 1000.
Calculate Month 1 End Balance (E2) = B2 + C2 + D2 (or B2 * 1.005 + 1000).
For Months 2 through 12, set each month's Start Balance equal to the previous month's End Balance (e.g., B3 = E2) and copy the formulas down to row 13 (Month 12).

Mentor tip

(c)

6 marks

Initialize PRINCIPAL = 30000, RATE = 1.005, and running TAX = 0.
Loop 12 times (for each month): compute the new balance with interest, determine the monthly PROFIT (interest earned), add the $1,000 deposit, and apply the conditional tax rate (25% if total <= 40,000, else 40%).
After the 12-month loop finishes, subtract the accumulated TAX from PRINCIPAL and output the final value:
PRINCIPAL = 30000 RATE = 1.005 TAX = 0 loop MONTH from 1 to 12 OLD_PRINCIPAL = PRINCIPAL INVESTMENT = OLD_PRINCIPAL * RATE PROFIT = INVESTMENT - OLD_PRINCIPAL PRINCIPAL = INVESTMENT + 1000 if PRINCIPAL <= 40000 then TAX = TAX + (PROFIT * 0.25) else TAX = TAX + (PROFIT * 0.40) end if output "End of month ", MONTH, " fund value: ", PRINCIPAL end loop FINAL_VALUE = PRINCIPAL - TAX output "Final Investment Value after tax: ", FINAL_VALUE

Mentor tip

Remember that tax is calculated each month on that month's profit and accumulated in a running total, but only deducted from the fund total at the end of the 12 months.

(d)

4 marks

Mathematical modelling enables investment companies to run 'what-if' scenarios by adjusting key variables before launching a financial product.
Analysts can track month-by-month fund growth and test how varying the interest rate impacts company profitability and client payouts.
They can also simulate how changes in customer monthly contributions or government tax brackets affect final investment returns.

Mentor tip

Mention 'what-if' scenario testing and give specific examples of varying interest rates, monthly deposits, and tax rates to observe their impact on profits.

Question 6

16 marks

Supermarket regional distribution centre location simulation

Modelling and simulation

(a)

4 marks

Reason 1 - Avoiding prohibitive capital cost: Constructing a physical distribution warehouse is extremely expensive, so it is financially impossible to build all four centres just to test which site performs best.
Reason 2 - Time compression and scenario testing: A computer simulation can model years of regional delivery traffic, roadworks, and worst-case congestion scenarios across all four sites in hours rather than waiting years for physical construction.

Mentor tip

Provide two distinct reasons (cost savings over physical trial-and-error, and speed/flexibility to test 'what-if' traffic scenarios) with a 1-mark expansion for each.

(b)(i)

2 marks

Input 1: Geographical coordinates/distances from each prospective site to the regional supermarkets and supply ports/rail hubs (or road network capacities and travel times at different hours).
Input 2: Land purchase prices and local labour wage rates at each of the four locations.

Mentor tip

List concrete raw data inputs such as road distances/travel times to supermarkets, land costs, or local wage rates.

(b)(ii)

2 marks

Criterion 1: Lowest total capital construction cost and ongoing operational/shipping cost.
Criterion 2: Shortest average delivery transit time to distribute goods to the supermarkets (or shortest construction completion time).

Mentor tip

Distinguish between raw data inputs in (b)(i) and decision evaluation criteria in (b)(ii) such as total construction/running cost and delivery turnaround time.

(b)(iii)

6 marks

Evaluating capital construction cost and build time: The simulation takes inputs such as land price, local labour wages, worker headcount, and material delivery schedules for each of the four sites, running scenarios (such as supply delays or wage changes) to project total build cost and completion time.
Evaluating ongoing distribution running costs: The model combines distances to supermarkets, lorry fuel economy, fuel prices, and shipment frequencies, simulating scenarios like fuel price spikes or increased order volumes to calculate annual operating expenditure.
Evaluating delivery transit times: Using road network maps, loading/unloading durations, and time-of-day traffic data, the simulation tests worst-case traffic congestion or roadworks to rank the four locations against the cost and speed criteria.

Mentor tip

(c)

2 marks

Developing a highly detailed simulation can be expensive and time-consuming, potentially outweighing the benefits if one site is already obviously superior.
Furthermore, real-world logistics involve too many unpredictable confounding variables (such as sudden political changes, extreme weather, or unmodelled human factors), meaning an oversimplified simulation may produce misleading recommendations.

Mentor tip

Mention both the cost/complexity of building the simulation and the risk of inaccurate results due to omitted or unpredictable real-world variables.

Question 7

12 marks

Tibetan antelope migration satellite tracking and real-time 2D visualization

Modelling and simulation

(a)

4 marks

Item 1: Real-time GPS coordinates of tagged antelopes transmitted continuously from satellite trackers.
Item 2: Current movement speed of the herd and the time required for the herd to cross a checkpoint.
Item 3: Historical migration route coordinates and fixed map geometry of the QT Railway and QT Highway.
Item 4: Calculated predicted path and estimated time of intersection with the highway/railway.

Mentor tip

Include four distinct data items: live GPS positions, herd speed/crossing duration, historical/road map coordinates, and predicted intersection timing.

(b)

4 marks

Reason 1 - Sufficiency of 2D spatial mapping: Antelopes travel across the ground surface, so their proximity to the railway and highway is fully represented on a 2D planar map without needing vertical 3D elevation rendering.
Reason 2 - Real-time rendering speed and lower hardware demands: A 2D visualization is computationally simpler and updates rapidly in real time on standard hardware, whereas complex 3D rendering requires powerful GPUs and could introduce processing delays when issuing urgent traffic warnings.

Mentor tip

Give two explained points: ground movement only needs 2D planar representation, and 2D updates much faster in real time without heavy GPU overhead.

(c)

4 marks

Challenge of manual observation: The Hoh Xil Nature Reserve is a vast, remote, and inhospitable mountainous region spanning over 300 km with harsh weather and poor visibility, making it impossible to station enough human observers along the entire railway and highway.
Necessity of satellite-fed visualization: Automated satellite telemetry displayed on a real-time visualization model allows authorities to monitor herd movements remotely 24/7 and proactively halt road or rail traffic before collisions occur.

Mentor tip

Spend 2 marks explaining the physical difficulties of human monitoring in remote/mountainous Tibet, and 2 marks on how satellite-driven visualization solves this.

Question 8

20 marks

Optical character recognition (OCR) and neural networks for handwriting and predictive text

Modelling and simulation

(a)

2 marks

Printed documents may suffer from degraded print quality (faint ink, smudges, or skewed scans) or use ornate/unusual fonts.
As a result, visually similar glyphs - such as the digit '5' and letter 'S', or '1' and 'I' in sans-serif typefaces - can be misclassified by the OCR scanner.

Mentor tip

State a cause (poor print quality or decorative/sans-serif fonts) and give a specific character confusion example (such as '5' vs 'S' or '1' vs 'I').

(b)

2 marks

Unlike rigid template matching that requires an exact pixel-for-pixel match, an Artificial Neural Network (ANN) can be trained on thousands of varied handwriting samples to generalize across different stroke styles.
The ANN extracts local sub-features of a character and calculates probabilistic similarity, allowing it to recognize a skewed or imperfect handwritten 'X' without requiring the entire grid to match Figure 4 identically.

Mentor tip

Emphasize the ANN's ability to learn from training examples and recognize partial/generalized stroke features rather than requiring an exact pixel match.

(c)

4 marks

Step 1 - Feature filtering / Convolution: The ANN breaks the 9x9 pixel grid into smaller local patches (such as 3x3 pixel filters representing diagonal lines and crossing intersections) and slides these filters across the image.
Step 2 - Pixel multiplication and scoring: At each position, the network multiplies image pixel values (+1 or -1) by the filter weights, sums them, and divides by the patch size to compute a feature match score (where 1.0 indicates a perfect local match).
Step 3 - Activation and Pooling: A Rectified Linear Unit (ReLU) layer zeroes out negative values to simplify computation, and pooling layers downsample the feature maps to make detection invariant to slight shifts in position.
Step 4 - Classification output: The stacked feature maps are fed into fully connected output layers and compared against stored character classes to identify the pattern as the letter 'X'.

Mentor tip

Walk through the convolutional neural network stages: sliding small pixel filters across the grid, computing local match scores, applying ReLU/pooling, and matching the combined features to 'X'.

(d)

2 marks

Feature 1: A stored lexicon / dictionary of valid words and grammatical syntax rules (nouns, verbs, adjectives).
Feature 2: Recurrent memory / feedback loops that retain the preceding words in the sentence as context inputs.

Mentor tip

Mention a vocabulary dictionary/grammar model and memory of previous words in the current sentence.

(e)

6 marks

Step 1 - Dual context inputs: The neural network receives two sets of inputs into its input layer: the sequence of preceding words ('The child is feeling') and the newly typed character prefix ('hu').
Step 2 - Weighted hidden layers: These inputs are combined and passed through multiple hidden layers where synaptic weights reflect learned statistical co-occurrence frequencies from training text.
Step 3 - Non-linear activation: Hidden nodes apply non-linear activation functions (such as a sigmoid or softmax function) to compute the conditional probability of prospective words.
Step 4 - Recurrent feedback and backpropagation: When the user types 'hu', the network merges the grammatical context (expecting an adjective after 'is feeling') with the prefix constraint ('hu*'), ranking 'hungry' as the highest-probability output while using feedback loops to refine future predictions.

Mentor tip

Trace the data flow through the ANN: input of prior words + input of new letters ('hu') -> weighted hidden layers -> sigmoid/probability calculation -> merging context and prefix to output 'hungry'.

(f)

4 marks

Problem 1 - Slang, text-speak, and poor training corpora: If users type informal abbreviations, slang, or 'text-speak' not present in the training dictionary (or if unsupervised learning absorbs misspelled user inputs), the ANN will fail to suggest appropriate words.
Problem 2 - Vanishing or exploding gradients during training: When backpropagating error signals across many recurrent steps or deep layers, repeated multiplication by the weight matrix can cause the gradient signal to shrink toward zero (vanishing gradient) or grow uncontrollably (exploding gradient), stalling or destabilizing learning.

Mentor tip

State two distinct training challenges: vocabulary/text-speak quality issues and the mathematical vanishing/exploding gradient problem.

Option C: Web science · Option

Questions 9–13 · 65 marks

Question 9

16 marks

OrderYourFoods.com web protocols, server-side PHP email validation, and black hat SEO

Web science

(a)(i)

1 mark

HTTPS (Hypertext Transfer Protocol Secure).

Mentor tip

Commercial ordering websites handling customer addresses and payments use HTTPS.

(a)(ii)

4 marks

Function 1 - Standardized communication and packet switching: Protocols define strict rules for formatting, addressing, and routing data packets across heterogeneous networks.
Function 2 - Data integrity and error correction: Protocols use checksums and acknowledgements to detect corrupted or dropped packets and trigger retransmission over unreliable channels.
Function 3 - Flow control and congestion management: Protocols regulate transmission rates so a fast server does not overwhelm a slower client device, and prevent network deadlock.
Function 4 - Security and encryption: Protocols like HTTPS/TLS authenticate servers and encrypt data in transit.

Mentor tip

Explain four distinct roles of network protocols: error checking/correction, flow control, congestion/deadlock management, and encryption/security.

(b)(i)

1 mark

Server-side scripting.

Mentor tip

PHP code executes on the web server rather than in the client's browser.

(b)(ii)

1 mark

State either built-in PHP function call with its arguments:
filter_var( $_{P} O S T [^{'} e mai l^{'}], F I L T E R_{V} A L I D A T E_{E} M A I L) - or mai l ($ to, $s u bj ec t,$ body, $h e a d er s, " - f ".$ from).

Mentor tip

Include both the function name (filter_var or mail) and the parameters inside the parentheses.

(b)(iii)

3 marks

Step 1: The PHP script retrieves the customer's submitted email address from the form via the $_POST['email'] associative array and assigns recipient/header variables.
Step 2: It validates the format of the submitted email address using filter_var($_POST['email'], FILTER_VALIDATE_EMAIL); if the email format is invalid, no further action or output occurs.
Step 3: If the email is valid, it calls the mail() function to send the subscription notification to info@orderyourfoods.com and outputs a confirmation message if mail() succeeds, or outputs an error message if sending fails.

Mentor tip

Be sure to note what happens in all branches: valid + mail sent -> confirmation echoed; valid + mail fails -> error echoed; invalid format -> nothing happens.

(c)

1 mark

Live online chat widget (or embedded social media links such as Facebook/Twitter, or a customer discussion forum).

Mentor tip

Any interactive web communication feature beyond static email/phone (such as live chat or social media integration) is valid.

(d)

5 marks

Short-term advantages: Black hat SEO techniques (such as keyword stuffing, hidden text, link farms, or doorway pages) can artificially inflate OrderYourFoods' ranking for 'food delivery' quickly, generating a short-term spike in website traffic and orders.
Long-term disadvantages: Black hat methods violate search engine guidelines and are considered deceptive and unethical.
Modern search engine crawlers actively detect manipulative patterns and will penalize or permanently blacklist (de-index) www.orderyourfoods.com from search results altogether.
Overall evaluation: While black hat SEO might yield brief initial visibility, the severe long-term risk of being banned from search engines and damaging the company's reputation makes it an unwise strategy compared to legitimate white hat SEO.

Mentor tip

Structure your evaluation with short-term visibility gains, long-term blacklisting/reputational risks, and a clear concluding judgement.

Question 10

15 marks

Search engines, PageRank vs HITS algorithms, cloud vs client-server, and image compression

Web science

(a)

1 mark

A search engine is a software system that queries an indexed database of web pages based on user-supplied keywords and returns a ranked list of relevant resources.

Mentor tip

Mention querying an indexed database of web pages to retrieve matching results.

(b)

4 marks

PageRank algorithm (Google): Evaluates the importance of a web page by recursively counting both the quantity and the quality (authority weight) of incoming links (in-links) pointing to it, under the assumption that authoritative sites receive more links from other important sites.
HITS algorithm (Ask - Hyperlink-Induced Topic Search): Evaluates pages dynamically for a query by calculating two distinct scores for every page: an Authority score (high number of in-links from hubs, indicating valuable content on the topic) and a Hub score (high number of out-links pointing to reputable authorities).

Mentor tip

Devote 2 marks to PageRank (count and quality of in-links) and 2 marks to HITS (distinguishing between Hubs with many out-links and Authorities with many in-links).

(c)

3 marks

Local client-server architecture: Hosted on dedicated servers maintained locally within the school/organization, providing low-latency local data transmission and direct physical security control for internal clients.
Cloud computing: Hosts data and applications on distributed remote server infrastructure accessed over the internet as an on-demand service.
Key difference: Cloud platforms can elastically scale storage and computing capacity up or down and allow ubiquitous remote access from anywhere, whereas local client-server networks have fixed hardware capacity and are managed at the organizational level.

Mentor tip

Define local client-server, define cloud computing, and state a clear operational contrast (such as elastic scalability, remote accessibility, or local latency/security).

(d)(i)

1 mark

Lossless compression.

Mentor tip

Lossless is the exact counterpart to lossy compression.

(d)(ii)

4 marks

Advantages: Lossy JPG compression drastically reduces image file sizes, allowing students to upload and download coursework documents much faster on the cloud platform and enabling smoother scrolling through multi-page essays.
Disadvantages: Because lossy compression permanently discards pixel data, high compression ratios introduce visible artifacts and blurriness, degrading visual quality when coursework is printed or when fine diagram details are examined.
Conclusion: Using compressed images is beneficial for general illustrative photos to keep coursework file sizes manageable, provided the compression ratio does not render essential academic diagrams illegible.

Mentor tip

Cover upload/download speed and document responsiveness (pros), reduced visual/print resolution (cons), and a contextual conclusion.

(e)

2 marks

Interoperability is the ability of different software applications, browsers, and operating systems to exchange and render information seamlessly.
Open standards are publicly available, royalty-free technical specifications (like the PNG format) that make interoperability across diverse platforms possible.

Mentor tip

Clearly distinguish the capability of systems working together (interoperability) from the publicly shared specifications that enable it (open standards).

Question 11

14 marks

World Wide Web network growth, browser and DNS functions, and intellectual property

Web science

(a)

3 marks

The World Wide Web is a scale-free distributed network where the total number of connected nodes (such as IB World Schools) grows exponentially.
However, due to small-world network topology and highly connected hubs, the average path length (number of routing hops between any two nodes) grows only logarithmically/linearly.
Consequently, schools worldwide can communicate and share resources with minimal latency regardless of physical distance, acting as a catalyst for global collaboration.

Mentor tip

Relate exponential node growth vs small hop counts / negligible latency to why global collaboration scales so effectively.

(b)

3 marks

Function - Fetching and rendering web pages (or bridging URLs to DNS):
The web browser takes the URL entered by the user, contacts a DNS server to resolve the domain into an IP address, and sends an HTTP/HTTPS request to the target web server.
It then receives the server's response files (HTML, CSS, JavaScript) and parses/interprets the markup code to render the visual, interactive web page on the user's screen.

Mentor tip

Trace the browser's end-to-end role: resolving the URL via DNS, fetching HTML/CSS/JS over HTTP(S), and parsing/rendering the page visually.

(c)

3 marks

Step 1: When a user enters a domain name into the browser, the request is sent to a Domain Name Server (DNS), which searches its local database for the matching IP address.
Step 2: If the domain is found, the DNS returns the IP address so the request can be routed to the destination web server.
Step 3: If the domain is not in its local database, the DNS queries other upstream DNS servers in the hierarchy until the IP address is resolved, or returns a 'site not found' error if no match exists anywhere.

Mentor tip

Include local database lookup, forwarding to other DNS servers on a cache miss, and returning either the resolved IP address or an error.

(d)

5 marks

Copyright issues: Because the World Wide Web makes digital text, images, and media effortlessly accessible across billions of pages, users can copy, redistribute, or publicly display works instantaneously - often infringing copyright laws unintentionally or due to ignorance of complex licensing terms.
Intellectual property (IP) issues: Web content frequently incorporates protected trademarks, patents, and proprietary designs; downloading, modifying, or remixing resources without the creator's authorization violates IP rights.
Jurisdictional complexity: Because the web is globally distributed across countries with conflicting copyright and IP statutes, an action that is legal in one jurisdiction may constitute infringement in another, making enforcement difficult.

Mentor tip

Question 12

14 marks

OBI European offices directed graph, web connectivity, multimedia search, and collective intelligence

Web science

(a)

2 marks

Represent the three offices as vertices (nodes) A, B, and C.
Draw a bidirectional arrow between A and B (A <---> B) to represent their mutual project collaboration.
Draw unidirectional arrows from A to C (A ---> C) and from B to C (B ---> C) to represent reporting to the head office.

Mentor tip

Double-check arrow directions: A <---> B (two-way collaboration) and A ---> C, B ---> C (one-way reporting).

(b)

3 marks

Graph theory models the World Wide Web by representing individual web pages as vertices (nodes) and hyperlinks between pages as directed edges (arrows indicating 1-way or 2-way links).
In graph theory, connectivity measures whether a directed path exists to traverse from any given vertex to another vertex.
Applying graph theory to the web allows researchers to analyse bow-tie structure, identify isolated components vs the strongly connected core, and measure how many hyperlink hops separate web pages.

Mentor tip

Explicitly map web pages to vertices/nodes, hyperlinks to directed edges, and path traversal between vertices to web connectivity.

(c)

4 marks

Issue 1 - Reliance on accurate manual metadata/tags: Search engines typically index multimedia (images, audio, video) using surrounding text or assigned meta-tags; if contributors omit or mislabel tags, the non-text files will not be found by keyword queries.
Issue 2 - Computational difficulty of content-based similarity search: Searching directly by visual or acoustic content (such as reverse image search) requires complex feature-extraction algorithms because there are no explicit text tokens to match, and fewer search engines support efficient multimedia indexing.

Mentor tip

Explain two distinct challenges: missing/inaccurate meta-tags on multimedia files, and the algorithmic difficulty of content-based image/audio similarity matching.

(d)

5 marks

Advantages: Utilizing collective intelligence allows OBI to tap into a massive global pool of contributors working simultaneously, completing large-scale data gathering far more rapidly and at a much lower cost per unit than hiring full-time specialists.
Disadvantages: Because crowdsourced environments are decentralized and loosely regulated, it is difficult to enforce rigorous quality control, verify the accuracy or originality of submissions, or evaluate thousands of competing ideas without clear hierarchy.
Conclusion: Collective intelligence is highly effective for OBI when rapidly gathering broad, non-safety-critical information, provided Office A implements verification workflows to filter out low-quality contributions.

Mentor tip

Weigh speed, scale, and low cost (pros) against lack of quality control, unverified originality, and coordination overhead (cons), finishing with a contextual conclusion.

Question 13

6 marks

Southend Council ambient intelligence and IoT humanoid robot (Pepper) evaluation

Web science

6 marks

Environmental complexity and sub-optimal decisions: Guiding elderly residents with memory loss from care homes to local shops exposes Pepper to unpredictable outdoor environments (traffic, roadworks, weather) that may exceed its ambient intelligence training, leading to navigation errors with serious safety consequences.
IoT connectivity black spots: Pepper relies on continuous wireless communication with Internet of Things (IoT) sensors and cloud servers; network dead zones or latency spikes along the route could leave the robot and resident stranded.
Over-reliance vs distrust by vulnerable residents: Residents suffering from dementia might either distrust the robot and become distressed, or over-trust Pepper and delegate critical road-crossing judgements that the robot is not equipped to handle.
Privacy and data security concerns: Families and care home regulators may object to Pepper's cameras and sensors continuously recording and transmitting sensitive location and health data over the IoT.
Legal accountability and liability: If an elderly resident is injured in an accident while following Pepper's directions, legal liability between Southend Council, the care home, and the robot manufacturer remains ambiguous.

Mentor tip

Option D: Object-oriented programming · Option

Questions 14–17 · 65 marks

Question 14

16 marks

Supermarket point-of-sale (POS) simulation: POSsystem, POSline, and Cart array manipulation

Object-oriented programming

(a)

1 mark

Aggregation (a 1-to-many 'has-a' relationship: one POSsystem aggregates multiple POSline objects).

Mentor tip

State 'aggregation' (or 'has-a' / 1-to-many relationship).

(b)

2 marks

Draw a hierarchy or UML aggregation diagram showing one POSsystem connected to multiple POSline objects, and each POSline connected to multiple Cart objects:
POSsystem (1) ◇---- (many) POSline (1) ◇---- (many) Cart

Mentor tip

Clearly show the two levels of 1-to-many relationships: POSsystem -> multiple POSlines, and each POSline -> multiple Carts.

(c)

1 mark

An identifier is the unique symbolic name assigned by a programmer to a program element such as a variable, method, or class.

Mentor tip

Must mention that it is the 'name' of a variable, method, or class.

(d)

3 marks

A class is the abstract template or blueprint that defines the attributes (data fields) and methods (behaviours) of an entity.
An instantiation is a concrete object created from that class blueprint and allocated memory at runtime with specific attribute values.
In the provided UML, POSline is the class blueprint defining id, active, and line, whereas an individual checkout counter in the supermarket is an instantiation of POSline storing instantiated Cart objects inside its line array.

Mentor tip

Define class, define instantiation, and explicitly reference POSline and Cart from the UML diagram.

(e)

2 marks

Use the new keyword to allocate an array of 20 Cart references:
private Cart[] line = new Cart[20]; (or line = new Cart[20];)

Mentor tip

Include both the new keyword and the size specification Cart[20].

(f)

3 marks

Start at index i = 0 and advance through the line array until the first null (empty) slot is found, then assign newCart to line[i]:
public void joinLine(Cart newCart) { int i = 0; while (this.line[i] != null) { i++; } this.line[i] = newCart; }

Mentor tip

Searching for the first null entry (this.line[i] != null) locates the end of the current queue in the array.

(g)

4 marks

Save the Cart at index n in a temporary variable result to return at the end.
Loop from i = n up to 18, shifting each subsequent element one position to the left (line[i] = line[i + 1]).
Set the final slot line[19] = null to clear the duplicate trailing reference, and return result:
public Cart leaveLine(int n) { Cart result = this.line[n]; for (int i = n; i < 19; i++) { this.line[i] = this.line[i + 1]; } this.line[19] = null; return result; }

Mentor tip

Don't forget to set this.line[19] = null after shifting elements down, and loop up to i < 19 so line[i + 1] does not cause an ArrayIndexOutOfBoundsException.

Question 15

15 marks

POS simulation inheritance, class libraries, model refactoring, and distributed teams

Object-oriented programming

(a)

2 marks

Inheritance is the OOP mechanism whereby a new class (subclass/child) is derived from an existing class (superclass/parent), automatically acquiring all of the parent's attributes and methods.

Mentor tip

Mention both deriving a subclass from a superclass and inheriting its attributes and methods.

(b)

3 marks

Inheritance promotes code reuse and extensibility.
Because the parent class POSline already defines common attributes (id, active, line) and queue operations (joinLine, leaveLine), developers can easily derive specialized subclasses such as CashOnlyPOSline or ExpressTenItemPOSline without rewriting shared queue logic, reducing development and testing time.

Mentor tip

State the advantage (code reuse / extensibility / reduced maintenance), explain how the parent holds shared code, and tie it directly to POSline and specialized checkout counters.

(c)

4 marks

Advantage 1 - Faster development and code reuse: Pre-written library classes provide ready-made data structures and utilities so programmers do not have to reinvent standard algorithms from scratch.
Advantage 2 - High reliability and abstraction: Library classes have been thoroughly tested and debugged by thousands of developers, and can be used as black-box abstractions without needing to know their internal implementation details.

Mentor tip

Give two distinct advantages (such as saving development time, pre-tested reliability, or black-box abstraction) and elaborate on each.

(d)

2 marks

In the original model, POSline tightly couples a single checkout counter with its own Cart[] line array.
To allow one shared queue feeding multiple counters, the model must decouple the waiting line from the checkout counter - for example, by creating a separate Counter class (or an array of active counters inside POSline/POSsystem) that pulls the next Cart from a single shared Cart[] line.

Mentor tip

Explain that the checkout counter entity and the waiting line of carts must be separated so multiple counters can draw from one shared line.

(e)

4 marks

Point 1: Teams working across different international time zones experience communication delays and cannot easily hold synchronous face-to-face meetings, slowing down debugging and integration.
Point 2: Language barriers and cultural differences between distributed teams can cause misunderstandings of system specifications.
Point 3: Different teams may follow inconsistent coding conventions, naming styles, or data formats (such as date/currency representations), causing interface mismatches when modules are integrated.
Point 4: Coordinating version control and project management across remote sites is more complex.

Mentor tip

Make four clear points covering time zones, language/communication barriers, differing coding/format conventions, and coordination delays.

Question 16

14 marks

Boolean checkout flags, splitting a POSline queue, and open-source software ethics

Object-oriented programming

(a)

2 marks

A checkout counter has only two mutually exclusive states: open (true) or closed (false).
A boolean data type models binary true/false states directly while using the minimum possible memory (1 bit).

Mentor tip

Mention both the two-state nature of an open/closed counter and memory efficiency.

(b)

8 marks

Instantiate a new POSline object newLine to hold the transferred carts.
First pass - Copy every second cart (odd indices 1, 3, 5, ...) from oldLine into newLine using oldLine.getLine(i) and newLine.joinLine(...).
Second pass - Remove those copied carts from oldLine using oldLine.leaveLine(i): note that since leaveLine(i) shifts all subsequent carts down by 1 index, either iterate in reverse from the end of the line down to 0, or increment i by 1 on each removal in a forward pass!
Alternatively, combine copying and removal in a single loop using a boolean toggle flag:
public POSline split(POSline oldLine) { POSline newLine = new POSline(); int i = 0; while (i < 20 && oldLine.getLine(i) != null) { if (i % 2 == 1) { newLine.joinLine(oldLine.getLine(i)); } i++; } i--; // start from last valid index and remove odd indices in reverse while (i > 0) { if (i % 2 == 1) { oldLine.leaveLine(i); } i--; } return newLine; }

Mentor tip

(c)

4 marks

Issue 1 - Intellectual property attribution and licensing compliance: Even though open-source code is publicly available, it is still intellectual property governed by licenses (such as GPL or MIT); the supermarket company has an ethical obligation to credit the original authors and respect license terms rather than passing off community work as proprietary.
Issue 2 - Customer security and reliability accountability: Because open-source code can be inspected by anyone (or may have been modified by unvetted contributors), failing to audit and test adapted modules thoroughly could expose supermarket customers to malware or payment data breaches.

Mentor tip

Describe two ethical issues with elaboration: proper attribution/licensing of open-source creators' work, and the duty of care to test open-source code so customer payment data isn't compromised.

Question 17

20 marks

Singly linked list of CartNode objects (POSlist), pointer manipulation, and coding conventions

Object-oriented programming

(a)

2 marks

An object reference is a variable that holds a pointer to the memory address where an instantiated object is stored.

Mentor tip

State both 'pointer/memory address' and 'where the object is stored in memory' for full marks.

(b)

4 marks

Declare Cart result = null; and check if the list is non-empty (head != null).
Extract the Cart object from the first node via head.getCart(), advance head = head.getNext() to unhook the first node, and return result:
public Cart removeFirst() { Cart result = null; if (head != null) { result = head.getCart(); head = head.getNext(); } return result; }

Mentor tip

Always extract head.getCart() before advancing head = head.getNext(), and return null safely when head == null.

(c)

2 marks

Trace the operations in sequence:
1) addLast(cart2), addLast(cart1), addLast(cart4) -> [cart2 -> cart1 -> cart4]
2) removeFirst() removes cart2 -> [cart1 -> cart4]
3) addLast(cart5), addLast(cart3) -> [cart1 -> cart4 -> cart5 -> cart3]
4) removeFirst() removes cart1 -> final list contains 3 nodes in order:
head ---> [ cart4 | • ] ---> [ cart5 | • ] ---> [ cart3 | null ]

Mentor tip

Include the head (root) pointer at the start, the three remaining nodes in exact FIFO order (cart4 -> cart5 -> cart3), and the terminating null pointer.

(d)

2 marks

A standard queue is a strict First-In, First-Out (FIFO) abstract data structure that only permits insertions at the rear (enqueue) and removals from the front (dequeue).
This is unsuitable for a real supermarket checkout line because impatient customers in the middle of the line may leave or switch to a newly opened counter at any position.

Mentor tip

Connect the strict FIFO restriction of a queue (only removing from the front) to the need for customers in the middle of a line to leave or switch lanes.

(e)

6 marks

Handle the special case where n == 1 (removing the first node) by calling removeFirst().
Otherwise, traverse the list using curr and prev pointers from i = 1 up to n so curr points to the nth CartNode and prev points to node n - 1.
Retrieve curr.getCart(), bypass the nth node via prev.setNext(curr.getNext()), and return the removed Cart:
public Cart leaveList(int n) { if (n == 1) { return removeFirst(); } CartNode prev = null; CartNode curr = head; int i = 1; while (i < n) { prev = curr; curr = curr.getNext(); i++; } Cart result = curr.getCart(); prev.setNext(curr.getNext()); return result; }

Mentor tip

Remember to handle n == 1 separately (using removeFirst() or updating head), then use prev.setNext(curr.getNext()) to unhook node n.

(f)

4 marks

Elements of style and naming conventions: Using meaningful variable/method identifiers, consistent capitalization (such as CamelCase for classes and lowerCamelCase for methods/variables), proper code indentation/whitespace, and inline comments.
Why they are important: They make source code self-documenting and easier to read, simplify debugging and future maintenance extensions, and allow multiple developers on a team to collaborate seamlessly without confusion.

Mentor tip

Include at least two concrete style elements (meaningful identifiers, indentation, capitalization, comments) and two benefits (readability, easier maintenance/debugging, team collaboration).

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: Relational vs object-oriented databases, election data warehousin · Q4
Present two clear 2-mark arguments: industry standardization/tooling/staffing and simpler tabular modelling compared to OOP complexity.
02
Modelling and simulation: Optical character recognition (OCR) and neural networks for handw · Q8
State a cause (poor print quality or decorative/sans-serif fonts) and give a specific character confusion example (such as '5' vs 'S' or '1' vs 'I').
03
Web science: OrderYourFoods.com web protocols, server-side PHP email validatio · Q9
Commercial ordering websites handling customer addresses and payments use HTTPS.
04
Object-oriented programming: Singly linked list of CartNode objects (POSlist), pointer manipul · Q17
State both 'pointer/memory address' and 'where the object is stored in memory' for full marks.

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