Assessment
Part II projects are assessed primarily on the dissertation which is awarded marks based on the following categories:
- Professional practice and presentation (14%)
- Introduction and preparation (26%)
- Implementation (40%)
- Evaluation and conclusions (20%)
Every dissertation will be read by at least two assessors. A viva voce examination or additional assessment by an expert may also be considered. Source code may be consulted. A proportion will also be read by an external examiner.
Professional Practice and Presentation
The assessors will determine whether you have taken a professional and ethical approach in your work. In particular, they will check that you have used appropriate methods and tools, understood software licenses, deployed appropriate review and evaluation techniques and been aware of the social and ethical impact of your work. You must demonstrate a structured design approach, including high-level design planning, design-for-test, consideration of human factors and systematic evaluation including confidence metrics within your evaluation where appropriate. You should explain how you would show conformance with appropriate legislation, such as that for intellectual property, data protection, human subjects and software licenses such as those for open source. Show that you understand the consequences of your project (or a more fully-formed variant of it) in terms of how it might affect commercial markets, contribute to society and/or the research community.
Regarding presentation, assessors primarily require the dissertation to be literate and tidy. It is not necessary to spend hours using an advanced graphics design package but it is necessary to write with correct grammar, in a clear and focused expository style using properly constructed sentences.
Strict adherence to the top-level arrangement of the chapters is regarded as part of the Presentation. Candidates who fail to put their names on the top right-hand corners of cover sheets, misunderstand the phrase "at most 100 words", or omit the Proforma altogether, will not receive high marks for Presentation.
The five chapters
Most of the marks are scored in the five chapters in the body of the dissertation. Assessors recognise that the precise partitioning prescribed by the five chapter headings will sometimes prove too serious a constraint. A writer might, for example, feel that it is essential to discuss some aspects of the Implementation in earlier chapters. Assessors will credit Implementation marks ahead of time in such circumstances. It is unnecessary to repeat discussion in the "correct" chapter in order to earn the marks.
The Appendices
The dissertation should be fully comprehensible without reference to any appendix. Do not rely on content in an appendix providing any examination credit because appendices are not marked. A consequence of following up a reference to an appendix may result in an adjustment to the mark for a chapter in the main body of the dissertation.
Difficulty
No marks are explicitly awarded for difficulty. Assessors are well aware that some projects are more challenging than others and take this into account as they read the dissertation.
A trivial example might be the comparison of two projects which are very much the same except that one is written in Java and the other in BCPL. The project written in BCPL will be regarded as a little more challenging if only because Java receives significant coverage in the Computer Science Tripos, while BCPL does not. In consequence an assessor might expect marginally more from the candidate who wrote in Java.
Late submission
The penalty for late submission of the dissertation is extremely severe. The formula is penalty = mark * (10 + n) / 40 where n is the integer part of the number of days late. This formula comes into play immediately after the noon deadline, when a quarter of the marks are lost.
If there should be substantial network or system failures within the University of Cambridge (including the Department, but not including Colleges), on the morning of the dissertation deadline, Student Administration will accept manual submissions. Whether substantial failures have occurred, and what the alternate procedure is, will be declared by The Examiners for Part II by the noon on the deadline day. A manual submission may involve handing in of a USB flash drive or uploading to an alternative online service.
Plagiarism and fraud
Project work is conducted in your own time and is not under constant control and supervision. It is expected that work will be done fairly, and that the dissertation will be a proper report on the work performed. If you get unusually large amounts of assistance during the year, or use code written by somebody else you must record it in your dissertation. Results shown in your dissertation must have been produced by your programs and not concocted. Obviously both general and particular claims (including ones made implicitly rather than explicitly) must be true. Note that none of these points prevent you from obtaining assistance with your project: they simply require that you present a sufficiently detailed explanation of how your results were achieved to allow the Assessors to assess the strengths of your contribution.
The University views fraud in examinations as a serious offence, and all staff members involved in the assessment of dissertations are expected to watch for and report any anomalies which could indicate its presence.
Please read the Department’s advice on avoiding plagiarism and make sure you give proper acknowledgement to the ideas and work of others. You should be aware that the electronic copy of your dissertation will be submitted to plagiarism-detection software such as TurnItIn for checking.
Viva voce examinations
The Examiners will issue a notice indicating whom they are calling for viva voce examination: only a small proportion of candidates are involved, and in recent years these have spanned the entire range of ability, not just concentrating on obvious borderlines. If selected for a viva voce examination you will be asked to make a short presentation and discuss your project. You might like to take along a copy of your program and any useful output from it not included in your dissertation. The viva voce examination is concerned only with your project, not with other aspects of the Computer Science course.
Marking guidelines for assessors
These guidelines are used by the Assessors when marking dissertations.
Note to students: It is important to be aware that the marking guidelines given below are guidelines, and not a detailed marking rubric. Unlike in a problem-solving exam, where everyone is asked the same question and is expected to produce more-or-less the same answer, Part II projects are different for every student. This means that marking necessarily relies upon the informed judgement of the Examiners. For example, in all cases we expect a solid evaluation, but what constitutes good evaluation practices can very quite widely -- consider the different standards needed to evaluate (1) a mechanised correctness proof of an algorithm, (2) an application of machine learning algorithms to a new domain, or (3) a UI/UX redesign aimed at making a piece of software more accessible to blind users. As a result, the guidelines should be treated as a qualitative guide towards writing a good dissertation, and honoring the internal logic of the project should take precedence over ticking all the boxes.
| Marks | Percent | Section |
|---|---|---|
| Professional practice and presentation | ||
| 0-5 | 0-36% | Write-up is minimal and unclear. Significant difficulty in understanding what has been done. Little evidence that a professional approach has been employed. |
| 6-9 | 43-64% | Write-up generally clear, with difficulties in some places. Significant errors in terms of planning and professional approach somewhat lacking in terms of project management. |
| 10-14 | 71-100% | Clear or excellent write-up, with at most a few minor errors. Good or excellent use of diagrams and figures. Professional approach taken across all phases of the project. |
| Introduction and Preparation | ||
| 0-10 | 0-38% | Poor or missing motivation. Incomplete or missing requirements. Little or no relevant background material presented. |
| 11-17 | 42-65% | Reasonable motivation. Some discussion on requirements, approach and tools used. Generally clear, relevant background material, with difficulties in places. |
| 18-26 | 69-100% | Clear motivation, justifying potential benefits of success. Good or excellent requirements analysis; justified and documented selection of suitable tools; good engineering approach. Clear presentation of challenging background material covering a range of computer science topics beyond Part IB. |
| Implementation | ||
| 0-9 | 0-23% | Substantially incomplete or missing implementation. Minimal discussion on approach, methods or tools. Little software written, hardware built, or analysis conducted. |
| 10-19 | 25-48% | Actual achievements may be limited or unclear, but a non-trivial deliverable is produced. Awareness of the need to select suitable methods and tools. |
| 20-27 | 50-68% | Project not particularly ambitious, or not entirely completed. A reasonable deliverable, largely making use of appropriate methods and tools. Minimal or unclear repository overview. Limitations in terms of execution or approach. |
| 28-33 | 70-83% | Challenging project meeting project success criteria. Appropriate use of mathematical, scientific and/or engineering techniques. Clear repository overview. Some limitations in terms of execution, but basically sound. |
| 34-40 | 83-100% | Contribution to the field. Application of extra-curricular reading and original interpretation of previous work from academia or industry. Challenging goals and substantial deliverables with excellent selection and application of appropriate mathematical, scientific and/or engineering techniques. Clear and justified repository overview. At most minor faults in execution or understanding. |
| Evaluation and Conclusions | ||
| 0-7 | 0-35% | Minimal or incomplete evaluation against the success criteria. Limited or incoherent results presented. Conclusions do not provide an effective summary of work completed. |
| 8-14 | 40-70% | Evaluation against the success criteria, with justification for any areas not completed. Reasonable presentation and interpretation of results which explore the effectiveness of the project. Conclusions provide a summary of work completed. |
| 15-20 | 75-100% | Clearly presented argument demonstrating success criteria met. Good or excellent evidence of critical thought and interpretation of the results which substantiate any claims of success, improvements or novelty. Conclusions provide an effective summary of work completed along with good future work. Personal reflection on the lessons learned. |
