Loading…

Worked solutions

IB Computer Science HL · Paper 2

Computer Science, Higher Level

Session: November 2018
Marks: 260
Duration: 1 hour 20 min
Calculator: Not allowed
Questions: 16

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

Keep your copy of the paper open alongside. We don't reproduce the questions: each part below is labelled to match the November 2018 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

17 marks

Information systems vs databases, banking ACID properties, and law enforcement access ethics

Databases

(a)

3 marks

Distinguish between an information system and a database:
1) A database is a structured collection of persistently stored data organized according to a specific data model (such as relational tables).
2) An information system is a comprehensive socio-technical system comprising hardware, software, telecommunications, business workflows, people, and the database itself.
3) The database functions as the foundational data storage engine within the broader information system.

Mentor tip

Describe a database as the organized data repository and an information system as the broader infrastructure encompassing people, procedures, and applications.

(b)

8 marks

Apply ACID transaction principles to a bank transfer between Account A and Account B:
1) Atomicity (All or Nothing): The entire transfer (debiting Account A and crediting Account B) constitutes a single indivisible unit of work. If a system failure happens after debiting A but before crediting B, the transaction is rolled back and no funds are deducted.
2) Consistency: The database must transition from one valid state to another, preserving financial integrity invariants (e.g. Account A balance $\geq 0$ and total bank ledger balance $Balance_{A} + Balance_{B}$ remains unchanged).
3) Isolation: Concurrent transactions operating simultaneously cannot observe intermediate states. If another client queries Account A while the transfer executes, they observe either the pre-transfer or post-transfer balance, preventing race conditions.
4) Durability: Once the transaction commits, all balance adjustments are permanently written to persistent storage (and transaction redo logs), surviving subsequent server crashes or power interruptions.

Mentor tip

Dedicate two distinct marking points to each of the four ACID letters, explicitly contextualizing every principle with the bank transfer scenario.

(c)

6 marks

Discuss government and law enforcement access to private financial databases:
1) Arguments in favour of authorized access (Security & Crime Prevention):
- Investigating money laundering, terrorist financing, and serious organized tax evasion requires following financial transaction trails.
- Judicial oversight via search warrants ensures access is targeted, justified, and proportional.
2) Arguments against unrestricted access (Privacy & Civil Liberties):
- Unrestricted mass surveillance breaches client financial confidentiality and fundamental privacy rights.
- Creating government backdoors or shared access vectors introduces severe cybersecurity vulnerabilities, risking identity theft and data leaks.
3) Synthesis / Conclusion:
- Unrestricted interrogation must be prohibited; access should be granted strictly under independent judicial warrant for substantiated criminal inquiries.

Mentor tip

Provide a balanced discussion contrasting national security/crime detection against customer privacy, and conclude with the necessity of judicial warrants.

Question 2

8 marks

DBMS definition, logical schema characteristics, DDL, and DBA security responsibilities

Databases

(a)

1 mark

Define Database Management System (DBMS):
A software suite that provides an interface and tools to define, create, query, update, administer, and maintain databases.

Mentor tip

State that it is software managing the storage, retrieval, and administration of structured data.

(b)

3 marks

Identify three key characteristics of a logical schema:
1) Independent of physical storage: It defines logical structures without specifying disk sector allocations, block sizes, or file paths.
2) Formal relational specifications: Specifies entities, attributes, primary keys, foreign keys, and domain constraints.
3) Business integrity rules: Encapsulates entity relationships (1:1, 1:M, M:N) and referential integrity constraints.

Mentor tip

Highlight physical independence, specification of tables/relationships, and enforcement of integrity constraints.

(c)

2 marks

Outline the purpose of Data Definition Language (DDL):
1) DDL commands (such as `CREATE`, `ALTER`, `DROP`) define and modify the structural blueprint and metadata of database objects.
2) It establishes table definitions, data types, indexes, and primary/foreign key constraints within the DBMS data dictionary.

Mentor tip

Emphasize creating and modifying table structures and constraints rather than manipulating data rows.

(d)

2 marks

Identify two tasks performed by a DBA to safeguard database security:
1) Role-based access control management: Creating user accounts, configuring access permissions, and assigning principle-of-least-privilege credentials.
2) Routine backup and encryption configuration: Scheduling encrypted periodic database backups and establishing restoration protocols against ransomware and hardware failure.

Mentor tip

Name two specific administrative actions: managing user permissions and configuring backups/encryption.

Question 3

20 marks

Greengrocer relation querying, update anomalies, and 3NF normalisation

Databases

(a)(i)

1 mark

Identify a suitable data type for Prod_Price:
Decimal / Currency / Fixed-point numeric (e.g. `DECIMAL(6,2)` or `NUMERIC`) to maintain precise monetary cents without floating point rounding error.

Mentor tip

Specify Currency, Decimal, or Real/Float.

(a)(ii)

3 marks

Construct the logical query steps:
Step 1: Inspect the GREENGROCER relation filtering rows where `Supp_Name = 'Veggy Co.'`.
Step 2: Apply the range predicate `Prod_Price >= 4.00 AND Prod_Price <= 10.00`.
Step 3: Project the `Prod_Name` attribute from matching records (yielding Celery and Pears).

Mentor tip

List the supplier equality condition, the closed range price check, and the projected attribute.

(b)

2 marks

Outline redundant data in the GREENGROCER relation:
1) Supplier contact details (`Supp_Contact` and `Supp_Phone`) are repeated across multiple rows whenever a supplier provides more than one product.
2) For instance, 'Fruit and Veggie' with 'Ines Bond' and '542 9800' is duplicated across Apples (P211) and Onions (P121), creating unnecessary storage waste.

Mentor tip

Point out the duplication of supplier contact names and phone numbers across multiple product rows.

(c)(i)

2 marks

Insertion anomaly problem and example:
Problem: When adding a new product supplied by an existing supplier, the supplier's contact details must be manually re-entered, risking inconsistent spelling or mismatched telephone entries.
Example: Inserting Potatoes supplied by 'Veggy Co.' with contact 'Mia Abiss' risks mismatching existing records where 'Veggy Co.' contact is 'Mike Oberst' (235 7656).

Mentor tip

Explain the risk of inconsistent re-entry of supplier data when creating new product records.

(c)(ii)

2 marks

Deletion anomaly problem and example:
Problem: Deleting the sole product supplied by a particular vendor inadvertently deletes all record of that vendor's existence.
Example: Deleting product P265 (Tomatoes) deletes the only row containing supplier 'New Fruits' (John Smith, 677 7895), erasing their contact info from the database.

Mentor tip

Demonstrate that deleting the only product from a vendor permanently destroys all record of that vendor.

(c)(iii)

2 marks

Modification anomaly problem and example:
Problem: Updating a supplier attribute requires identifying and altering every corresponding row across the table; missing even one tuple causes data inconsistency.
Example: If 'Fruit and Veggie' changes telephone to '542 1811', updating row P211 without updating row P121 leaves the database with contradictory phone records for the same firm.

Mentor tip

Illustrate partial updates leaving conflicting phone records across multiple product rows.

(d)

8 marks

Decompose GREENGROCER into 3NF relations:
1) Analysis of functional dependencies:
- Prod_ID -> Prod_Name, Prod_Price, Supp_ID
- Supp_ID -> Supp_Name, Supp_Phone
- Supp_Phone -> Supp_Contact (or Contact associated directly with Supplier)
2) Normalised Relations:
PRODUCT (Prod_ID [PK], Prod_Name, Prod_Price, Supp_ID [FK])
SUPPLIER (Supp_ID [PK], Supp_Name, Supp_Phone [FK])
SUPPLIER_CONTACT (Supp_Phone [PK], Supp_Contact)
3) Verification of normal forms:
- 1NF: All attribute values are atomic; no repeating groups exist.
- 2NF: The primary keys are non-composite (single attributes), so partial functional dependencies cannot exist.
- 3NF: Transitive dependencies are eliminated by placing supplier contact details in separate relations referencing foreign keys.

Mentor tip

Define Product, Supplier, and SupplierContact with primary and foreign keys clearly denoted.

Question 4

20 marks

Ministry of Tourism data warehousing, ETL migration, privacy ethics, cluster and link analysis

Databases

(a)

4 marks

Identify four information sources for the tourism data warehouse:
1) Transportation providers: Airport flight manifests, seaport passenger records, and rail/bus transit ticketing systems.
2) Accommodation providers: Hotel, hostel, and holiday rental booking registries detailing stay durations and room choices.
3) Financial / Card transaction data: Anonymized merchant payment transaction feeds from retail, dining, and currency exchanges.
4) Tourism agency portals: Online visitor reviews, guided tour bookings, and national park visitor center passes.

Mentor tip

Name four distinct sources: transport gateways, hotels/lodging, commercial card transactions, and visitor booking portals.

(b)

4 marks

Outline two differences between a transactional database and a data warehouse:
Difference 1: Operational vs Analytical Purpose:
- An operational database supports real-time OLTP transactions (read/write updates for daily transactions).
- A data warehouse is optimized for OLAP analytics, executing read-only aggregated queries across massive historical datasets for strategic decision support.
Difference 2: Temporal Scope and Normalisation:
- Operational databases hold current, highly normalized snapshots to eliminate write anomalies.
- Data warehouses hold non-volatile, denormalized historical time-series spanning multiple years.

Mentor tip

Contrast OLTP (current, normalized, transactional) with OLAP (historical, denormalized, analytical).

(c)

3 marks

Explain how ETL processes resolve data migration challenges:
1) Extraction: Pulls raw heterogeneous data from incompatible legacy databases (SQL, spreadsheets, flat files) without interrupting live operations.
2) Transformation: Cleanses data by removing whitespace, resolving character encoding mismatches, standardizing date/currency formats, and deduplicating records.
3) Loading: Loads validated, uniform data records into targeted multidimensional star/snowflake schemas in the warehouse.

Mentor tip

Walk through Extract (pulling heterogeneous sources), Transform (cleansing, standardizing formats), and Load (inserting into warehouse).

(d)

2 marks

Outline one ethical dilemma resulting from tourism data mining:
Invasive behavioral profiling and loss of privacy: Correlating transit, hotel, and financial records creates intimate behavioral profiles tracking tourists' movements and lifestyle habits without their explicit informed consent.

Mentor tip

Focus on tracking individual movements and profiling habits without informed consent.

(e)

4 marks

Explain how cluster analysis enhances tourism marketing campaigns:
1) Segmenting tourist demographics: Unsupervised cluster algorithms group visitors into distinct cohorts based on shared attributes (e.g. expenditure patterns, origin country, family status).
2) Tailored campaign strategy:
- Cohort 1 ('Budget Eco-travelers'): Market affordable youth hostels, hiking trails, and eco-tours.
- Cohort 2 ('Luxury Cultural visitors'): Advertise high-end boutique hotels, historical museums, and fine dining.
3) Maximizing ROI: Customizing advertising to receptive audiences yields higher conversion rates and maximizes municipal tourism revenue.

Mentor tip

Define partitioning visitors into homogeneous clusters and give two specific tourism campaign examples.

(f)

3 marks

Explain the significance of link analysis in data mining:
1) Link analysis examines relationships and interconnected networks between disparate entities (e.g. tourists, hotels, flights, credit cards).
2) It uncovers hidden structural associations (e.g. discovering that tourists visiting Attraction A frequently patronize Restaurant B in another district).
3) In fraud and security detection, it identifies coordinated booking rings or irregular financial pathways across multiple databases.

Mentor tip

Explain analyzing connections between entities to discover hidden relationship networks and fraud patterns.

Option B: Modelling and simulation · Option

Questions 5–8 · 65 marks

Question 5

15 marks

Water utility billing model, rollover logic, and pseudocode

Modelling and simulation

(a)

3 marks

Specify data types and value ranges for model variables:
1) `METER_TYPE`: String or Enumerated type {'old', 'new'}.
2) `WATER_IN`: Integer in the range [0, 999999] litres.
3) `WATER_OUT`: Integer in the range [0, 999999] litres.
4) `INFRA_CHARGE`: Currency / Real number $\geq 0$ (default $204.80$ ).

Mentor tip

State both the programmatic data type and the allowable numeric or enumerated range.

(b)

2 marks

Explain how yearly water consumption is calculated accounting for meter rollover:
1) Normal condition: If $NewReading \geq OldReading$ , then $Volume = NewReading - OldReading$ .
2) Rollover condition: If $NewReading < OldReading$ (the 6-digit display rolled over past 999,999 to 0), then $Volume = (1000000 - OldReading) + NewReading$ .

Mentor tip

Describe standard subtraction when reading increases and handling rollover past 999,999.

(c)

7 marks

Construct pseudocode to calculate annual household water bills:
```text
input METER_TYPE, PREV_IN, CURR_IN
if CURR_IN >= PREV_IN then
WATER_SUPPLIED = CURR_IN - PREV_IN
else
WATER_SUPPLIED = (1000000 - PREV_IN) + CURR_IN
end if
WATER_CHARGE = WATER_SUPPLIED * 1.90
INFRA_CHARGE = 204.80
if METER_TYPE == 'new' then
input PREV_OUT, CURR_OUT
if CURR_OUT >= PREV_OUT then
WATER_RETURNED = CURR_OUT - PREV_OUT
else
WATER_RETURNED = (1000000 - PREV_OUT) + CURR_OUT
end if
if WATER_RETURNED >= 0.95 * WATER_SUPPLIED then
INFRA_CHARGE = INFRA_CHARGE * 0.90
end if
end if
TOTAL_BILL = WATER_CHARGE + INFRA_CHARGE
output TOTAL_BILL
```

Mentor tip

Ensure water charges ($1.90/L), meter rollover subtraction, and 10% discount when returned >= 95% are properly sequenced.

(d)

3 marks

Identify three actions customers can take to reduce infrastructure charges:
1) Upgrade equipment: Upgrade to a new smart meter capable of measuring returned waste water.
2) Maximize wastewater return: Ensure domestic wastewater drains directly into municipal sewage pipes rather than external garden runoff, exceeding the 95% return threshold.
3) Repair domestic leaks: Fix pipe leaks on private property to prevent water from escaping unmetered before returning to the sewage plant.

Mentor tip

Suggest upgrading to the new meter, minimizing unreturned garden usage, and eliminating household leaks.

Question 6

15 marks

Household water consumption model vs simulation and spreadsheet distribution

Modelling and simulation

(a)

4 marks

Distinguish between a model and a simulation using the water scenario:
1) A model is a conceptual and mathematical representation: Here, the model comprises the table of static activity values (15 L per dishwasher cycle, 70 L per bath) and rules governing weekly totals.
2) A simulation is the execution and dynamic exploration of the model over time: Here, running the simulation involves varying activity frequencies (e.g. 5 showers vs 3 baths per week) to evaluate projected weekly water usage and test scenarios.

Mentor tip

Contrast the static representation of mathematical equations/data with the dynamic execution and scenario testing.

(b)

4 marks

Explain how data accuracy affects simulation quality:
1) Garbage In, Garbage Out: The table provides average estimates that ignore appliance variances (e.g. high-efficiency vs old washing machines) and personal behavioral durations (e.g. a 4-minute shower vs a 15-minute shower).
2) Systematic error propagation: If baseline estimates are inaccurate, projected totals will deviate significantly from actual meter readings, reducing trust in the model for decision-making.

Mentor tip

Discuss appliance variations, variable human durations, and how flawed initial inputs compromise output reliability.

(c)

5 marks

Outline spreadsheet organization to maintain consumption under 2700 L/week:
1) Layout structure: Create columns for Activity Name, Baseline Litres Per Event, Weekly Frequency (variable input cells), and Total Litres ( $= Litres \times Frequency$ ).
2) Aggregation: A summary cell calculates `=SUM(Total Litres Range)`.
3) Conditional constraints: Incorporate conditional formatting (highlighting red if total $> 2700$ ) or an alert formula `=IF(Total <= 2700, "Target Met", "Over Target")`.
4) What-if experimentation: Family members adjust frequency parameters (e.g. reducing bath frequency) to bring the total below 2700 L.

Mentor tip

Specify input cells, product formulas, the SUM function, conditional alerts for > 2700 L, and parameter adjustments.

(d)

2 marks

Identify two reasons why the model fails for house-sharing young adults:
1) Independent, non-cooperative consumption: Roommates run separate partial laundry cycles and cook individually, generating far more frequent appliance usages than a cohesive family.
2) Longer and more varied personal habits: Individual shower durations and visitor water usage are substantially higher than conservative family averages.

Mentor tip

Point out independent non-consolidated appliance runs and uncoordinated usage habits.

Question 7

15 marks

Surgical simulation training, smart glove haptics, and 3D visualization

Modelling and simulation

(a)

4 marks

Outline one advantage and one disadvantage of surgical simulation software:
1) Advantage: Patient safety and risk-free repeatability. Trainees practice delicate operative techniques repeatedly under varied pathological conditions without risking patient morbidity or malpractice.
2) Disadvantage: Absence of full physical fidelity. Simulations cannot perfectly recreate tissue tactile resistance, unexpected arterial hemorrhaging, or the psychological stress of live surgery.

Mentor tip

Balance safe repetitive skill mastery with the lack of tactile fidelity and unpredictable biological complexity.

(b)

3 marks

Identify three technical requirements for smart surgery gloves:
1) Spatial tracking accuracy: High-resolution inertial/optical sensors providing six degrees of freedom (6-DoF) position and orientation tracking with sub-millimeter precision.
2) Haptic force feedback: Actuators providing variable resistance to simulate cutting or suturing different tissue densities.
3) Ultra-low latency communication: Real-time data transmission to the display engine ( $< 10$ ms) to prevent lag and visual-tactile desynchronization.

Mentor tip

Specify precise position/orientation tracking, tactile force feedback, and low-latency response.

(c)

2 marks

Define the term visualization:
The graphical representation of abstract data or spatial models into intuitive visual imagery (2D or 3D displays) to facilitate human cognitive comprehension and interactive analysis.

Mentor tip

Define converting structured data or physical models into visual graphical representations.

(d)

6 marks

Discuss whether upgrading to a 3D display improves surgical training:
1) Arguments in favour of 3D display:
- Depth perception: Stereoscopic 3D provides accurate depth cues and spatial awareness, essential for estimating incision depth and avoiding accidental puncture of underlying blood vessels.
- Realistic anatomical spatial relationships: Organs and vasculature are layered in 3D volume; viewing true perspective improves hand-eye coordination.
2) Counterarguments / Limitations:
- Hardware fatigue: Wearing 3D glasses or headsets during extended simulations can induce eye strain, motion sickness, and ergonomic fatigue.
- High computational expense: Rendering stereoscopic 3D requires high-end GPU hardware to maintain high frame rates without stuttering.
3) Conclusion: A 3D system substantially elevates anatomical realism and depth judgment, outweighing hardware overhead for surgical education.

Mentor tip

Examine depth perception and spatial accuracy vs visual fatigue and computing power, drawing a reasoned conclusion.

Question 8

20 marks

Genetic algorithm stages, neural network supervised learning, and voice response design

Modelling and simulation

(a)

6 marks

Outline the iterative stages of a Genetic Algorithm (GA):
1) Initial population generation: Initialize a diverse population of trial solutions encoded as genetic strings (chromosomes).
2) Fitness evaluation: Score each chromosome using a defined objective fitness function measuring solution quality.
3) Selection: Select pairs of chromosomes for reproduction using fitness-proportionate selection (e.g. roulette wheel or tournament selection).
4) Crossover (Recombination): Combine segments from selected parent chromosomes at specified crossover points to produce offspring.
5) Mutation: Apply low-probability random bit flips or perturbations to offspring genes to maintain genetic diversity and prevent premature convergence.
6) Replacement and Termination: Replace the previous generation with the new offspring population and repeat until convergence criteria or generational limits are reached.

Mentor tip

List the six sequential steps: initialize population, evaluate fitness, selection, crossover, mutation, and replacement/termination.

(b)(i)

4 marks

Outline why neural networks require training before aerial image analysis:
1) Random initial weights: At initialization, synaptic weights connecting network neurons have random values, producing arbitrary outputs.
2) Feature pattern abstraction: Training on aerial imagery allows the network to adjust weights iteratively, learning to recognize complex visual features (leaf discoloration, canopy thinning, necrotic textures).
3) Generalization: Proper training ensures the network generalizes to unobserved groves across diverse lighting and terrain without overfitting.

Mentor tip

Explain starting from random weights, learning visual patterns (discoloration/canopy loss), and achieving generalization.

(b)(ii)

6 marks

Explain why supervised learning is preferred over unsupervised learning for disease prediction:
1) Ground-truth classification: Supervised learning utilizes labelled ground-truth datasets where each aerial image is paired with verified botanical diagnostics ('infected' vs 'healthy').
2) Direct error minimization: The network calculates the precise classification error between its prediction and verified reality, backpropagating weight updates to minimize misdiagnoses.
3) Limitations of unsupervised learning: Unsupervised clustering groups images purely on visual similarities (e.g. shadows, soil color, tree density) without understanding which variations signify parasitic infection, leading to unaligned clusters.
4) Predictive precision: For disease progression forecasting, supervised regression/classification directly correlates canopy symptoms with disease severity.

Mentor tip

Contrast labelled ground truth and error minimization in supervised learning with unguided clustering based on irrelevant visual features in unsupervised learning.

(c)

4 marks

Explain why telephone voice response system prompts must be carefully curated:
1) Natural language ambiguity and speech recognition constraints: Prompts must present clear, closed-ended choices (e.g. 'Say billing or support') so acoustic models can accurately resolve vocabulary within a constrained lexicon.
2) User cognitive load and frustration mitigation: Overly lengthy or convoluted audio menus increase caller cognitive fatigue, leading to incorrect selections, abandoned calls, and customer dissatisfaction.

Mentor tip

Explain constraining the vocabulary to improve speech recognition accuracy and keeping prompts concise to prevent user frustration.

Option C: Web science · Option

Questions 9–12 · 65 marks

Question 9

16 marks

Online newspaper audio podcast publishing, compression, and client personalization

Web science

(a)

2 marks

Identify two actions relating to podcast provenance prior to publication:
1) Copyright and syndication verification: Secure explicit copyright licenses, syndication permissions, and broadcast rights from originating agencies.
2) Content validation and attribution: Verify factual accuracy and ensure appropriate intellectual property attribution is credited to original creators.

Mentor tip

Focus on licensing/copyright clearance and editorial/source attribution.

(b)

2 marks

Outline one technical reason for limiting available online podcasts to eight:
1) Server bandwidth and storage management: Limiting concurrent hosted audio streams prevents network saturation and eases server disk consumption.
2) Mobile UI rendering and page weight: Restricting list length keeps web page DOM size small, improving mobile download speed and preventing vertical scrolling clutter.

Mentor tip

Explain server bandwidth/storage control or lightweight client-side DOM loading on mobile devices.

(c)

1 mark

State one standard audio file format:
MP3 (or AAC, WAV, FLAC, OGG).

Mentor tip

State MP3, AAC, or WAV.

(d)

5 marks

Compare lossy vs lossless audio compression for streaming and downloading:
1) Lossy compression (e.g. MP3, AAC): Permanently removes psychoacoustically inaudible frequencies, achieving massive file size reductions ( $80 - 90%$ ). Spoken dialogue remains crystal clear, and reduced file size enables fast downloads and interruption-free streaming on mobile networks.
2) Lossless compression (e.g. FLAC, ALAC): Reconstructs original acoustic waveforms perfectly without data loss, but yields much larger files ( $50 - 60%$ of original size), consuming unnecessary bandwidth with no audible benefit for speech.
3) Evaluation: For 10-minute spoken interviews, lossy compression is optimal because transmission speed and buffer reliability outweigh minor acoustic fidelity differences.

Mentor tip

Contrast file size reduction and streaming reliability with acoustic fidelity, noting that spoken podcasts benefit most from lossy formats.

(e)

2 marks

Outline one technical cause of streaming audio buffering:
Network throughput fluctuation / Bandwidth bottleneck: If local Internet download speed drops below the media bit rate (due to Wi-Fi interference or ISP peak congestion), the playback buffer drains faster than it fills, halting audio until the buffer replenishes.

Mentor tip

Connect throughput dropping below stream bitrate to the client buffer running dry.

(f)

4 marks

Explain the technology enabling personalized podcast ordering:
1) Client tracking: Client interactions are recorded via persistent cookies, localStorage, or user login profile histories tracking preferences for science topics.
2) Dynamic server/client scripting: When loading the podcast section, an API request passes user profile metadata to the backend or executes client-side JavaScript.
3) Dynamic DOM reordering: Scripting sorts podcast elements, querying metadata tags (e.g. `genre = 'science'`) to hoist matching items to the top of the list before rendering.

Mentor tip

Explain tracking user category preference via cookies/history and using dynamic scripting to re-order the items.

Question 10

14 marks

E-commerce URL security, DNS resolution hierarchy, and PHP session cart management

Web science

(a)

2 marks

Describe how the HTTPS URL ensures communication security:
1) Protocol identifier: The `https://` scheme invokes Transport Layer Security (TLS/SSL).
2) Cryptographic protection: It guarantees symmetric encryption of all transmitted packet payloads and authenticates the BesTea server using X.509 certificates to thwart eavesdropping and spoofing.

Mentor tip

Highlight the https protocol, TLS encryption, and certificate-based server authentication.

(b)

4 marks

Identify the hierarchical steps in DNS resolution for `www.BesTea.com`:
1) Local resolver cache check: The browser and local OS check their DNS resolver caches; if missed, a query is dispatched to the recursive DNS server (ISP resolver).
2) Root nameserver inquiry: The recursive resolver contacts a Root DNS nameserver, which returns the IP address for the `.com` Top-Level Domain (TLD) nameserver.
3) TLD nameserver inquiry: The `.com` TLD server directs the resolver to the authoritative nameserver for `BesTea.com`.
4) Authoritative lookup and return: The authoritative nameserver looks up the A/AAAA record for `www.BesTea.com` and returns the destination IP address to the resolver, which caches it and provides it to the browser.

Mentor tip

Walk through: local cache -> recursive resolver -> Root server -> TLD server -> authoritative server -> browser.

(c)

2 marks

Outline how the footer is maintained uniformly across all web pages:
1) Server-side templating / includes: The footer HTML is maintained in a single separate component file (e.g. `footer.php`) incorporated into each page template via `include('footer.php');`.
2) Global CSS stylesheet: Global styling rules in a shared external CSS file format the footer consistently across the entire site.

Mentor tip

Mention server-side includes (e.g. PHP include/template partial) or a shared external CSS stylesheet.

(d)

4 marks

Explain the function of the provided PHP script fragment:
1) Session initialization: `$_SESSION['sessionUserID'] = new_random_number();` generates and stores a unique session identifier in the global session superglobal to track the visiting user.
2) Database query execution: Constructs and executes an SQL SELECT query (`SELECT * FROM users_db WHERE id = ...`) querying `users_db` for the record matching the session ID.
3) Record fetching: Fetches the resulting row into the `$userRecord` object/array, allowing the checkout workflow to populate user details.

Mentor tip

Explain generating a session identifier, querying the database with it, and fetching the record into an object.

(e)

2 marks

Identify user-provided data from the PHP code snippet:
1) Action command: The parameter `action` containing `'removeBasketItem'`.
2) Item identifier: The parameter `id` indicating the specific product to be deleted from the shopping basket.

Mentor tip

Name the action string ('removeBasketItem') and the product id passed in the request.

Question 11

15 marks

Charity rural connectivity challenges, cloud migration, and search engine optimization ethics

Web science

(a)

2 marks

Outline one technical challenge for hosting a website on local rural equipment:
Satellite latency and erratic power/bandwidth: Satellite connections suffer high latency and frequent weather disruptions, while aging local computers lack server redundancy and unmetered upstream bandwidth to handle inbound visitor traffic.

Mentor tip

Mention high satellite latency, limited uplink bandwidth, or power/hardware unreliability.

(b)

4 marks

Explain how cloud-based services overcome local hardware and connectivity constraints:
1) Offloaded hosting and global availability: Hosting the website on public cloud servers ensures high uptime, automated backups, and low latency globally regardless of local internet blackouts at Harvest's base.
2) Eliminating local hardware maintenance: Cloud SaaS/PaaS shifts the burden of server maintenance, security patching, and processing power away from obsolete local machines to enterprise data centers.

Mentor tip

Focus on high availability, external server maintenance, and freeing local machines from hosting burdens.

(c)

4 marks

Describe two problems arising from hosting data with an overseas MEDC cloud provider:
1) Data sovereignty and cross-border regulatory jurisdiction: Storing charity data in a foreign jurisdiction subjects files to foreign surveillance laws, potentially conflicting with local privacy protections.
2) Vendor lock-in and foreign currency subscription overhead: Ongoing cloud subscription costs priced in foreign currencies fluctuate with exchange rates, straining charity budgets, while migrating off proprietary platforms is technically arduous.

Mentor tip

Describe data sovereignty/legal jurisdiction issues and recurring foreign currency costs/vendor lock-in.

(d)

5 marks

Discuss whether Harvest should use black hat SEO techniques:
1) Potential short-term perceived benefit: Techniques like keyword stuffing, hidden text, and link farms might temporarily boost search rankings, attracting immediate donors.
2) Severe technical and ethical consequences:
- Search engine penalties: Algorithmic crawlers detect manipulative tactics and completely de-index the domain, eradicating web visibility.
- Reputational damage: A charitable non-profit relying on deceptive tricks destroys institutional trust with international donors.
3) Conclusion: Black hat SEO presents intolerable operational and reputational risks; Harvest must adopt sustainable white hat practices (high quality authentic content, semantic markup, and credible outreach).

Mentor tip

Contrast temporary ranking boosts against devastating algorithmic bans and loss of donor trust; conclude strongly against black hat.

Question 12

20 marks

Tapetum lucidum semantic search, ontologies vs folksonomies, and collective intelligence

Web science

(a)

4 marks

Explain why a semantic multimedia search is preferable to a keyword search:
1) Acoustic / Visual query input: The student may possess an audio recording of a nocturnal animal's call or a photo of eyes glowing in the dark without knowing the species name.
2) Semantic web conceptual linking: Instead of matching the string 'tapetum lucidum', semantic web engines query multimedia ontologies linking visual eye-shine and acoustic calls to nocturnal biological taxonomies, identifying the animal directly.

Mentor tip

Contrast keyword matching with querying semantic multimedia ontologies that link recorded sounds or eye-shine images to species taxonomies.

(b)

3 marks

Identify three distinctions between an ontology and a folksonomy:
1) Authorship: Ontologies are authored by domain knowledge engineers; folksonomies are crowdsourced informally by web users.
2) Vocabulary control: Ontologies enforce strict, controlled vocabularies and formal hierarchical logic (taxonomies); folksonomies are uncontrolled, free-form tags.
3) Semantic expressivity: Ontologies express precise relationships (e.g. `subClassOf`, `partOf`); folksonomies consist of flat, ambiguous associative tags.

Mentor tip

Highlight expert formal design vs crowdsourced tags, controlled vs uncontrolled vocabulary, and formal taxonomy vs flat associations.

(c)

6 marks

Discuss integrating folksonomy expressivity with ontology authoritativeness:
1) Benefits of integration:
- Folksonomies capture contemporary linguistic usage, emerging trends, and diverse user viewpoints that rigid ontologies miss.
- Ontologies provide structural disambiguation, mapping ambiguous tags (e.g. 'cat's eye' marble vs feline eye) to formal semantic concepts.
2) Integration mechanisms and challenges:
- Social tagging clustering: Applying NLP and clustering to social tags extracts potential terms to expand and enrich ontologies dynamically.
- Noise and vandalism risk: Uncurated user tags introduce misspellings, spam, and subjective labels, necessitating automated confidence thresholds and human expert moderation.
3) Conclusion: Hybrid systems bridging folksonomic tags into ontology concepts via weighted semantic mappings achieve both adaptability and rigor.

Mentor tip

Discuss using folksonomies to surface emerging vocabulary while using ontologies to resolve ambiguity, noting noise/spam moderation challenges.

(d)

4 marks

Explain the role of web crawlers in building search engine indexes:
1) Continuous discovery: Crawlers systematically follow hyperlinks starting from seed URLs to discover newly published and modified web documents.
2) Content harvesting: Crawlers fetch raw HTML content, image metadata, and text documents, caching them locally in document repositories.
3) Indexing pipeline feed: Downloaded pages are parsed, tokenized, and passed to indexing algorithms that construct and update inverted index posting lists.

Mentor tip

Describe traversing links from seeds, retrieving document content, and feeding raw text into the indexing pipeline.

(e)

3 marks

Explain how collective intelligence fosters successful open-source software development:
1) Decentralized peer contribution: Thousands of independent developers inspect code, report edge-case bugs, and submit pull requests across global time zones.
2) Rapid code refinement: Linus's Law ('Given enough eyeballs, all bugs are shallow') ensures security flaws and performance bottlenecks are audited and resolved rapidly.
3) Meritocratic governance: Projects like Linux utilize rigorous peer review and automated testing frameworks, integrating the best global engineering ideas into a stable public good.

Mentor tip

Explain distributed global peer code review, Linus's Law for rapid bug remediation, and meritocratic pull request integration.

Option D: Object-oriented programming · Option

Questions 13–16 · 65 marks

Question 13

21 marks

Airport Flight and Arrival classes, OOP principles, UML diagram, and linear search

Object-oriented programming

(a)(i)

2 marks

Outline the nature of an object in OOP:
1) An object is an instantiated runtime instance of a class that encapsulates both state (data held in instance variables) and behavior (functions/methods).
2) It acts as an autonomous software entity interacting with other objects through message passing (method calls).

Mentor tip

Mention that an object encapsulates state (attributes) and behavior (methods) as an instance of a class.

(a)(ii)

4 marks

Describe two disadvantages of Object-Oriented Programming:
1) Memory and runtime execution overhead: Encapsulation, dynamic method dispatch, and object reference wrappers require more memory and processing cycles than procedural code.
2) Steep learning curve and architectural complexity: Designing class hierarchies, inheritance, and interfaces requires substantial upfront planning; improper design leads to bloated, overly complex codebases.

Mentor tip

Cite memory/processing overhead from dynamic dispatch and complex architectural design requirements.

(a)(iii)

2 marks

Outline one advantage of modularity in software development:
Reusability and maintainability: Independent, self-contained modules can be tested, debugged, and maintained in isolation without side effects on the rest of the codebase, and can be reused across future projects.

Mentor tip

Focus on isolated testing, easier debugging, and reusable modules.

(b)

1 mark

State the relationship between Flight and Arrival:
Aggregation (or Association): An `Arrival` object 'has-a' `Flight` object as an instance variable (`myFlight`).

Mentor tip

State Aggregation (or 'has-a' association).

(c)

4 marks

Construct the UML diagram for the Arrival class:
+---------------------------------------------------+
| Arrival |
+---------------------------------------------------+
| - myFlight : Flight |
| - sta : String |
| - runway : int |
| - gate : String |
| - delay : int |
| - landed : boolean |
+---------------------------------------------------+
| + Arrival(myFlight : Flight, sta : String) |
| + addDelay(newDelay : int) : void |
| + getETA() : String |
| + compareWith(flightID : String) : int |
| + compareWith(anotherArrival : Arrival) : int |
+---------------------------------------------------+

Mentor tip

Draw the 3 compartments: class name, private attributes with types, and public methods with parameters and return types.

(d)

4 marks

Construct the `search(String flightID)` linear search method:
```java
private int search(String flightID)
{
for (int i = 0; i <= last; i++)
{
if (inbound[i].compareWith(flightID) == 0)
{
return i;
}
}
return -1;
}
```
Explanation: Loop from index 0 up to `last`. For each arrival, call `compareWith(flightID)`. When it returns 0, the target is located; return the index `i`.

Mentor tip

Iterate from 0 to last, call compareWith(flightID) == 0, and return the matching index.

(e)(i)

2 marks

Outline one advantage of binary search over linear search:
Logarithmic time efficiency: Binary search runs in $O (lo g n)$ time by repeatedly halving the search space, performing far fewer comparisons than linear search ( $O (n)$ ) on large sorted datasets.

Mentor tip

State $O (lo g n)$ logarithmic speed and dividing the search space in half.

(e)(ii)

2 marks

Outline one disadvantage of binary search:
Mandatory sorted order requirement: Binary search requires the array to remain strictly sorted by the search key; searching on an unsorted attribute (or maintaining sorting during frequent insertions) incurs significant overhead.

Mentor tip

State the strict requirement that elements must already be sorted by the search attribute.

Question 14

8 marks

Data hiding security, method overloading, and arrival ETA comparison

Object-oriented programming

(a)

2 marks

Outline how data hiding acts as a security feature in OOP:
1) Declaring instance variables `private` encapsulates internal state and prevents direct external tampering by client classes.
2) State mutations can only occur via controlled public accessor/mutator methods that enforce business validation rules.

Mentor tip

Explain private encapsulation preventing arbitrary external tampering, and enforcing validation through methods.

(b)(i)

2 marks

Outline why two `compareWith` methods do not cause a conflict:
1) Method overloading allows multiple methods to share the same name provided their parameter signatures differ.
2) The compiler differentiates them by their arguments: `compareWith(String)` takes a `String`, whereas `compareWith(Arrival)` takes an `Arrival` object.

Mentor tip

Explain that distinct parameter types allow the compiler to distinguish between the two calls.

(b)(ii)

1 mark

State the name of this OOP property:
Method overloading (or compile-time / static polymorphism).

Mentor tip

State method overloading or polymorphism.

(c)

3 marks

Construct `compareWith(Arrival anotherArrival)`:
```java
public int compareWith(Arrival anotherArrival)
{
String thisETA = this.getETA();
String otherETA = anotherArrival.getETA();
return thisETA.compareTo(otherETA);
}
```
Explanation: Retrieve the estimated arrival time strings for both objects using `getETA()`, invoke `compareTo()`, and return the integer result directly.

Mentor tip

Obtain getETA() from both this and anotherArrival, then return thisETA.compareTo(otherETA).

Question 15

16 marks

Method signature definition, delayed flight display, and array insertion

Object-oriented programming

(a)

2 marks

Define method signature:
The method name combined with its ordered parameter list (the number, order, and data types of its formal parameters). Return types and access modifiers are not part of the signature in Java.

Mentor tip

State that a signature consists of the method identifier and the parameter types/order.

(b)

4 marks

Construct `showDelayed(String t)`:
```java
public void showDelayed(String t)
{
for (int i = 0; i <= last; i++)
{
Arrival arr = inbound[i];
if (arr.getDelay() > 0 && !arr.isLanded())
{
if (arr.getETA().compareTo(t) < 0)
{
System.out.println(arr.getMyFlight().getId());
}
}
}
}
```
Explanation: Loop from 0 to `last`. For each flight, verify that delay $> 0$ , `landed == false`, and `getETA().compareTo(t) < 0`. If all match, output the flight ID.

Mentor tip

Check three conditions: delay > 0, not landed, and ETA lexicographically before t.

(c)

6 marks

Construct `add(Arrival newArrival)` maintaining sorted order by ETA:
```java
public void add(Arrival newArrival)
{
int i = last;
while (i >= 0 && inbound[i].compareWith(newArrival) > 0)
{
inbound[i + 1] = inbound[i]; // shift right
i--;
}
inbound[i + 1] = newArrival; // insert at sorted position
last++;
}
```
Explanation: Implement standard insertion sort insertion: start at index `last`, shift all elements with larger ETA one position to the right, insert `newArrival` into the vacated slot, and increment `last`.

Mentor tip

Shift elements right while greater than newArrival using compareWith, place newArrival, and increment last.

(d)

4 marks

Describe how to update flight delay and maintain sorted order using provided methods:
1) Step 1: Call `remove(flightID)` to extract and remove the target `Arrival` object from the array; `remove` shifts subsequent elements left and decrements `last`.
2) Step 2: Invoke `addDelay(newDelay)` on the removed `Arrival` object to update its delay and recalculate its ETA.
3) Step 3: Call `add(arrivalObject)` passing the updated arrival object; `add` inserts it into its newly sorted position according to its revised ETA.

Mentor tip

Clearly describe the 3-step sequence: remove the flight using remove(), update its delay using addDelay(), and re-insert it using add().

Question 16

20 marks

Dual-runway airport LinkedList operations, recursion tracing, and list merging algorithm

Object-oriented programming

(a)

2 marks

Outline one advantage of utilizing standard software libraries:
Reliability and development productivity: Library collections (e.g. `java.util.LinkedList`) are pre-tested, debugged, and optimized by systems engineers, accelerating development and preventing algorithmic errors.

Mentor tip

State pre-tested reliability, performance optimization, and saving development time.

(b)

4 marks

Evaluate dynamic linked lists vs static arrays for managing airport arrivals:
1) Dynamic sizing vs Fixed allocation:
- A static array has a fixed capacity ( $200$ ); an unexpected surge in incoming flights causes overflow crashes, while low traffic wastes memory.
- A linked list allocates nodes dynamically on demand, scaling gracefully without artificial size limits.
2) Insertion / Deletion efficiency:
- Inserting an arrival into a sorted array requires shifting on average $n /2$ elements ( $O (n)$ memory writes).
- In a linked list, inserting or reordering nodes simply updates pointer references without physical array copying.

Mentor tip

Compare fixed vs dynamic memory sizing and shifting elements in arrays vs updating pointers in linked lists.

(c)

2 marks

Determine the position and sketch the linked list after inserting RO225:
Existing Runway 1 list sorted by ETA:
1) KL1102 (STA 11:05, delay 0 -> ETA 11:05)
2) MH9077 (STA 11:45, delay 45 -> ETA 12:30)
3) LH8803 (STA 12:55, delay 0 -> ETA 12:55)
New flight to insert: RO225 (STA 12:05, delay 0 -> ETA 12:05).
Comparing ETAs: 11:05 < 12:05 < 12:30.
RO225 is inserted between KL1102 and MH9077:
[Head: KL1102 (11:05)] -> [RO225 (12:05)] -> [MH9077 (12:30)] -> [LH8803 (12:55)] -> null

Mentor tip

Calculate ETA 12:05 for RO225 and place it between KL1102 (11:05) and MH9077 (12:30).

(d)(i)

1 mark

Define recursion:
A programming technique in which a method calls itself directly or indirectly to solve smaller sub-problems until reaching a base terminating condition.

Mentor tip

Define a method calling itself with a base condition.

(d)(ii)

4 marks

Trace the execution of `process(runway1, 0)`:
Method code:
```java
void process(LinkedList<Arrival> myList, int i) {
if (i < myList.size()) {
process(myList, i + 1);
System.out.println(myList.get(i).getETA());
}
}
```
List elements: index 0 = KL1102 (ETA 11:05), index 1 = MH9077 (ETA 12:30), index 2 = LH8803 (ETA 12:55). `size() == 3`.
Tracing call stack progression:
`i = 0`: calls `process(runway1, 1)`
`i = 1`: calls `process(runway1, 2)`
`i = 2`: calls `process(runway1, 3)`
`i = 3`: condition `3 < 3` is false; returns.
Unwinding call frames (post-order printing):
Frame `i = 2`: prints `myList.get(2).getETA()` -> 12:55
Frame `i = 1`: prints `myList.get(1).getETA()` -> 12:30
Frame `i = 0`: prints `myList.get(0).getETA()` -> 11:05
Construct trace table:
i | id | output
0 | KL1102 | -
1 | MH9077 | -
2 | LH8803 | -
3 | - | -
2 | LH8803 | 12:55
1 | MH9077 | 12:30
0 | KL1102 | 11:05

Mentor tip

Notice that printing occurs after the recursive call, outputting ETAs in reverse order: 12:55, 12:30, 11:05.

(e)

7 marks

Construct `mergeLists()` to merge two sorted LinkedLists:
```java
public LinkedList<Arrival> mergeLists()
{
LinkedList<Arrival> result = new LinkedList<Arrival>();
while (!runway1.isEmpty() && !runway2.isEmpty())
{
if (runway1.getFirst().compareWith(runway2.getFirst()) <= 0)
{
result.addLast(runway1.removeFirst());
}
else
{
result.addLast(runway2.removeFirst());
}
}
while (!runway1.isEmpty())
{
result.addLast(runway1.removeFirst());
}
while (!runway2.isEmpty())
{
result.addLast(runway2.removeFirst());
}
return result;
}
```
Algorithm: Continuously compare head elements of both lists using `compareWith`. Remove the earlier flight with `removeFirst()` and append it to `result` with `addLast()`. Once one list empties, drain any remaining flights from the non-empty list.

Mentor tip

Compare head elements using getFirst().compareWith(), remove the earlier flight with removeFirst(), addLast() to result, and drain residual nodes.

Our analysis

Where the marks are

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

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

Mentor insights

Traps to watch for

01
Confusing supervised vs unsupervised learning for disease diagnostics · Q8
Unsupervised learning groups imagery by visual similarities such as shadows or soil variations without clinical correlation. Supervised training uses verified ground truth to guide weight adjustments.
02
Ontologies vs folksonomies architectural roles · Q12
Ontologies provide formal, expert-designed hierarchical taxonomies with controlled vocabularies. Folksonomies are user-generated, flat, non-hierarchical tagging systems.
03
Recursive call stack printing order · Q16
In method process(), the recursive call occurs before the print statement, resulting in post-order traversal that outputs flights in reverse order.

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 2018
Marks: 260
Duration: 1 hour 20 min
Calculator: Not allowed
Questions: 16

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

Keep your copy of the paper open alongside. We don't reproduce the questions: each part below is labelled to match the November 2018 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

17 marks

Information systems vs databases, banking ACID properties, and law enforcement access ethics

Databases

(a)

3 marks

Distinguish between an information system and a database:
1) A database is a structured collection of persistently stored data organized according to a specific data model (such as relational tables).
2) An information system is a comprehensive socio-technical system comprising hardware, software, telecommunications, business workflows, people, and the database itself.
3) The database functions as the foundational data storage engine within the broader information system.

Mentor tip

Describe a database as the organized data repository and an information system as the broader infrastructure encompassing people, procedures, and applications.

(b)

8 marks

Apply ACID transaction principles to a bank transfer between Account A and Account B:
1) Atomicity (All or Nothing): The entire transfer (debiting Account A and crediting Account B) constitutes a single indivisible unit of work. If a system failure happens after debiting A but before crediting B, the transaction is rolled back and no funds are deducted.
2) Consistency: The database must transition from one valid state to another, preserving financial integrity invariants (e.g. Account A balance $\geq 0$ and total bank ledger balance $Balance_{A} + Balance_{B}$ remains unchanged).
3) Isolation: Concurrent transactions operating simultaneously cannot observe intermediate states. If another client queries Account A while the transfer executes, they observe either the pre-transfer or post-transfer balance, preventing race conditions.
4) Durability: Once the transaction commits, all balance adjustments are permanently written to persistent storage (and transaction redo logs), surviving subsequent server crashes or power interruptions.

Mentor tip

Dedicate two distinct marking points to each of the four ACID letters, explicitly contextualizing every principle with the bank transfer scenario.

(c)

6 marks

Discuss government and law enforcement access to private financial databases:
1) Arguments in favour of authorized access (Security & Crime Prevention):
- Investigating money laundering, terrorist financing, and serious organized tax evasion requires following financial transaction trails.
- Judicial oversight via search warrants ensures access is targeted, justified, and proportional.
2) Arguments against unrestricted access (Privacy & Civil Liberties):
- Unrestricted mass surveillance breaches client financial confidentiality and fundamental privacy rights.
- Creating government backdoors or shared access vectors introduces severe cybersecurity vulnerabilities, risking identity theft and data leaks.
3) Synthesis / Conclusion:
- Unrestricted interrogation must be prohibited; access should be granted strictly under independent judicial warrant for substantiated criminal inquiries.

Mentor tip

Provide a balanced discussion contrasting national security/crime detection against customer privacy, and conclude with the necessity of judicial warrants.

Question 2

8 marks

DBMS definition, logical schema characteristics, DDL, and DBA security responsibilities

Databases

(a)

1 mark

Define Database Management System (DBMS):
A software suite that provides an interface and tools to define, create, query, update, administer, and maintain databases.

Mentor tip

State that it is software managing the storage, retrieval, and administration of structured data.

(b)

3 marks

Identify three key characteristics of a logical schema:
1) Independent of physical storage: It defines logical structures without specifying disk sector allocations, block sizes, or file paths.
2) Formal relational specifications: Specifies entities, attributes, primary keys, foreign keys, and domain constraints.
3) Business integrity rules: Encapsulates entity relationships (1:1, 1:M, M:N) and referential integrity constraints.

Mentor tip

Highlight physical independence, specification of tables/relationships, and enforcement of integrity constraints.

(c)

2 marks

Outline the purpose of Data Definition Language (DDL):
1) DDL commands (such as `CREATE`, `ALTER`, `DROP`) define and modify the structural blueprint and metadata of database objects.
2) It establishes table definitions, data types, indexes, and primary/foreign key constraints within the DBMS data dictionary.

Mentor tip

Emphasize creating and modifying table structures and constraints rather than manipulating data rows.

(d)

2 marks

Identify two tasks performed by a DBA to safeguard database security:
1) Role-based access control management: Creating user accounts, configuring access permissions, and assigning principle-of-least-privilege credentials.
2) Routine backup and encryption configuration: Scheduling encrypted periodic database backups and establishing restoration protocols against ransomware and hardware failure.

Mentor tip

Name two specific administrative actions: managing user permissions and configuring backups/encryption.

Question 3

20 marks

Greengrocer relation querying, update anomalies, and 3NF normalisation

Databases

(a)(i)

1 mark

Identify a suitable data type for Prod_Price:
Decimal / Currency / Fixed-point numeric (e.g. `DECIMAL(6,2)` or `NUMERIC`) to maintain precise monetary cents without floating point rounding error.

Mentor tip

Specify Currency, Decimal, or Real/Float.

(a)(ii)

3 marks

Construct the logical query steps:
Step 1: Inspect the GREENGROCER relation filtering rows where `Supp_Name = 'Veggy Co.'`.
Step 2: Apply the range predicate `Prod_Price >= 4.00 AND Prod_Price <= 10.00`.
Step 3: Project the `Prod_Name` attribute from matching records (yielding Celery and Pears).

Mentor tip

List the supplier equality condition, the closed range price check, and the projected attribute.

(b)

2 marks

Outline redundant data in the GREENGROCER relation:
1) Supplier contact details (`Supp_Contact` and `Supp_Phone`) are repeated across multiple rows whenever a supplier provides more than one product.
2) For instance, 'Fruit and Veggie' with 'Ines Bond' and '542 9800' is duplicated across Apples (P211) and Onions (P121), creating unnecessary storage waste.

Mentor tip

Point out the duplication of supplier contact names and phone numbers across multiple product rows.

(c)(i)

2 marks

Insertion anomaly problem and example:
Problem: When adding a new product supplied by an existing supplier, the supplier's contact details must be manually re-entered, risking inconsistent spelling or mismatched telephone entries.
Example: Inserting Potatoes supplied by 'Veggy Co.' with contact 'Mia Abiss' risks mismatching existing records where 'Veggy Co.' contact is 'Mike Oberst' (235 7656).

Mentor tip

Explain the risk of inconsistent re-entry of supplier data when creating new product records.

(c)(ii)

2 marks

Deletion anomaly problem and example:
Problem: Deleting the sole product supplied by a particular vendor inadvertently deletes all record of that vendor's existence.
Example: Deleting product P265 (Tomatoes) deletes the only row containing supplier 'New Fruits' (John Smith, 677 7895), erasing their contact info from the database.

Mentor tip

Demonstrate that deleting the only product from a vendor permanently destroys all record of that vendor.

(c)(iii)

2 marks

Modification anomaly problem and example:
Problem: Updating a supplier attribute requires identifying and altering every corresponding row across the table; missing even one tuple causes data inconsistency.
Example: If 'Fruit and Veggie' changes telephone to '542 1811', updating row P211 without updating row P121 leaves the database with contradictory phone records for the same firm.

Mentor tip

Illustrate partial updates leaving conflicting phone records across multiple product rows.

(d)

8 marks

Decompose GREENGROCER into 3NF relations:
1) Analysis of functional dependencies:
- Prod_ID -> Prod_Name, Prod_Price, Supp_ID
- Supp_ID -> Supp_Name, Supp_Phone
- Supp_Phone -> Supp_Contact (or Contact associated directly with Supplier)
2) Normalised Relations:
PRODUCT (Prod_ID [PK], Prod_Name, Prod_Price, Supp_ID [FK])
SUPPLIER (Supp_ID [PK], Supp_Name, Supp_Phone [FK])
SUPPLIER_CONTACT (Supp_Phone [PK], Supp_Contact)
3) Verification of normal forms:
- 1NF: All attribute values are atomic; no repeating groups exist.
- 2NF: The primary keys are non-composite (single attributes), so partial functional dependencies cannot exist.
- 3NF: Transitive dependencies are eliminated by placing supplier contact details in separate relations referencing foreign keys.

Mentor tip

Define Product, Supplier, and SupplierContact with primary and foreign keys clearly denoted.

Question 4

20 marks

Ministry of Tourism data warehousing, ETL migration, privacy ethics, cluster and link analysis

Databases

(a)

4 marks

Identify four information sources for the tourism data warehouse:
1) Transportation providers: Airport flight manifests, seaport passenger records, and rail/bus transit ticketing systems.
2) Accommodation providers: Hotel, hostel, and holiday rental booking registries detailing stay durations and room choices.
3) Financial / Card transaction data: Anonymized merchant payment transaction feeds from retail, dining, and currency exchanges.
4) Tourism agency portals: Online visitor reviews, guided tour bookings, and national park visitor center passes.

Mentor tip

Name four distinct sources: transport gateways, hotels/lodging, commercial card transactions, and visitor booking portals.

(b)

4 marks

Outline two differences between a transactional database and a data warehouse:
Difference 1: Operational vs Analytical Purpose:
- An operational database supports real-time OLTP transactions (read/write updates for daily transactions).
- A data warehouse is optimized for OLAP analytics, executing read-only aggregated queries across massive historical datasets for strategic decision support.
Difference 2: Temporal Scope and Normalisation:
- Operational databases hold current, highly normalized snapshots to eliminate write anomalies.
- Data warehouses hold non-volatile, denormalized historical time-series spanning multiple years.

Mentor tip

Contrast OLTP (current, normalized, transactional) with OLAP (historical, denormalized, analytical).

(c)

3 marks

Explain how ETL processes resolve data migration challenges:
1) Extraction: Pulls raw heterogeneous data from incompatible legacy databases (SQL, spreadsheets, flat files) without interrupting live operations.
2) Transformation: Cleanses data by removing whitespace, resolving character encoding mismatches, standardizing date/currency formats, and deduplicating records.
3) Loading: Loads validated, uniform data records into targeted multidimensional star/snowflake schemas in the warehouse.

Mentor tip

Walk through Extract (pulling heterogeneous sources), Transform (cleansing, standardizing formats), and Load (inserting into warehouse).

(d)

2 marks

Outline one ethical dilemma resulting from tourism data mining:
Invasive behavioral profiling and loss of privacy: Correlating transit, hotel, and financial records creates intimate behavioral profiles tracking tourists' movements and lifestyle habits without their explicit informed consent.

Mentor tip

Focus on tracking individual movements and profiling habits without informed consent.

(e)

4 marks

Explain how cluster analysis enhances tourism marketing campaigns:
1) Segmenting tourist demographics: Unsupervised cluster algorithms group visitors into distinct cohorts based on shared attributes (e.g. expenditure patterns, origin country, family status).
2) Tailored campaign strategy:
- Cohort 1 ('Budget Eco-travelers'): Market affordable youth hostels, hiking trails, and eco-tours.
- Cohort 2 ('Luxury Cultural visitors'): Advertise high-end boutique hotels, historical museums, and fine dining.
3) Maximizing ROI: Customizing advertising to receptive audiences yields higher conversion rates and maximizes municipal tourism revenue.

Mentor tip

Define partitioning visitors into homogeneous clusters and give two specific tourism campaign examples.

(f)

3 marks

Explain the significance of link analysis in data mining:
1) Link analysis examines relationships and interconnected networks between disparate entities (e.g. tourists, hotels, flights, credit cards).
2) It uncovers hidden structural associations (e.g. discovering that tourists visiting Attraction A frequently patronize Restaurant B in another district).
3) In fraud and security detection, it identifies coordinated booking rings or irregular financial pathways across multiple databases.

Mentor tip

Explain analyzing connections between entities to discover hidden relationship networks and fraud patterns.

Option B: Modelling and simulation · Option

Questions 5–8 · 65 marks

Question 5

15 marks

Water utility billing model, rollover logic, and pseudocode

Modelling and simulation

(a)

3 marks

Specify data types and value ranges for model variables:
1) `METER_TYPE`: String or Enumerated type {'old', 'new'}.
2) `WATER_IN`: Integer in the range [0, 999999] litres.
3) `WATER_OUT`: Integer in the range [0, 999999] litres.
4) `INFRA_CHARGE`: Currency / Real number $\geq 0$ (default $204.80$ ).

Mentor tip

State both the programmatic data type and the allowable numeric or enumerated range.

(b)

2 marks

Explain how yearly water consumption is calculated accounting for meter rollover:
1) Normal condition: If $NewReading \geq OldReading$ , then $Volume = NewReading - OldReading$ .
2) Rollover condition: If $NewReading < OldReading$ (the 6-digit display rolled over past 999,999 to 0), then $Volume = (1000000 - OldReading) + NewReading$ .

Mentor tip

Describe standard subtraction when reading increases and handling rollover past 999,999.

(c)

7 marks

Construct pseudocode to calculate annual household water bills:
```text
input METER_TYPE, PREV_IN, CURR_IN
if CURR_IN >= PREV_IN then
WATER_SUPPLIED = CURR_IN - PREV_IN
else
WATER_SUPPLIED = (1000000 - PREV_IN) + CURR_IN
end if
WATER_CHARGE = WATER_SUPPLIED * 1.90
INFRA_CHARGE = 204.80
if METER_TYPE == 'new' then
input PREV_OUT, CURR_OUT
if CURR_OUT >= PREV_OUT then
WATER_RETURNED = CURR_OUT - PREV_OUT
else
WATER_RETURNED = (1000000 - PREV_OUT) + CURR_OUT
end if
if WATER_RETURNED >= 0.95 * WATER_SUPPLIED then
INFRA_CHARGE = INFRA_CHARGE * 0.90
end if
end if
TOTAL_BILL = WATER_CHARGE + INFRA_CHARGE
output TOTAL_BILL
```

Mentor tip

Ensure water charges ($1.90/L), meter rollover subtraction, and 10% discount when returned >= 95% are properly sequenced.

(d)

3 marks

Identify three actions customers can take to reduce infrastructure charges:
1) Upgrade equipment: Upgrade to a new smart meter capable of measuring returned waste water.
2) Maximize wastewater return: Ensure domestic wastewater drains directly into municipal sewage pipes rather than external garden runoff, exceeding the 95% return threshold.
3) Repair domestic leaks: Fix pipe leaks on private property to prevent water from escaping unmetered before returning to the sewage plant.

Mentor tip

Suggest upgrading to the new meter, minimizing unreturned garden usage, and eliminating household leaks.

Question 6

15 marks

Household water consumption model vs simulation and spreadsheet distribution

Modelling and simulation

(a)

4 marks

Distinguish between a model and a simulation using the water scenario:
1) A model is a conceptual and mathematical representation: Here, the model comprises the table of static activity values (15 L per dishwasher cycle, 70 L per bath) and rules governing weekly totals.
2) A simulation is the execution and dynamic exploration of the model over time: Here, running the simulation involves varying activity frequencies (e.g. 5 showers vs 3 baths per week) to evaluate projected weekly water usage and test scenarios.

Mentor tip

Contrast the static representation of mathematical equations/data with the dynamic execution and scenario testing.

(b)

4 marks

Explain how data accuracy affects simulation quality:
1) Garbage In, Garbage Out: The table provides average estimates that ignore appliance variances (e.g. high-efficiency vs old washing machines) and personal behavioral durations (e.g. a 4-minute shower vs a 15-minute shower).
2) Systematic error propagation: If baseline estimates are inaccurate, projected totals will deviate significantly from actual meter readings, reducing trust in the model for decision-making.

Mentor tip

Discuss appliance variations, variable human durations, and how flawed initial inputs compromise output reliability.

(c)

5 marks

Outline spreadsheet organization to maintain consumption under 2700 L/week:
1) Layout structure: Create columns for Activity Name, Baseline Litres Per Event, Weekly Frequency (variable input cells), and Total Litres ( $= Litres \times Frequency$ ).
2) Aggregation: A summary cell calculates `=SUM(Total Litres Range)`.
3) Conditional constraints: Incorporate conditional formatting (highlighting red if total $> 2700$ ) or an alert formula `=IF(Total <= 2700, "Target Met", "Over Target")`.
4) What-if experimentation: Family members adjust frequency parameters (e.g. reducing bath frequency) to bring the total below 2700 L.

Mentor tip

Specify input cells, product formulas, the SUM function, conditional alerts for > 2700 L, and parameter adjustments.

(d)

2 marks

Identify two reasons why the model fails for house-sharing young adults:
1) Independent, non-cooperative consumption: Roommates run separate partial laundry cycles and cook individually, generating far more frequent appliance usages than a cohesive family.
2) Longer and more varied personal habits: Individual shower durations and visitor water usage are substantially higher than conservative family averages.

Mentor tip

Point out independent non-consolidated appliance runs and uncoordinated usage habits.

Question 7

15 marks

Surgical simulation training, smart glove haptics, and 3D visualization

Modelling and simulation

(a)

4 marks

Outline one advantage and one disadvantage of surgical simulation software:
1) Advantage: Patient safety and risk-free repeatability. Trainees practice delicate operative techniques repeatedly under varied pathological conditions without risking patient morbidity or malpractice.
2) Disadvantage: Absence of full physical fidelity. Simulations cannot perfectly recreate tissue tactile resistance, unexpected arterial hemorrhaging, or the psychological stress of live surgery.

Mentor tip

Balance safe repetitive skill mastery with the lack of tactile fidelity and unpredictable biological complexity.

(b)

3 marks

Identify three technical requirements for smart surgery gloves:
1) Spatial tracking accuracy: High-resolution inertial/optical sensors providing six degrees of freedom (6-DoF) position and orientation tracking with sub-millimeter precision.
2) Haptic force feedback: Actuators providing variable resistance to simulate cutting or suturing different tissue densities.
3) Ultra-low latency communication: Real-time data transmission to the display engine ( $< 10$ ms) to prevent lag and visual-tactile desynchronization.

Mentor tip

Specify precise position/orientation tracking, tactile force feedback, and low-latency response.

(c)

2 marks

Define the term visualization:
The graphical representation of abstract data or spatial models into intuitive visual imagery (2D or 3D displays) to facilitate human cognitive comprehension and interactive analysis.

Mentor tip

Define converting structured data or physical models into visual graphical representations.

(d)

6 marks

Discuss whether upgrading to a 3D display improves surgical training:
1) Arguments in favour of 3D display:
- Depth perception: Stereoscopic 3D provides accurate depth cues and spatial awareness, essential for estimating incision depth and avoiding accidental puncture of underlying blood vessels.
- Realistic anatomical spatial relationships: Organs and vasculature are layered in 3D volume; viewing true perspective improves hand-eye coordination.
2) Counterarguments / Limitations:
- Hardware fatigue: Wearing 3D glasses or headsets during extended simulations can induce eye strain, motion sickness, and ergonomic fatigue.
- High computational expense: Rendering stereoscopic 3D requires high-end GPU hardware to maintain high frame rates without stuttering.
3) Conclusion: A 3D system substantially elevates anatomical realism and depth judgment, outweighing hardware overhead for surgical education.

Mentor tip

Examine depth perception and spatial accuracy vs visual fatigue and computing power, drawing a reasoned conclusion.

Question 8

20 marks

Genetic algorithm stages, neural network supervised learning, and voice response design

Modelling and simulation

(a)

6 marks

Outline the iterative stages of a Genetic Algorithm (GA):
1) Initial population generation: Initialize a diverse population of trial solutions encoded as genetic strings (chromosomes).
2) Fitness evaluation: Score each chromosome using a defined objective fitness function measuring solution quality.
3) Selection: Select pairs of chromosomes for reproduction using fitness-proportionate selection (e.g. roulette wheel or tournament selection).
4) Crossover (Recombination): Combine segments from selected parent chromosomes at specified crossover points to produce offspring.
5) Mutation: Apply low-probability random bit flips or perturbations to offspring genes to maintain genetic diversity and prevent premature convergence.
6) Replacement and Termination: Replace the previous generation with the new offspring population and repeat until convergence criteria or generational limits are reached.

Mentor tip

List the six sequential steps: initialize population, evaluate fitness, selection, crossover, mutation, and replacement/termination.

(b)(i)

4 marks

Outline why neural networks require training before aerial image analysis:
1) Random initial weights: At initialization, synaptic weights connecting network neurons have random values, producing arbitrary outputs.
2) Feature pattern abstraction: Training on aerial imagery allows the network to adjust weights iteratively, learning to recognize complex visual features (leaf discoloration, canopy thinning, necrotic textures).
3) Generalization: Proper training ensures the network generalizes to unobserved groves across diverse lighting and terrain without overfitting.

Mentor tip

Explain starting from random weights, learning visual patterns (discoloration/canopy loss), and achieving generalization.

(b)(ii)

6 marks

Explain why supervised learning is preferred over unsupervised learning for disease prediction:
1) Ground-truth classification: Supervised learning utilizes labelled ground-truth datasets where each aerial image is paired with verified botanical diagnostics ('infected' vs 'healthy').
2) Direct error minimization: The network calculates the precise classification error between its prediction and verified reality, backpropagating weight updates to minimize misdiagnoses.
3) Limitations of unsupervised learning: Unsupervised clustering groups images purely on visual similarities (e.g. shadows, soil color, tree density) without understanding which variations signify parasitic infection, leading to unaligned clusters.
4) Predictive precision: For disease progression forecasting, supervised regression/classification directly correlates canopy symptoms with disease severity.

Mentor tip

Contrast labelled ground truth and error minimization in supervised learning with unguided clustering based on irrelevant visual features in unsupervised learning.

(c)

4 marks

Explain why telephone voice response system prompts must be carefully curated:
1) Natural language ambiguity and speech recognition constraints: Prompts must present clear, closed-ended choices (e.g. 'Say billing or support') so acoustic models can accurately resolve vocabulary within a constrained lexicon.
2) User cognitive load and frustration mitigation: Overly lengthy or convoluted audio menus increase caller cognitive fatigue, leading to incorrect selections, abandoned calls, and customer dissatisfaction.

Mentor tip

Explain constraining the vocabulary to improve speech recognition accuracy and keeping prompts concise to prevent user frustration.

Option C: Web science · Option

Questions 9–12 · 65 marks

Question 9

16 marks

Online newspaper audio podcast publishing, compression, and client personalization

Web science

(a)

2 marks

Identify two actions relating to podcast provenance prior to publication:
1) Copyright and syndication verification: Secure explicit copyright licenses, syndication permissions, and broadcast rights from originating agencies.
2) Content validation and attribution: Verify factual accuracy and ensure appropriate intellectual property attribution is credited to original creators.

Mentor tip

Focus on licensing/copyright clearance and editorial/source attribution.

(b)

2 marks

Outline one technical reason for limiting available online podcasts to eight:
1) Server bandwidth and storage management: Limiting concurrent hosted audio streams prevents network saturation and eases server disk consumption.
2) Mobile UI rendering and page weight: Restricting list length keeps web page DOM size small, improving mobile download speed and preventing vertical scrolling clutter.

Mentor tip

Explain server bandwidth/storage control or lightweight client-side DOM loading on mobile devices.

(c)

1 mark

State one standard audio file format:
MP3 (or AAC, WAV, FLAC, OGG).

Mentor tip

State MP3, AAC, or WAV.

(d)

5 marks

Compare lossy vs lossless audio compression for streaming and downloading:
1) Lossy compression (e.g. MP3, AAC): Permanently removes psychoacoustically inaudible frequencies, achieving massive file size reductions ( $80 - 90%$ ). Spoken dialogue remains crystal clear, and reduced file size enables fast downloads and interruption-free streaming on mobile networks.
2) Lossless compression (e.g. FLAC, ALAC): Reconstructs original acoustic waveforms perfectly without data loss, but yields much larger files ( $50 - 60%$ of original size), consuming unnecessary bandwidth with no audible benefit for speech.
3) Evaluation: For 10-minute spoken interviews, lossy compression is optimal because transmission speed and buffer reliability outweigh minor acoustic fidelity differences.

Mentor tip

Contrast file size reduction and streaming reliability with acoustic fidelity, noting that spoken podcasts benefit most from lossy formats.

(e)

2 marks

Outline one technical cause of streaming audio buffering:
Network throughput fluctuation / Bandwidth bottleneck: If local Internet download speed drops below the media bit rate (due to Wi-Fi interference or ISP peak congestion), the playback buffer drains faster than it fills, halting audio until the buffer replenishes.

Mentor tip

Connect throughput dropping below stream bitrate to the client buffer running dry.

(f)

4 marks

Explain the technology enabling personalized podcast ordering:
1) Client tracking: Client interactions are recorded via persistent cookies, localStorage, or user login profile histories tracking preferences for science topics.
2) Dynamic server/client scripting: When loading the podcast section, an API request passes user profile metadata to the backend or executes client-side JavaScript.
3) Dynamic DOM reordering: Scripting sorts podcast elements, querying metadata tags (e.g. `genre = 'science'`) to hoist matching items to the top of the list before rendering.

Mentor tip

Explain tracking user category preference via cookies/history and using dynamic scripting to re-order the items.

Question 10

14 marks

E-commerce URL security, DNS resolution hierarchy, and PHP session cart management

Web science

(a)

2 marks

Describe how the HTTPS URL ensures communication security:
1) Protocol identifier: The `https://` scheme invokes Transport Layer Security (TLS/SSL).
2) Cryptographic protection: It guarantees symmetric encryption of all transmitted packet payloads and authenticates the BesTea server using X.509 certificates to thwart eavesdropping and spoofing.

Mentor tip

Highlight the https protocol, TLS encryption, and certificate-based server authentication.

(b)

4 marks

Identify the hierarchical steps in DNS resolution for `www.BesTea.com`:
1) Local resolver cache check: The browser and local OS check their DNS resolver caches; if missed, a query is dispatched to the recursive DNS server (ISP resolver).
2) Root nameserver inquiry: The recursive resolver contacts a Root DNS nameserver, which returns the IP address for the `.com` Top-Level Domain (TLD) nameserver.
3) TLD nameserver inquiry: The `.com` TLD server directs the resolver to the authoritative nameserver for `BesTea.com`.
4) Authoritative lookup and return: The authoritative nameserver looks up the A/AAAA record for `www.BesTea.com` and returns the destination IP address to the resolver, which caches it and provides it to the browser.

Mentor tip

Walk through: local cache -> recursive resolver -> Root server -> TLD server -> authoritative server -> browser.

(c)

2 marks

Outline how the footer is maintained uniformly across all web pages:
1) Server-side templating / includes: The footer HTML is maintained in a single separate component file (e.g. `footer.php`) incorporated into each page template via `include('footer.php');`.
2) Global CSS stylesheet: Global styling rules in a shared external CSS file format the footer consistently across the entire site.

Mentor tip

Mention server-side includes (e.g. PHP include/template partial) or a shared external CSS stylesheet.

(d)

4 marks

Explain the function of the provided PHP script fragment:
1) Session initialization: `$_SESSION['sessionUserID'] = new_random_number();` generates and stores a unique session identifier in the global session superglobal to track the visiting user.
2) Database query execution: Constructs and executes an SQL SELECT query (`SELECT * FROM users_db WHERE id = ...`) querying `users_db` for the record matching the session ID.
3) Record fetching: Fetches the resulting row into the `$userRecord` object/array, allowing the checkout workflow to populate user details.

Mentor tip

Explain generating a session identifier, querying the database with it, and fetching the record into an object.

(e)

2 marks

Identify user-provided data from the PHP code snippet:
1) Action command: The parameter `action` containing `'removeBasketItem'`.
2) Item identifier: The parameter `id` indicating the specific product to be deleted from the shopping basket.

Mentor tip

Name the action string ('removeBasketItem') and the product id passed in the request.

Question 11

15 marks

Charity rural connectivity challenges, cloud migration, and search engine optimization ethics

Web science

(a)

2 marks

Outline one technical challenge for hosting a website on local rural equipment:
Satellite latency and erratic power/bandwidth: Satellite connections suffer high latency and frequent weather disruptions, while aging local computers lack server redundancy and unmetered upstream bandwidth to handle inbound visitor traffic.

Mentor tip

Mention high satellite latency, limited uplink bandwidth, or power/hardware unreliability.

(b)

4 marks

Explain how cloud-based services overcome local hardware and connectivity constraints:
1) Offloaded hosting and global availability: Hosting the website on public cloud servers ensures high uptime, automated backups, and low latency globally regardless of local internet blackouts at Harvest's base.
2) Eliminating local hardware maintenance: Cloud SaaS/PaaS shifts the burden of server maintenance, security patching, and processing power away from obsolete local machines to enterprise data centers.

Mentor tip

Focus on high availability, external server maintenance, and freeing local machines from hosting burdens.

(c)

4 marks

Describe two problems arising from hosting data with an overseas MEDC cloud provider:
1) Data sovereignty and cross-border regulatory jurisdiction: Storing charity data in a foreign jurisdiction subjects files to foreign surveillance laws, potentially conflicting with local privacy protections.
2) Vendor lock-in and foreign currency subscription overhead: Ongoing cloud subscription costs priced in foreign currencies fluctuate with exchange rates, straining charity budgets, while migrating off proprietary platforms is technically arduous.

Mentor tip

Describe data sovereignty/legal jurisdiction issues and recurring foreign currency costs/vendor lock-in.

(d)

5 marks

Discuss whether Harvest should use black hat SEO techniques:
1) Potential short-term perceived benefit: Techniques like keyword stuffing, hidden text, and link farms might temporarily boost search rankings, attracting immediate donors.
2) Severe technical and ethical consequences:
- Search engine penalties: Algorithmic crawlers detect manipulative tactics and completely de-index the domain, eradicating web visibility.
- Reputational damage: A charitable non-profit relying on deceptive tricks destroys institutional trust with international donors.
3) Conclusion: Black hat SEO presents intolerable operational and reputational risks; Harvest must adopt sustainable white hat practices (high quality authentic content, semantic markup, and credible outreach).

Mentor tip

Contrast temporary ranking boosts against devastating algorithmic bans and loss of donor trust; conclude strongly against black hat.

Question 12

20 marks

Tapetum lucidum semantic search, ontologies vs folksonomies, and collective intelligence

Web science

(a)

4 marks

Explain why a semantic multimedia search is preferable to a keyword search:
1) Acoustic / Visual query input: The student may possess an audio recording of a nocturnal animal's call or a photo of eyes glowing in the dark without knowing the species name.
2) Semantic web conceptual linking: Instead of matching the string 'tapetum lucidum', semantic web engines query multimedia ontologies linking visual eye-shine and acoustic calls to nocturnal biological taxonomies, identifying the animal directly.

Mentor tip

Contrast keyword matching with querying semantic multimedia ontologies that link recorded sounds or eye-shine images to species taxonomies.

(b)

3 marks

Identify three distinctions between an ontology and a folksonomy:
1) Authorship: Ontologies are authored by domain knowledge engineers; folksonomies are crowdsourced informally by web users.
2) Vocabulary control: Ontologies enforce strict, controlled vocabularies and formal hierarchical logic (taxonomies); folksonomies are uncontrolled, free-form tags.
3) Semantic expressivity: Ontologies express precise relationships (e.g. `subClassOf`, `partOf`); folksonomies consist of flat, ambiguous associative tags.

Mentor tip

Highlight expert formal design vs crowdsourced tags, controlled vs uncontrolled vocabulary, and formal taxonomy vs flat associations.

(c)

6 marks

Discuss integrating folksonomy expressivity with ontology authoritativeness:
1) Benefits of integration:
- Folksonomies capture contemporary linguistic usage, emerging trends, and diverse user viewpoints that rigid ontologies miss.
- Ontologies provide structural disambiguation, mapping ambiguous tags (e.g. 'cat's eye' marble vs feline eye) to formal semantic concepts.
2) Integration mechanisms and challenges:
- Social tagging clustering: Applying NLP and clustering to social tags extracts potential terms to expand and enrich ontologies dynamically.
- Noise and vandalism risk: Uncurated user tags introduce misspellings, spam, and subjective labels, necessitating automated confidence thresholds and human expert moderation.
3) Conclusion: Hybrid systems bridging folksonomic tags into ontology concepts via weighted semantic mappings achieve both adaptability and rigor.

Mentor tip

Discuss using folksonomies to surface emerging vocabulary while using ontologies to resolve ambiguity, noting noise/spam moderation challenges.

(d)

4 marks

Explain the role of web crawlers in building search engine indexes:
1) Continuous discovery: Crawlers systematically follow hyperlinks starting from seed URLs to discover newly published and modified web documents.
2) Content harvesting: Crawlers fetch raw HTML content, image metadata, and text documents, caching them locally in document repositories.
3) Indexing pipeline feed: Downloaded pages are parsed, tokenized, and passed to indexing algorithms that construct and update inverted index posting lists.

Mentor tip

Describe traversing links from seeds, retrieving document content, and feeding raw text into the indexing pipeline.

(e)

3 marks

Explain how collective intelligence fosters successful open-source software development:
1) Decentralized peer contribution: Thousands of independent developers inspect code, report edge-case bugs, and submit pull requests across global time zones.
2) Rapid code refinement: Linus's Law ('Given enough eyeballs, all bugs are shallow') ensures security flaws and performance bottlenecks are audited and resolved rapidly.
3) Meritocratic governance: Projects like Linux utilize rigorous peer review and automated testing frameworks, integrating the best global engineering ideas into a stable public good.

Mentor tip

Explain distributed global peer code review, Linus's Law for rapid bug remediation, and meritocratic pull request integration.

Option D: Object-oriented programming · Option

Questions 13–16 · 65 marks

Question 13

21 marks

Airport Flight and Arrival classes, OOP principles, UML diagram, and linear search

Object-oriented programming

(a)(i)

2 marks

Outline the nature of an object in OOP:
1) An object is an instantiated runtime instance of a class that encapsulates both state (data held in instance variables) and behavior (functions/methods).
2) It acts as an autonomous software entity interacting with other objects through message passing (method calls).

Mentor tip

Mention that an object encapsulates state (attributes) and behavior (methods) as an instance of a class.

(a)(ii)

4 marks

Describe two disadvantages of Object-Oriented Programming:
1) Memory and runtime execution overhead: Encapsulation, dynamic method dispatch, and object reference wrappers require more memory and processing cycles than procedural code.
2) Steep learning curve and architectural complexity: Designing class hierarchies, inheritance, and interfaces requires substantial upfront planning; improper design leads to bloated, overly complex codebases.

Mentor tip

Cite memory/processing overhead from dynamic dispatch and complex architectural design requirements.

(a)(iii)

2 marks

Outline one advantage of modularity in software development:
Reusability and maintainability: Independent, self-contained modules can be tested, debugged, and maintained in isolation without side effects on the rest of the codebase, and can be reused across future projects.

Mentor tip

Focus on isolated testing, easier debugging, and reusable modules.

(b)

1 mark

State the relationship between Flight and Arrival:
Aggregation (or Association): An `Arrival` object 'has-a' `Flight` object as an instance variable (`myFlight`).

Mentor tip

State Aggregation (or 'has-a' association).

(c)

4 marks

Construct the UML diagram for the Arrival class:
+---------------------------------------------------+
| Arrival |
+---------------------------------------------------+
| - myFlight : Flight |
| - sta : String |
| - runway : int |
| - gate : String |
| - delay : int |
| - landed : boolean |
+---------------------------------------------------+
| + Arrival(myFlight : Flight, sta : String) |
| + addDelay(newDelay : int) : void |
| + getETA() : String |
| + compareWith(flightID : String) : int |
| + compareWith(anotherArrival : Arrival) : int |
+---------------------------------------------------+

Mentor tip

Draw the 3 compartments: class name, private attributes with types, and public methods with parameters and return types.

(d)

4 marks

Construct the `search(String flightID)` linear search method:
```java
private int search(String flightID)
{
for (int i = 0; i <= last; i++)
{
if (inbound[i].compareWith(flightID) == 0)
{
return i;
}
}
return -1;
}
```
Explanation: Loop from index 0 up to `last`. For each arrival, call `compareWith(flightID)`. When it returns 0, the target is located; return the index `i`.

Mentor tip

Iterate from 0 to last, call compareWith(flightID) == 0, and return the matching index.

(e)(i)

2 marks

Outline one advantage of binary search over linear search:
Logarithmic time efficiency: Binary search runs in $O (lo g n)$ time by repeatedly halving the search space, performing far fewer comparisons than linear search ( $O (n)$ ) on large sorted datasets.

Mentor tip

State $O (lo g n)$ logarithmic speed and dividing the search space in half.

(e)(ii)

2 marks

Outline one disadvantage of binary search:
Mandatory sorted order requirement: Binary search requires the array to remain strictly sorted by the search key; searching on an unsorted attribute (or maintaining sorting during frequent insertions) incurs significant overhead.

Mentor tip

State the strict requirement that elements must already be sorted by the search attribute.

Question 14

8 marks

Data hiding security, method overloading, and arrival ETA comparison

Object-oriented programming

(a)

2 marks

Outline how data hiding acts as a security feature in OOP:
1) Declaring instance variables `private` encapsulates internal state and prevents direct external tampering by client classes.
2) State mutations can only occur via controlled public accessor/mutator methods that enforce business validation rules.

Mentor tip

Explain private encapsulation preventing arbitrary external tampering, and enforcing validation through methods.

(b)(i)

2 marks

Outline why two `compareWith` methods do not cause a conflict:
1) Method overloading allows multiple methods to share the same name provided their parameter signatures differ.
2) The compiler differentiates them by their arguments: `compareWith(String)` takes a `String`, whereas `compareWith(Arrival)` takes an `Arrival` object.

Mentor tip

Explain that distinct parameter types allow the compiler to distinguish between the two calls.

(b)(ii)

1 mark

State the name of this OOP property:
Method overloading (or compile-time / static polymorphism).

Mentor tip

State method overloading or polymorphism.

(c)

3 marks

Construct `compareWith(Arrival anotherArrival)`:
```java
public int compareWith(Arrival anotherArrival)
{
String thisETA = this.getETA();
String otherETA = anotherArrival.getETA();
return thisETA.compareTo(otherETA);
}
```
Explanation: Retrieve the estimated arrival time strings for both objects using `getETA()`, invoke `compareTo()`, and return the integer result directly.

Mentor tip

Obtain getETA() from both this and anotherArrival, then return thisETA.compareTo(otherETA).

Question 15

16 marks

Method signature definition, delayed flight display, and array insertion

Object-oriented programming

(a)

2 marks

Define method signature:
The method name combined with its ordered parameter list (the number, order, and data types of its formal parameters). Return types and access modifiers are not part of the signature in Java.

Mentor tip

State that a signature consists of the method identifier and the parameter types/order.

(b)

4 marks

Construct `showDelayed(String t)`:
```java
public void showDelayed(String t)
{
for (int i = 0; i <= last; i++)
{
Arrival arr = inbound[i];
if (arr.getDelay() > 0 && !arr.isLanded())
{
if (arr.getETA().compareTo(t) < 0)
{
System.out.println(arr.getMyFlight().getId());
}
}
}
}
```
Explanation: Loop from 0 to `last`. For each flight, verify that delay $> 0$ , `landed == false`, and `getETA().compareTo(t) < 0`. If all match, output the flight ID.

Mentor tip

Check three conditions: delay > 0, not landed, and ETA lexicographically before t.

(c)

6 marks

Construct `add(Arrival newArrival)` maintaining sorted order by ETA:
```java
public void add(Arrival newArrival)
{
int i = last;
while (i >= 0 && inbound[i].compareWith(newArrival) > 0)
{
inbound[i + 1] = inbound[i]; // shift right
i--;
}
inbound[i + 1] = newArrival; // insert at sorted position
last++;
}
```
Explanation: Implement standard insertion sort insertion: start at index `last`, shift all elements with larger ETA one position to the right, insert `newArrival` into the vacated slot, and increment `last`.

Mentor tip

Shift elements right while greater than newArrival using compareWith, place newArrival, and increment last.

(d)

4 marks

Describe how to update flight delay and maintain sorted order using provided methods:
1) Step 1: Call `remove(flightID)` to extract and remove the target `Arrival` object from the array; `remove` shifts subsequent elements left and decrements `last`.
2) Step 2: Invoke `addDelay(newDelay)` on the removed `Arrival` object to update its delay and recalculate its ETA.
3) Step 3: Call `add(arrivalObject)` passing the updated arrival object; `add` inserts it into its newly sorted position according to its revised ETA.

Mentor tip

Clearly describe the 3-step sequence: remove the flight using remove(), update its delay using addDelay(), and re-insert it using add().

Question 16

20 marks

Dual-runway airport LinkedList operations, recursion tracing, and list merging algorithm

Object-oriented programming

(a)

2 marks

Outline one advantage of utilizing standard software libraries:
Reliability and development productivity: Library collections (e.g. `java.util.LinkedList`) are pre-tested, debugged, and optimized by systems engineers, accelerating development and preventing algorithmic errors.

Mentor tip

State pre-tested reliability, performance optimization, and saving development time.

(b)

4 marks

Evaluate dynamic linked lists vs static arrays for managing airport arrivals:
1) Dynamic sizing vs Fixed allocation:
- A static array has a fixed capacity ( $200$ ); an unexpected surge in incoming flights causes overflow crashes, while low traffic wastes memory.
- A linked list allocates nodes dynamically on demand, scaling gracefully without artificial size limits.
2) Insertion / Deletion efficiency:
- Inserting an arrival into a sorted array requires shifting on average $n /2$ elements ( $O (n)$ memory writes).
- In a linked list, inserting or reordering nodes simply updates pointer references without physical array copying.

Mentor tip

Compare fixed vs dynamic memory sizing and shifting elements in arrays vs updating pointers in linked lists.

(c)

2 marks

Determine the position and sketch the linked list after inserting RO225:
Existing Runway 1 list sorted by ETA:
1) KL1102 (STA 11:05, delay 0 -> ETA 11:05)
2) MH9077 (STA 11:45, delay 45 -> ETA 12:30)
3) LH8803 (STA 12:55, delay 0 -> ETA 12:55)
New flight to insert: RO225 (STA 12:05, delay 0 -> ETA 12:05).
Comparing ETAs: 11:05 < 12:05 < 12:30.
RO225 is inserted between KL1102 and MH9077:
[Head: KL1102 (11:05)] -> [RO225 (12:05)] -> [MH9077 (12:30)] -> [LH8803 (12:55)] -> null

Mentor tip

Calculate ETA 12:05 for RO225 and place it between KL1102 (11:05) and MH9077 (12:30).

(d)(i)

1 mark

Define recursion:
A programming technique in which a method calls itself directly or indirectly to solve smaller sub-problems until reaching a base terminating condition.

Mentor tip

Define a method calling itself with a base condition.

(d)(ii)

4 marks

Trace the execution of `process(runway1, 0)`:
Method code:
```java
void process(LinkedList<Arrival> myList, int i) {
if (i < myList.size()) {
process(myList, i + 1);
System.out.println(myList.get(i).getETA());
}
}
```
List elements: index 0 = KL1102 (ETA 11:05), index 1 = MH9077 (ETA 12:30), index 2 = LH8803 (ETA 12:55). `size() == 3`.
Tracing call stack progression:
`i = 0`: calls `process(runway1, 1)`
`i = 1`: calls `process(runway1, 2)`
`i = 2`: calls `process(runway1, 3)`
`i = 3`: condition `3 < 3` is false; returns.
Unwinding call frames (post-order printing):
Frame `i = 2`: prints `myList.get(2).getETA()` -> 12:55
Frame `i = 1`: prints `myList.get(1).getETA()` -> 12:30
Frame `i = 0`: prints `myList.get(0).getETA()` -> 11:05
Construct trace table:
i | id | output
0 | KL1102 | -
1 | MH9077 | -
2 | LH8803 | -
3 | - | -
2 | LH8803 | 12:55
1 | MH9077 | 12:30
0 | KL1102 | 11:05

Mentor tip

Notice that printing occurs after the recursive call, outputting ETAs in reverse order: 12:55, 12:30, 11:05.

(e)

7 marks

Construct `mergeLists()` to merge two sorted LinkedLists:
```java
public LinkedList<Arrival> mergeLists()
{
LinkedList<Arrival> result = new LinkedList<Arrival>();
while (!runway1.isEmpty() && !runway2.isEmpty())
{
if (runway1.getFirst().compareWith(runway2.getFirst()) <= 0)
{
result.addLast(runway1.removeFirst());
}
else
{
result.addLast(runway2.removeFirst());
}
}
while (!runway1.isEmpty())
{
result.addLast(runway1.removeFirst());
}
while (!runway2.isEmpty())
{
result.addLast(runway2.removeFirst());
}
return result;
}
```
Algorithm: Continuously compare head elements of both lists using `compareWith`. Remove the earlier flight with `removeFirst()` and append it to `result` with `addLast()`. Once one list empties, drain any remaining flights from the non-empty list.

Mentor tip

Compare head elements using getFirst().compareWith(), remove the earlier flight with removeFirst(), addLast() to result, and drain residual nodes.

Our analysis

Where the marks are

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

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

Mentor insights

Traps to watch for

01
Confusing supervised vs unsupervised learning for disease diagnostics · Q8
Unsupervised learning groups imagery by visual similarities such as shadows or soil variations without clinical correlation. Supervised training uses verified ground truth to guide weight adjustments.
02
Ontologies vs folksonomies architectural roles · Q12
Ontologies provide formal, expert-designed hierarchical taxonomies with controlled vocabularies. Folksonomies are user-generated, flat, non-hierarchical tagging systems.
03
Recursive call stack printing order · Q16
In method process(), the recursive call occurs before the print statement, resulting in post-order traversal that outputs flights in reverse order.

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