2023 CCEA A-Level Digital Technology 2650 Practice Paper with Answers

A server-based LAN needs several resources beyond individual network cards. A server is needed to store and manage the shared patient records centrally so all PCs and the laptop can access the same data. Because one device (the laptop) connects wirelessly, a wireless access point is needed to let it join the wired LAN. Checking against the full resource list for this scenario (server, switched hub, wireless access point, and possibly a repeater or media converter if cabling runs are long or mixed media are used) confirms that server and wireless access point are the two most clearly required resources given the description. Answer: server; wireless access point.

[1] mark for each correctly identified resource, up to a maximum of [2]. Accept any two from: server; wireless access point; switched hub; repeater; media converter. Do not accept network card (excluded by the question) or software-only answers.

Network scope is classified by the geographical area covered. A network confined to one building or site with a single owning organisation and short cable runs (up to about 1 km) is, by definition, a local area network (LAN); the dental practice's in-house network from part (a) is a direct example. A network that spans a city, linking multiple sites of one organisation over leased lines, is a metropolitan area network (MAN); a hospital trust connecting its various clinics across Belfast fits this scope exactly, since it is larger than a single-site LAN but smaller than a national/international WAN. Cross-checking against the three defined scopes (LAN/MAN/WAN), the 1 km single-site description can only be a LAN and the city-wide multi-site description can only be a MAN, confirming both answers. Answer: Row 1 — LAN, e.g. the dental practice's own office network; Row 2 — MAN, e.g. a hospital trust linking its clinics across Belfast.

[1] mark for 'LAN'; [1] mark for a valid LAN example (single site/organisation); [1] mark for 'MAN'; [1] mark for a valid MAN example (city-wide, same organisation, multiple sites). Total [4]. Accept any sensible real-world example matching the scope described.

Each device serves a distinct role in extending or organising the network. (i) The switched hub is the central wired connection point: unlike a simple (unswitched) hub which broadcasts every incoming frame to all ports, a switch reads the destination MAC address of each frame and forwards it only to the port for that specific device, so it purposefully reduces collisions and unnecessary traffic on the LAN. (ii) The wireless access point bridges the wired and wireless parts of the network: it receives radio-frequency signals from the wireless laptop, converts them into the electrical/optical signals used on the wired LAN (and vice versa), which is exactly what allows the one wireless device in the scenario to reach the shared server. (iii) The repeater's purpose is signal regeneration: as a signal travels along a cable or through the air it loses strength (attenuates), and a repeater amplifies and retransmits a clean copy of the signal so it can continue over a longer distance without becoming corrupted. Checking each description against its device name confirms a one-to-one match with no overlap: switch = selective forwarding, access point = wired/wireless bridging, repeater = signal regeneration. Answer: (i) selectively forwards data to the correct wired device using its MAC address; (ii) bridges wireless devices to the wired LAN; (iii) regenerates a weakening signal to extend its range.

(i) [3]: [1] connects/forwards data between wired devices; [1] reference to using the destination address/MAC address to send only to the intended port; [1] reduces collisions/unnecessary traffic — accept comparison with a simple hub. (ii) [3]: [1] allows wireless devices to join the network; [1] converts between wireless (radio) signals and wired signals; [1] reference to acting as a bridge/link between wireless and wired segments. (iii) [3]: [1] regenerates/amplifies/boosts a signal; [1] reference to signal loss/attenuation over distance; [1] allows data to travel further without loss of quality/corruption. Total [9].

A ring topology is defined by how devices are physically and logically connected. First, every device connects to exactly two others, its immediate neighbours, so the cabling forms a closed loop with no free ends; data (often carried by a token) passes from device to device around the loop until it arrives at the addressed device. Second, since there is no central switch or hub controlling traffic, each device in the ring itself regenerates the signal as it forwards it to the next device, which helps maintain signal strength over the whole ring; however, this also means the ring depends on every device/cable segment functioning, so a single break in a simple (single) ring can take down the entire network, unlike a star topology where one faulty cable only affects one device. Re-reading the two features confirms they are genuinely distinct (physical/logical connection pattern vs signal-regeneration-and-fault-sensitivity) and both correctly describe a ring rather than a bus or star. Answer: (1) each device connects to two neighbours in a closed loop with data passing round the ring to its destination; (2) each device regenerates the signal as it forwards data, but a single break can disrupt the whole ring.

[2] marks per feature, to a maximum of [4]: [1] for stating a valid feature, [1] for a developed/explained detail. Accept any two of: closed-loop connection with each device linked to two neighbours; unidirectional (or dual-ring bidirectional) data flow around the ring; each device regenerates the signal; single point of cable/device failure can break the whole ring (unless dual-ring); no central device/hub is required; performance can degrade as more devices are added since data may pass through many devices.

In a star topology, every device (including the server) connects by its own individual cable to a central switch, whereas in a bus topology all devices share a single central cable (the backbone) with terminators at each end. Considering reliability: in the star, if one cable or one PC fails, only that single device loses connection while the rest of the practice, including the shared patient-records server, keeps working normally; in the bus, a break anywhere on the shared cable, or a missing/faulty terminator, brings down the entire network because every device relies on that one cable, which would be highly disruptive for a dental practice needing continuous access to patient records. Considering cost: the bus topology is cheaper to install because it needs less cable overall and no central switching hardware, whereas the star topology needs one length of cable per device plus a switch, so it is more expensive to set up and slightly more expensive to extend. Weighing these against each other for this scenario, the reliability of continuous access to a shared patient-records server is more important for a healthcare setting than the modest extra cabling cost of a star, and the star also makes it much easier to add a new surgery PC later by connecting one more cable to the switch, without disturbing the existing network, unlike a bus that would need re-terminating. On balance, the star topology is the more suitable choice, since its reliability and scalability benefits are judged to outweigh its higher cost for a small practice that cannot afford network downtime. Answer: the star topology is more suitable overall, since its much greater fault tolerance (one cable/device failure affects only one PC, not the whole network) and easier expansion outweigh its higher cabling/hardware cost, which matters less for a small practice.

Level-of-response marking (3 bands), maximum [6]. Band 2 [5-6 marks]: detailed, accurate description of both topologies with balanced comparison covering reliability AND cost, a clear justified conclusion recommending one topology for the scenario, fluent use of DT terminology (e.g. central switch, backbone cable, terminator, single point of failure), high standard of spelling, punctuation and grammar. Band 1 [3-4 marks]: competent description of both topologies, some comparison of reliability or cost (not necessarily both in depth), a conclusion given but with limited justification, some relevant DT terminology, reasonable SPG. Band 0 [1-2 marks]: brief or one-sided description (e.g. only one topology described), little or no comparison, no clear conclusion, limited terminology, meaning not always clear. 0 marks: no creditable response. Answer above corresponds to a Band 2 response.

Each statement describes the defining purpose of one protocol from the list. Statement (i) describes collision detection on a shared medium, which is precisely the role of Carrier Sense Multiple Access with Collision Detection (CSMA/CD): devices 'listen' before transmitting and detect if a collision occurs, so the answer is CSMA/CD. Statement (ii) describes packaging, addressing, routing and reassembling data for transport across networks such as the internet, which is the core function of the Transmission Control Protocol/Internet Protocol (TCP/IP) suite, so the answer is TCP/IP. Statement (iii) describes framing data and MAC addressing for wired LANs, which is the defining role of Ethernet, so the answer is Ethernet. Checking that no protocol is used twice and every statement is matched to exactly one protocol from the given list confirms the mapping is consistent and complete. Answer: (i) CSMA/CD; (ii) TCP/IP; (iii) Ethernet.

[1] mark for each correct gap, to a maximum of [3]: (i) CSMA/CD; (ii) TCP/IP; (iii) Ethernet. No partial credit for an incorrect but related term (e.g. 'CSMA' alone without CD).

The OSI model organises network communication into seven layers, each with a defined function, stacked from the physical transmission of bits up to the software the user interacts with. Starting at the bottom: Layer 1 is the Physical layer (transmits raw bits over the physical medium), Layer 2 is the Data Link layer (organises bits into frames and handles MAC addressing), Layer 3 is the Network layer (handles logical addressing and routing), Layer 4 is the Transport layer (provides end-to-end delivery, flow control and error recovery), Layer 5 is the Session layer (establishes, manages and terminates sessions), Layer 6 is the Presentation layer (formats/translates/encrypts data for the application), and Layer 7 is the Application layer (provides the interface to network-aware software). Counting the list confirms exactly seven layers with no repeats and no omissions, matching the standard OSI model. Answer: 1 Physical; 2 Data Link; 3 Network; 4 Transport; 5 Session; 6 Presentation; 7 Application.

[1] mark for the seven layer names substantially correct (any order); [1] additional mark for all seven correctly ordered from Physical (1) to Application (7). Total [2]. Common mnemonic accepted as working: 'Please Do Not Throw Sausage Pizza Away'.

Each function maps to one specific OSI layer by definition. Routing data between different networks using logical addressing (IP addresses) is the defining job of the Network layer (Layer 3), which is responsible for determining the best path for data across interconnected networks. Providing end-to-end flow control and error recovery between the original sender and final receiver, regardless of the route taken, is the defining job of the Transport layer (Layer 4), which manages segmentation, sequencing and reliable delivery. Cross-checking against the full layer list from the previous part confirms Network = Layer 3 and Transport = Layer 4, and that neither function belongs to a layer below (Data Link, Physical, which are link-local, not end-to-end) or above (Session, Presentation, Application, which do not handle routing or flow control). Answer: (i) Network layer; (ii) Transport layer.

[1] mark for 'Network layer' (i); [1] mark for 'Transport layer' (ii). Total [2]. Accept 'Layer 3' / 'Layer 4' as equivalent to the layer names.

(i) The Data Link layer sits directly above the Physical layer and is responsible for turning the raw stream of bits the Physical layer transmits into structured frames; it adds header information including the source and destination MAC addresses so that devices on the same local network segment can identify each other, and it typically includes a check (such as a frame check sequence/CRC) to detect transmission errors on that single link. (ii) The Application layer is the topmost layer and does not do the actual network transmission itself; instead it is the layer that network-aware software (such as a web browser, email client, or file-transfer utility) uses to access network services, presenting data to and receiving data from the user via protocols appropriate to that application. Checking both descriptions against their position in the seven-layer stack from the earlier part confirms Data Link (Layer 2) is concerned with local framing/addressing/error-checking, clearly distinct from Application (Layer 7), which is concerned with providing the user-facing network interface — the two descriptions do not overlap in function. Answer: (i) organises bits into frames, adds MAC addressing and performs link-level error detection; (ii) provides the interface between user-facing network software and the network services it uses.

(i) [4]: [1] organises/structures bits into frames; [1] reference to MAC addressing / physical addressing of devices on the local network; [1] reference to error detection/checking (e.g. frame check sequence, CRC) on the link; [1] correctly identifies this operates over a single physical link/local segment. (ii) [4]: [1] identifies this as the layer closest to the user/topmost layer; [1] provides the interface between network services and application software; [1] valid named example of application-layer software/protocol (e.g. web browser/HTTP, email client/SMTP); [1] reference to enabling the application to send/receive data over the network. Total [8].

A layered model breaks the complex task of network communication into smaller, well-defined stages, and this structure brings real design benefits. First, because each layer only needs to know how to communicate with the layer directly above and below it (through a standard interface), a manufacturer can change the internal workings of one layer, for example, upgrading Physical-layer hardware from copper cable to fibre optic, without needing to redesign the layers above it; this modularity makes development, testing and maintenance simpler. Second, because the layers and their interfaces are standardised (as in the OSI model), equipment and software produced by different manufacturers, following the same standard, can be relied upon to interoperate correctly, which is why, for example, a laptop from one manufacturer can connect to a switch and router from entirely different manufacturers. Re-reading both explanations confirms they describe genuinely different benefits (ease of independent development/modularity vs cross-vendor interoperability from standardisation) rather than restating the same point twice. Answer: (1) layers can be developed/updated independently since only the interface to adjacent layers must stay fixed (modularity); (2) standardised layers allow equipment/software from different manufacturers to interoperate.

[2] marks per advantage, to a maximum of [4]: [1] for stating a valid advantage, [1] for a developed explanation of why it is an advantage. Accept any two of: modularity/independent development of layers; easier troubleshooting/isolating faults to a specific layer; interoperability between different manufacturers' hardware/software; standardisation aids teaching/documentation/design; simplifies complex network communication into manageable stages.

(i) VoIP works by digitising the caller's analogue voice signal, compressing it, and splitting it into data packets which are then transmitted across an IP-based network (the internet) using the normal internet protocols for addressing and routing, rather than using a dedicated circuit-switched telephone line; at the receiving end, the packets are reassembled and converted back into an analogue sound signal, allowing a phone call to be carried as ordinary internet traffic. (ii) RFID works by attaching a small tag, containing a microchip that stores an identifying code plus a small antenna, to the object being tracked; when this tag comes within range of an RFID reader, the reader emits a radio signal that powers/activates the tag (for passive tags), and the tag responds by transmitting its stored data back to the reader by radio waves, which the reader then processes — critically, this requires no physical contact or direct line-of-sight, unlike a barcode scanner. Checking both descriptions against their protocol names confirms VoIP concerns carrying voice calls over data networks, and RFID concerns wireless identification/tracking of tagged objects — two clearly distinct technologies with no overlap. Answer: (i) VoIP digitises and packetises voice for transmission over an IP network, reassembling it back to sound at the receiver; (ii) RFID uses a tag and reader communicating by radio waves to wirelessly identify/track a tagged object without contact or line-of-sight.

(i) [4]: [1] voice/sound is converted to digital data/packets; [1] transmitted over an IP/internet-based network rather than a traditional phone line; [1] reference to compression and/or packet routing; [1] reassembled/converted back to audio at the receiving end. (ii) [4]: [1] reference to a tag containing a microchip/antenna attached to an object; [1] reference to an RFID reader that detects/reads the tag; [1] communication is by radio waves; [1] reference to no line-of-sight/contact required (or reference to passive/active tags). Total [8].

(i) Bandwidth measures the capacity of a transmission medium: it is the maximum volume of data that can theoretically pass through the connection per unit of time, conventionally expressed in bits per second and its multiples (kbps, Mbps, Gbps) — a higher bandwidth means more data can be transferred in the same time. (ii) Broadband describes a class of internet connection technology characterised by high bandwidth and the ability to carry many signals/frequencies simultaneously over one physical medium, which is what allows a broadband connection to support activities like video streaming alongside other traffic at the same time, in clear contrast to older narrowband (dial-up) connections that were far slower and carried only a single signal. Re-reading both definitions confirms 'bandwidth' names the measurable capacity/rate, while 'broadband' names the technology/connection type that provides high bandwidth — these are related but distinct concepts, matching what the question asks for. Answer: (i) bandwidth = the maximum data transfer rate of a medium, measured in bits per second; (ii) broadband = a high-bandwidth connection type able to carry multiple signals at once, much faster than narrowband/dial-up.

(i) [2]: [1] reference to maximum amount/rate of data transferred; [1] reference to a time period and/or correct unit (bits per second/Mbps). (ii) [2]: [1] reference to high-bandwidth/high-speed connection; [1] reference to carrying multiple signals/channels at once or comparison with narrowband/dial-up. Total [4].

Fibre optic cable is a physical transmission medium consisting of a thin glass or plastic core surrounded by a cladding layer with a different refractive index, and (unlike copper cable) it carries information as light rather than electricity. To transmit data, a light source at the sending end (typically a laser diode or LED) converts the electrical data signal into pulses of light, switching rapidly to represent the binary data. As this light travels along the fibre core, it strikes the boundary with the cladding at an angle that causes total internal reflection, meaning the light bounces back into the core rather than escaping, so it continues travelling along the length of the fibre with very little energy lost, even over long distances. At the receiving end, a photodetector converts the incoming light pulses back into an electrical signal that the receiving device can process. Checking this description covers both what the medium physically is (glass/plastic strands with core and cladding) and how data actually moves along it (light pulses via total internal reflection) confirms it fully answers the two-part command 'describe... including how data is transmitted'. Answer: fibre optic cable transmits data as pulses of light generated by a laser/LED, which travel along the glass/plastic core by total internal reflection off the cladding with very little signal loss, and are converted back to an electrical signal by a receiver.

[1] reference to the physical structure (thin strands of glass/plastic, core and cladding); [1] data is transmitted as pulses of light, not electrical signals; [1] reference to a light source (laser/LED) generating the pulses; [1] reference to total internal reflection keeping the light within the core along the cable; [1] reference to a receiver/photodetector converting light back into an electrical signal, or reference to very low signal loss/attenuation over distance. Total [5].

Metal (copper) cable transmits data as varying electrical current, and any electrical current produces an electromagnetic field around the cable; this field can potentially be detected and decoded by an attacker using specialised equipment placed near the cable, without the attacker needing to physically cut into or otherwise damage the cable, which makes covert interception (wiretapping) comparatively straightforward and hard to notice. Fibre optic cable, by contrast, carries data as pulses of light that are contained within the glass/plastic core through total internal reflection and do not radiate a detectable field outside the cable in the same way; to intercept the signal, an attacker generally has to physically access and bend or splice into the fibre itself, which is technically more difficult, and doing so typically causes a measurable loss of light/signal strength that a network administrator monitoring the line is likely to detect. Weighing these points together, fibre optic cable is judged to be considerably more secure against unauthorised interception than metal cable, because covert, non-invasive tapping (the main practical threat for copper) is not realistically possible with fibre, and physical tapping of fibre is both harder to perform and easier to detect. It should be noted this evaluation is specifically about interception/eavesdropping security; other factors such as cost and installation ease are not part of this comparison. Answer: fibre optic cable is more secure against interception because it does not emit a detectable electromagnetic field like copper cable, so covert electrical tapping is not possible, and any physical tap on the fibre itself is harder to achieve and more likely to be detected through signal loss.

Level-of-response marking (3 bands), maximum [6]. Band 2 [5-6]: accurate, detailed description of how each medium can/cannot be intercepted (electromagnetic radiation from copper vs light contained in fibre), a balanced comparison specifically focused on security, a clear justified conclusion, fluent DT terminology (e.g. electromagnetic field, total internal reflection, wiretapping, signal loss), high standard of SPG. Band 1 [3-4]: competent description of at least one medium's security characteristics, some comparison attempted, a conclusion given with limited justification, some DT terminology, reasonable SPG. Band 0 [1-2]: brief/superficial description, little or no genuine comparison, unclear or missing conclusion, limited terminology. 0: no creditable response.

Parity checking works by adding one extra bit to each byte before it is sent. Using, say, even parity, the sender sets this parity bit to 0 or 1 so that the total count of 1-bits across the whole byte, including the parity bit, is an even number; both sender and receiver must have agreed in advance whether even or odd parity is being used. When the byte arrives, the receiver independently counts the number of 1-bits in the byte it received (again including the parity bit) and checks whether that count matches the agreed rule (even, in this example). If the count is odd instead of even, this mismatch tells the receiver that at least one bit must have changed value during transmission, i.e. an error occurred, and the receiver can then request that the byte be resent; if the count matches, the receiver treats the byte as (probably) correctly received. Re-checking this explanation: it correctly captures that parity only detects a single-bit (or any odd number of bit) error, and cannot identify which bit was wrong or correct it, and cannot reliably detect an even number of bit errors — this matches the standard limitation of simple parity checking as a detection-only method. Answer: a parity bit is added to make the total 1-bit count of the byte even (or odd) by agreement; the receiver recounts the 1-bits on arrival, and a mismatch with the agreed parity reveals that a transmission error has occurred.

[1] mark for reference to an extra parity bit added to the byte before transmission; [1] mark for reference to making the total number of 1-bits even (even parity) or odd (odd parity) according to an agreed rule; [1] mark for reference to the receiver recounting the 1-bits on arrival and comparing against the agreed parity to detect a mismatch/error. Total [3]. Accept description using either even or odd parity.

CRC is a more powerful, mathematically based error-detection method than a simple parity bit. Before transmission, the sender treats the block of data to be sent as one large binary number and divides it, using a specific binary (modulo-2) division method, by a fixed divisor value that both sender and receiver have agreed on in advance; the remainder left over from this division, the CRC value, is appended to the end of the original data as a form of checksum. When the data block plus its appended CRC arrives at the receiver, the receiver carries out exactly the same division, using the same agreed divisor, on the data it actually received; if no error occurred during transmission, this division should produce a remainder of zero. If the resulting remainder is not zero, the receiver knows that one or more bits were altered in transmission, i.e. an error has been detected. Regarding correction: CRC itself is fundamentally a detection method rather than a method that can pinpoint and directly repair the specific corrupted bit(s); in practice, once an error is detected this way, correction is achieved indirectly, by the receiver requesting that the sender retransmit the affected block of data, and the process repeats until a transmission is received with a zero remainder. Checking this description: it clearly separates the mathematical detection mechanism (modulo-2 division producing a remainder/CRC value, and re-checking that remainder at the receiver) from how correction is actually achieved (retransmission on request), which correctly matches how CRC is used in real systems and avoids overstating CRC as a self-correcting code. Answer: CRC detects errors by having the receiver repeat the sender's modulo-2 division of the data by an agreed divisor and checking for a zero remainder; a non-zero remainder reveals an error, which is then corrected by requesting retransmission of the data block.

[1] the data is divided by a fixed, pre-agreed divisor value using binary/modulo-2 division; [1] the remainder of this division (the CRC) is appended to the data before transmission; [1] the receiver repeats the same division on the received data (including the appended CRC); [1] a non-zero remainder at the receiver indicates that an error has occurred (a zero remainder indicates no detected error); [1] correction is achieved by the receiver requesting retransmission of the data, rather than CRC directly repairing the corrupted bit(s); [1] for overall accurate, coherent description linking detection to the correction mechanism. Total [6].

(i) A primary key's defining property is uniqueness: its value must be different for every single row in the table (no two records may share the same primary key value), and it must never be left empty (null), because its whole purpose is to let the database engine, and other tables, reliably identify one specific record. (ii) A foreign key's defining property is that it creates a link between two tables: it is a field placed in the 'many' side table whose values must match values of the primary key in the related ('one' side) table, which is exactly the mechanism that lets, for example, a BOOKING record be tied back to the correct MEMBER record. Re-reading both definitions confirms they describe different roles — a primary key identifies records within its own table, while a foreign key references another table's primary key to link tables together — which matches the distinct, complementary purposes required by the question. Answer: (i) primary key = a field that uniquely identifies each record in its own table, never duplicated or null; (ii) foreign key = a field that references/matches the primary key of a related table to link the two tables.

(i) [2]: [1] reference to uniquely identifying each row/record; [1] reference to no duplicates/no null values allowed. (ii) [2]: [1] reference to a field that links to/matches another table's primary key; [1] reference to creating a relationship between two tables. Total [4].

First normal form requires every field in a table to hold a single, atomic value, and it does not allow repeating groups of fields within one row. In the BOOKING table, however, the class-related information (ClassCode, ClassName, InstructorID, InstructorName, InstructorPhone, BookingDate) is repeated twice in every row, once for 'Class 1' and again for 'Class 2', purely to allow each member to book up to two classes on a single row; this is a classic repeating group, since it is the same set of attributes duplicated side-by-side rather than each attribute appearing once. A second, related symptom confirms the diagnosis: because the table can only ever hold two classes per member, a member who books just one class (such as M002) leaves all the 'Class 2' fields empty, wasting storage, and a member wanting to book a third class could not be accommodated at all without restructuring the table — both are direct consequences of using repeating groups instead of a properly normalised structure. Answer: the table is not in 1NF because it contains a repeating group (the Class 1 / Class 2 sets of fields duplicated within one row), which also wastes space with empty fields and cannot support more than two bookings per member.

[1] mark for identifying the repeating group (Class 1/Class 2 fields duplicated in the same row); [1] mark for an explained consequence — e.g. wasted/empty fields for members with fewer bookings, and/or the fixed limit of two classes per member being an inflexible/artificial restriction. Total [2].

To reach 1NF, the repeating group identified in the previous part must be removed by giving each class booking its own row, rather than packing two bookings side-by-side into one row. This means collapsing the 'Class 1' and 'Class 2' field-sets into a single set of class fields (ClassCode, ClassName, InstructorID, InstructorName, InstructorPhone, BookingDate) that is now allowed to repeat as many times as needed, once per row, for the same member. Because a single MemberID can now appear on more than one row (one per class booked), MemberID alone can no longer uniquely identify a row, so the primary key must become the composite (MemberID, ClassCode), which correctly identifies one specific booking of one specific class by one specific member. Checking against the two example rows: M001 would now become two separate rows (M001+YOG01, and M001+SPN02), and M002 would become just one row (M002+YOG01) with no empty fields, which resolves both problems identified in part (b) (the repeating group and the wasted empty fields). Answer: BOOKING_1NF(MemberID, MemberName, MemberPhone, ClassCode, ClassName, InstructorID, InstructorName, InstructorPhone, BookingDate), primary key (MemberID, ClassCode).

[1] mark for removing the repeating group by giving one row per member-per-class (a single, non-repeated set of class fields); [1] mark for a correct composite primary key of (MemberID, ClassCode) reflecting that MemberID alone no longer uniquely identifies a row. Total [2]. Accept an equivalent field list/table name.

Second normal form requires that, in a table with a composite primary key, every non-key field must depend on the whole key, not on just part of it (no partial dependency). In BOOKING_1NF, the composite key is (MemberID, ClassCode); checking each non-key field against this: MemberName and MemberPhone depend only on MemberID (they would be the same regardless of which class is being booked), which is a partial dependency and therefore breaks 2NF; likewise ClassName, InstructorID, InstructorName and InstructorPhone depend only on ClassCode (they would be the same regardless of which member booked the class), which is also a partial dependency. Only BookingDate genuinely depends on the combination of a specific member and a specific class (i.e. depends on the whole key), so it may remain with the composite key. Removing the partial dependencies means splitting BOOKING_1NF into three tables: MEMBER, holding the fields that depend only on MemberID; CLASS, holding the fields that depend only on ClassCode; and a slimmed-down BOOKING table holding only the composite key plus BookingDate, linked back to MEMBER and CLASS by foreign keys. Re-checking the three resulting tables confirms every non-key field now depends on the whole of its own table's key (MEMBER and CLASS each have a single-field key, so partial dependency cannot occur; BOOKING's only non-key field, BookingDate, depends on the whole composite key), satisfying 2NF. Answer: MEMBER(MemberID, MemberName, MemberPhone); CLASS(ClassCode, ClassName, InstructorID, InstructorName, InstructorPhone); BOOKING(MemberID, ClassCode, BookingDate), with MemberID and ClassCode as foreign keys in BOOKING.

[1] mark for correctly identifying/removing the partial dependency of MemberName/MemberPhone on MemberID alone (MEMBER table); [1] mark for correctly identifying/removing the partial dependency of ClassName/InstructorID/InstructorName/InstructorPhone on ClassCode alone (CLASS table); [1] mark for a correctly retained BOOKING table with composite key (MemberID, ClassCode) and BookingDate, with appropriate foreign keys. Total [3].

Third normal form requires that every non-key field depends only on the primary key, and not on another non-key field (no transitive dependency). Examining the CLASS table from part (d), CLASS(ClassCode, ClassName, InstructorID, InstructorName, InstructorPhone): ClassName correctly depends directly on ClassCode (which class has which name), but InstructorName and InstructorPhone actually depend on InstructorID (which instructor has which name and phone number), and only depend on ClassCode indirectly, through InstructorID — i.e. ClassCode determines InstructorID, which in turn determines InstructorName and InstructorPhone, which is precisely the definition of a transitive dependency and breaks 3NF (also risking data duplication/inconsistency if the same instructor teaches two classes, since their name/phone would be stored twice). To fix this, InstructorName and InstructorPhone are removed from CLASS and placed in a new INSTRUCTOR table keyed on InstructorID, leaving CLASS holding only ClassCode, ClassName and the foreign key InstructorID linking to INSTRUCTOR. The MEMBER and BOOKING tables from part (d) contain no similar transitive dependencies (MEMBER's fields depend only on MemberID directly; BOOKING's only non-key field, BookingDate, depends on the whole composite key directly), so they remain unchanged. Re-checking the four final tables confirms every non-key field now depends on its own table's key only, directly, with no field depending on another non-key field, which satisfies 3NF. Answer: MEMBER(MemberID, MemberName, MemberPhone); INSTRUCTOR(InstructorID, InstructorName, InstructorPhone); CLASS(ClassCode, ClassName, InstructorID); BOOKING(MemberID, ClassCode, BookingDate).

[1] mark for correctly identifying the transitive dependency (InstructorName/InstructorPhone depend on InstructorID, not directly on ClassCode); [1] mark for correctly creating a new INSTRUCTOR table keyed on InstructorID holding InstructorName and InstructorPhone; [1] mark for a correctly slimmed CLASS table (ClassCode, ClassName, InstructorID) with InstructorID as a foreign key; [1] mark for correctly leaving MEMBER and BOOKING unchanged from 2NF. Total [4].

Building the ER model starts from the four 3NF tables and their keys: MEMBER(MemberID*, MemberName, MemberPhone), INSTRUCTOR(InstructorID*, InstructorName, InstructorPhone), CLASS(ClassCode*, ClassName, InstructorID), and BOOKING(MemberID*, ClassCode*, BookingDate), where * marks (part of) the primary key. Each foreign key defines one relationship: BOOKING's MemberID foreign key links to MEMBER, and since one member can appear in many BOOKING rows (one per class booked) but each BOOKING row has exactly one MemberID, this is a one-to-many relationship from MEMBER to BOOKING; similarly, BOOKING's ClassCode foreign key links to CLASS, and since one class can appear in many BOOKING rows but each BOOKING row has exactly one ClassCode, this is a one-to-many relationship from CLASS to BOOKING. Separately, CLASS's InstructorID foreign key links to INSTRUCTOR, and since one instructor can teach many classes but each CLASS row has exactly one InstructorID, this is a one-to-many relationship from INSTRUCTOR to CLASS. Crucially, the underlying real-world relationship between MEMBER and CLASS is many-to-many (a member books several classes; a class has several members), and the question specifically requires no many-to-many relationships in the model — this is exactly why BOOKING exists as a linking (associative) entity: it breaks the single many-to-many relationship into two one-to-many relationships (MEMBER-to-BOOKING and CLASS-to-BOOKING), which is the standard, correct way to resolve a many-to-many relationship in an ER model. Checking the final model: three relationships are named, all are one-to-many, none is many-to-many, and every foreign key identified in the 3NF tables is accounted for by exactly one relationship, confirming the model is complete and consistent with the normalised tables. Answer: MEMBER to BOOKING (one-to-many), CLASS to BOOKING (one-to-many), INSTRUCTOR to CLASS (one-to-many), with BOOKING resolving the underlying many-to-many relationship between MEMBER and CLASS.

[1] all four entities correctly identified (MEMBER, INSTRUCTOR, CLASS, BOOKING); [1] correct primary keys shown for each entity (composite key correctly shown for BOOKING); [1] correct foreign keys shown (MemberID and ClassCode in BOOKING; InstructorID in CLASS); [1] MEMBER-to-BOOKING relationship correctly stated as one-to-many; [1] CLASS-to-BOOKING relationship correctly stated as one-to-many; [1] INSTRUCTOR-to-CLASS relationship correctly stated as one-to-many; [1] correct recognition/explanation that BOOKING resolves the underlying many-to-many relationship between MEMBER and CLASS, so no many-to-many relationship appears in the model. Total [7].

Reading the statement clause by clause: CREATE TABLE BOOK begins the creation of a new table named BOOK; each line inside the brackets defines one field, giving its name and data type — ISBN as a fixed-length 13-character text field, Title as variable-length text up to 100 characters, Author up to 60 characters, Genre up to 30 characters, and Stock as a whole-number (integer) field; the words PRIMARY KEY attached to ISBN mean the database will enforce that every ISBN value is unique across the table and cannot be left blank, since it is the field used to uniquely identify each book record. The overall effect of running this statement is that a brand-new, completely empty BOOK table is added to the database's structure, ready to have book records inserted into it — no data rows are created by this statement itself, since CREATE TABLE only defines structure (it is DDL), not data. Checking against the DDL/DML distinction: statements that define the structure of the database (CREATE, ALTER, DROP) are DDL, while statements that manipulate the data within that structure (SELECT, INSERT, UPDATE, DELETE) are DML — CREATE TABLE is unambiguously DDL, confirming the classification. Answer: the statement is DDL; it creates a new, empty BOOK table with fields ISBN (13-char text, primary key), Title, Author, Genre (text) and Stock (integer).

[1] mark for correctly identifying this as DDL (defines table structure, not data); [1] mark for stating it creates a new table called BOOK with the five named fields/data types; [1] mark for correctly explaining ISBN as the primary key (unique, not null, identifies each record). Total [3].

The SELECT clause names exactly which fields to retrieve, Title and Author (not every field in the table), the FROM clause names the table being queried, BOOK, and the WHERE clause filters the rows so that only records where Genre equals the text 'Science Fiction' are included in the result. The effect of running this query is therefore a results list showing just the title and author of the science-fiction books held by the library — none of the underlying data in the BOOK table is altered, since SELECT only reads/retrieves data. Checking the statement is read correctly: it does not use SELECT *, so fields other than Title and Author (such as ISBN or Stock) are correctly excluded from the result, and the WHERE condition correctly restricts rows rather than fields. Answer: it is a DML SELECT query that retrieves and displays the Title and Author of all books with Genre 'Science Fiction', without altering the table's data.

[1] mark for correctly stating it retrieves/displays only the Title and Author fields; [1] mark for correctly stating the WHERE condition restricts the result to books with Genre = 'Science Fiction'. Total [2].

The UPDATE BOOK clause names the table whose data will be changed; the SET clause specifies which field is being changed and how — Stock = Stock - 1 takes the current value of Stock for the affected row and reduces it by exactly 1; the WHERE clause restricts this change to only the one row where ISBN matches '9780141439518', rather than affecting every book in the table. The overall effect is that the number of available copies recorded for that one specific book decreases by one, which would typically be run when a copy of that book is borrowed from the library, while the ISBN, Title, Author and Genre fields, and every other book's record, remain completely unaffected. Checking the logic: because Stock is reassigned relative to its own current value (Stock - 1) rather than to a fixed number, this statement correctly decrements rather than resets the stock count, and the WHERE clause correctly targets exactly one record via the unique ISBN primary key. Answer: it is a DML UPDATE statement that decreases the Stock value by 1 for the single book with ISBN '9780141439518', leaving all other fields and records unchanged.

[1] mark for correctly identifying this as a DML UPDATE (modifies existing data); [1] mark for correctly explaining that Stock is decreased by 1 (relative to its current value); [1] mark for correctly explaining the WHERE clause restricts the change to the one matching ISBN record only. Total [3].

The DELETE FROM BOOK clause specifies that rows are to be removed from the BOOK table, and the WHERE Stock = 0 condition restricts this removal to only those rows where the Stock field currently holds the value 0. The effect is therefore that every book record with zero available copies is permanently deleted from the table (not just hidden or marked); any book record with one or more copies in stock is unaffected and remains in the table exactly as before. Checking the statement: unlike UPDATE, which changes field values within a row, DELETE removes the whole row/record, and because a WHERE clause is present, this is not an unconditional delete-everything statement — only rows genuinely matching Stock = 0 are removed, so books currently in stock are correctly preserved. Answer: it is a DML DELETE statement that permanently removes every book record with Stock equal to 0 from the BOOK table, leaving books that have stock untouched.

[1] mark for correctly identifying this as a DML DELETE (removes whole records); [1] mark for correctly explaining that only records with Stock = 0 are removed, all other records unaffected. Total [2].

(i) The knowledge base's purpose is purely to hold the content of the system's expertise: facts about the domain (e.g. symptoms, components, conditions) and the rules (typically IF...THEN rules) that link those facts to conclusions, built up by a knowledge engineer from a human expert's knowledge. (ii) The inference engine's purpose is to use that stored knowledge: it takes the facts entered about a specific case by the current user and systematically applies the rules from the knowledge base, chaining through them logically, to reach a conclusion or recommendation and to determine which further question would most usefully narrow down the answer next. Contrasting the two: the knowledge base can be thought of as the expert system's 'memory' of domain expertise, which does nothing by itself, while the inference engine is the 'reasoning' component that actively processes that memory against the current case to actually produce an answer — without the inference engine, the knowledge base's rules would simply sit unused, and without the knowledge base, the inference engine would have no domain rules to reason with. Answer: (i) the knowledge base stores the domain facts and rules; (ii) the inference engine applies logical reasoning over those rules and the user's input facts to reach conclusions; the knowledge base is passive storage of expertise, while the inference engine is the active reasoning process that uses it.

[1] mark for correctly stating the purpose of the knowledge base (stores facts/rules of the domain); [1] mark for correctly stating the purpose of the inference engine (applies reasoning/rules to reach a conclusion/decide next question); [1] mark for a valid contrast between the two roles (e.g. passive storage vs active reasoning). Total [3].

Heuristics address the problem-solving strategy of the expert system: rather than exhaustively checking every possible rule/combination to guarantee the mathematically best answer (which could be very slow for a complex problem), the system uses experience-based shortcuts or rules of thumb that usually lead to a good, workable answer more quickly, accepting that the answer found is not guaranteed to be perfect or unique. Fuzzy logic, by contrast, addresses how the expert system handles uncertain or imprecise data values: instead of forcing every fact into a strict true/false or a single crisp number, fuzzy logic allows a value to partially belong to more than one category at once (with a degree of membership, e.g. 70% 'warm' and 30% 'hot'), which better reflects how real-world, imprecise concepts (like temperature, age, or risk) actually behave. Contrasting the two directly: heuristics are about how the system searches for/reaches a decision efficiently, while fuzzy logic is about how the system represents and reasons with imprecise values/data — a system could use one, both, or neither, since they solve different problems (search efficiency vs handling vagueness). Answer: heuristics are practical rule-of-thumb strategies used to reach a workable decision efficiently without guaranteeing the optimal answer; fuzzy logic instead allows reasoning with degrees of truth/partial membership rather than strict true/false values, so it handles imprecise data rather than search efficiency.

[1] mark for a valid explanation of heuristics (rule-of-thumb/experience-based strategy to reach a workable, not necessarily optimal, answer more efficiently); [1] mark for a valid explanation of fuzzy logic (reasoning with degrees of truth/partial membership rather than strict true/false); [1] mark for an explicit, valid contrast between the two (e.g. search-strategy efficiency vs handling imprecise data/values). Total [3].

Considering the advantages: a medical expert system can hold and instantly apply a knowledge base drawing on far more recorded medical knowledge and past cases than any single doctor could remember, and it applies its rules with total consistency every time (unaffected by tiredness, workload or a doctor's individual experience/specialism), while also being available continuously, which is useful in an out-of-hours or remote setting where an appropriate human specialist may not be immediately available; it can also act as a useful second opinion or triage tool, flagging cases that need urgent human attention. Considering the disadvantages: an expert system can only reason using the rules and facts it has been given, so a patient whose case is unusual, or involves symptoms/interactions not anticipated by the knowledge engineer when building the system, may be diagnosed incorrectly or not at all, whereas an experienced human doctor can draw on intuition, wider context about the patient (e.g. their history, emotional state, things not directly asked about) and can adapt their questioning creatively in a way a rule-based system may not; a doctor also provides reassurance, empathy and the ability to explain a diagnosis sensitively, which a purely computerised system cannot fully replicate, and there are also serious questions of legal/ethical accountability if a wrong diagnosis is given by software rather than a licensed professional. Weighing these against each other, the expert system's advantages of consistency, breadth of knowledge and availability are valuable, but its disadvantages, particularly the risk of a wrong diagnosis outside its programmed rules and the lack of human judgement/empathy, mean it should be judged as most appropriate as a decision-support tool used alongside a human doctor, rather than as a full replacement for one, especially for complex or unusual cases. Answer: an expert system offers consistent, broad, always-available diagnostic support but cannot match a human doctor's judgement, adaptability and empathy, and risks misdiagnosing cases outside its programmed rules, so it is best evaluated as a valuable support tool for a doctor rather than a full replacement.

Level-of-response marking (3 bands), maximum [6]. Band 2 [5-6]: detailed, accurate description of at least two genuine advantages and at least two genuine disadvantages, a balanced evaluation directly comparing the expert system with a human doctor, a clear justified overall conclusion, fluent DT terminology (e.g. knowledge base, rule-based reasoning, decision support), high standard of SPG. Band 1 [3-4]: competent description of some advantages and disadvantages (may be unbalanced or list-like rather than evaluative), a conclusion given with limited justification, some relevant terminology, reasonable SPG. Band 0 [1-2]: brief/superficial points, mostly one-sided (advantages only or disadvantages only), no clear conclusion, limited terminology. 0: no creditable response.

Speech begins as sound waves — variations in air pressure — which the microphone's diaphragm physically vibrates in response to, converting this vibration into a corresponding analogue electrical signal whose voltage varies continuously in the same pattern as the original sound wave. Because this analogue signal is continuous and a computer can only work with discrete digital (binary) values, an analogue-to-digital converter (ADC) is used next: it repeatedly measures (samples) the voltage of the analogue signal at a very high, regular rate, and converts (quantises) each individual measurement into a binary number representing that sample's amplitude. The resulting stream of binary numbers is a digital representation of the original speech signal, which can now be stored, processed and compared against reference patterns by the voice recognition software — something that would not be possible with the original continuous analogue signal. Re-checking this explanation, it correctly places the two components in their true order of operation (microphone captures/converts sound to analogue electricity first, ADC digitises that analogue signal second) and correctly explains why digitisation is necessary (computers process binary data only). Answer: the microphone converts sound waves into an analogue electrical signal, and the ADC samples that signal at regular intervals, converting each sample into binary, to produce a digital version of the speech for the software to process.

[1] mark for correctly explaining the microphone converts sound waves into an analogue electrical signal; [1] mark for correctly explaining the ADC samples/measures the analogue signal at regular intervals; [1] mark for correctly explaining each sample is converted into a binary/digital value, producing digital data the computer can process. Total [3].

Once the speech has been digitised (as covered in the previous part), the voice recognition software analyses the digital signal and breaks it down into small segments corresponding to individual sound units, or phonemes, since human speech is built from a limited set of these basic sounds. Each segment's characteristics (such as its frequency pattern) are then compared, using pattern-matching techniques, against a stored database/library of reference patterns that the system holds for every phoneme or word in its recognised vocabulary; the software calculates how closely the input segment matches each stored pattern and selects the reference pattern that is the closest (best statistical) match. By matching a whole sequence of input segments against sequences of stored word patterns in this way, and often also applying knowledge of which word sequences are grammatically/statistically likely (a language model), the system builds up and outputs the most probable word, or sequence of words, that the user actually spoke. Checking the logic of this description confirms it correctly separates the two stages already covered — digitisation happens first (previous part), and pattern matching against stored reference data happens second (this part) — and correctly identifies that the output of pattern matching is the recognised text/word(s), which is the ultimate purpose of the voice recognition system. Answer: the digitised signal is broken into phoneme-sized segments, each compared by pattern matching against a stored database of reference sound patterns, and the closest-matching stored pattern(s) determine the recognised word(s) that are output.

[1] mark for reference to the digitised speech being broken down/segmented into phonemes or similar units; [1] mark for reference to comparing these units against a stored database/library of reference patterns; [1] mark for reference to selecting the closest match to determine/output the recognised word(s). Total [3].

(i) Virtualisation works by using special software (a hypervisor) to divide the resources of one physical machine, such as its processor, memory and storage, into several separate, isolated virtual machines, each of which behaves to its users and software as if it were its own independent computer, even though several virtual machines may actually be sharing the same underlying physical hardware; this is what allows a cloud provider to run many customers' virtual servers on a smaller number of physical machines. (ii) Clustering works by connecting several separate physical servers together, using software that coordinates them, so that from the outside they appear and function as one combined, more powerful system; the workload (e.g. incoming requests) can be shared/balanced across the servers in the cluster, and if one physical server in the cluster fails, the remaining servers can continue handling the workload, giving the overall service greater reliability/availability than relying on a single server. Re-reading both explanations confirms virtualisation is about dividing one physical machine into many virtual ones, while clustering is about combining many physical machines into what behaves as one system — these are, appropriately, near-opposite (though complementary) cloud infrastructure techniques, matching what distinct terms should describe. Answer: (i) virtualisation creates multiple independent virtual machines/resources on a single physical machine; (ii) clustering links multiple physical servers together to act as one system, sharing workload and providing redundancy if one server fails.

(i) [2]: [1] reference to creating a virtual/software-based version of a physical resource; [1] reference to multiple virtual instances running on shared physical hardware. (ii) [2]: [1] reference to multiple physical servers linked/working together as one system; [1] reference to sharing workload and/or providing redundancy/reliability if one server fails. Total [4].

(i) For data storage, the cloud provider maintains large-scale storage hardware within its own data centre(s); when a user saves a file 'to the cloud', that file is actually transmitted over the internet and written to the provider's storage hardware, and the provider handles all the underlying management (capacity, redundancy across multiple physical disks/locations, hardware maintenance) on the user's behalf; the user can then retrieve or edit that same file from any internet-connected device, since the data is not tied to one specific local machine. (ii) For backup, the key extra feature compared with plain storage is that copying happens automatically and regularly (e.g. continuously, hourly, or daily, according to a schedule the user or organisation sets), without requiring the user to manually save a copy each time; this creates an independent, up-to-date copy of the data that is physically separate from the user's own device/site, so that if the original data is lost or damaged for any reason local to the user (hardware failure, fire, theft, ransomware), the organisation can restore its data from the cloud copy rather than losing it permanently. Checking that these two descriptions are genuinely distinct: data storage is fundamentally about where everyday, actively-used data is kept and accessed from, whereas backup is specifically about maintaining a separate, scheduled protective copy for disaster recovery — the two services can use similar underlying cloud storage hardware but serve different purposes, which the descriptions correctly reflect. Answer: (i) cloud storage saves data on remote provider-owned hardware, accessible over the internet from any device; (ii) cloud backup automatically and regularly copies data to the cloud so it can be restored if the original local data is lost or damaged.

(i) [4]: [1] reference to data being stored on remote/provider-owned hardware in a data centre; [1] reference to access over the internet; [1] reference to access from any (authorised) device/location; [1] reference to the provider managing/maintaining the storage hardware/capacity. (ii) [4]: [1] reference to automatic/scheduled/regular copying of data; [1] reference to the copy being stored remotely/separately from the original; [1] reference to restoring data from the cloud copy if the original is lost/damaged/corrupted; [1] reference to protection against a specific risk (e.g. hardware failure, disaster, theft, ransomware). Total [8].

The Computer Misuse Act was introduced specifically to criminalise hacking and other unauthorised interference with computer systems, because such actions were not clearly covered by earlier law. Its main features are structured as a set of distinct offences of increasing seriousness. The first, most basic offence is unauthorised access to computer material, which covers simply viewing or attempting to view data or a program on a computer system without permission to do so, for example by guessing, stealing or misusing someone else's login details — no further criminal intent beyond the access itself needs to be proven for this offence. The second offence is unauthorised access with intent to commit, or to help someone else commit, a further offence, such as gaining unauthorised access specifically in order to then commit fraud or theft using the data found; this is treated more seriously than simple unauthorised access because of the additional criminal intent involved. The third offence is unauthorised modification of computer material, which covers deliberately changing, deleting, corrupting or otherwise impairing data, programs or the operation of a computer without permission, which includes actions such as knowingly spreading a computer virus or other malware, or deliberately deleting an organisation's files; this offence carries the most severe penalties of the three because of the direct harm/damage caused. Checking the three offences against each other confirms they form a genuinely escalating structure (access only, then access with further criminal intent, then actual damaging modification), matching how the Act is generally taught and applied. Answer: the Computer Misuse Act creates three main offences of increasing severity — unauthorised access to computer material; unauthorised access with intent to commit a further offence; and unauthorised modification of computer material (e.g. spreading malware or deleting data) — each carrying increasingly serious penalties.

[2] marks for accurately describing unauthorised access to computer material (basic offence, e.g. viewing data without permission); [2] marks for accurately describing unauthorised access with intent to commit a further offence; [2] marks for accurately describing unauthorised modification of computer material (e.g. deleting/corrupting data, spreading malware). Total [6]. Award marks within each 2-mark block as [1] for naming the offence and [1] for a valid example/detail.

For an organisation such as the leisure centre, the Data Protection Act imposes concrete legal obligations on how it handles the personal data it collects from members (e.g. name, phone number, booking history from Question 4): it must only collect data that is genuinely needed for a stated purpose, must keep that data accurate, must protect it with adequate security measures against loss or unauthorised access, and must not retain it indefinitely once it is no longer needed; failing to meet these obligations can expose the organisation to formal complaints, fines or other legal consequences, so in practice it also means the organisation has to spend time and money on things like staff training, secure storage systems and clear data-handling procedures. For a member of the public whose data is held by the organisation, the Act's impact is mainly in terms of rights gained: the individual can ask the organisation what personal data it holds about them and request a copy of it, can ask for incorrect data to be corrected, and can, in appropriate circumstances, ask for their data to be deleted or for its use to be restricted; this gives the individual meaningful control and reassurance over their own information, for example knowing the leisure centre cannot simply pass their phone number to a third party without proper justification. Checking that the two parts of the answer are properly distinct and complementary: part (i) describes the organisation's obligations/burdens under the Act, while part (ii) describes the individual's resulting rights/benefits, which correctly reflects that data protection law works by placing duties on data holders specifically in order to protect the people whose data is held. Answer: (i) the organisation must collect, use, store and retain members' data responsibly and securely, facing legal consequences if it fails to do so; (ii) members of the public gain rights to know about, access, correct and, in some cases, have deleted, the personal data the organisation holds on them.

(i) [3]: [1] reference to a specific data-handling obligation (e.g. collect only necessary data, keep it accurate, store it securely, not keep it longer than necessary); [1] reference to a second, different obligation from the list above; [1] reference to a consequence for the organisation of non-compliance (e.g. fines, legal action, loss of trust). (ii) [3]: [1] reference to the right to know about/access their own data; [1] reference to the right to have inaccurate data corrected and/or deleted/restricted; [1] reference to the resulting benefit/reassurance for the individual (e.g. control over/protection of their personal information). Total [6].