Undergraduate/Postgraduate project ideas (engineering projects) - the level of complexity can be adjusted for the appropriate level
Recommender Systems (think Amazon or Netflix)
- Unit recommender system for optional units in the School of Computing; Build a web-based platform for students to see what options they have available and allow them to rate units they have taken; calculate recommendations for new users based on ratings of previous users; display personalised recommendations; test the system (both the recommendations and the usability of the system) with users;
- Combine individual preferences for group recommendations; most recommender systems provide recommendations to individuals; using an algorithmic approach, define a way of combining individual preferences so that the group recommendations are likely to satisfy the majority of the group members; test with users;
- Include mood in the recommendation mechanism to provide individual/group suggestions in line with users' mood; test with users;
- Use partial profile matching to recommend products for new users (i.e. address the cold start problem); using an algorithmic approach, match new users with existing users and compute recommendations based on the best matches.
Data mining
- Are certain areas more prone to certain types of crimes? Using crime data from police UK, prepare and analyse the data to determine if there is a link between certain areas (e.g. areas in Portsmouth; areas in the south of England, etc.) and certain types of crimes;
- Is there a link between population density and certain types of crime? Using crime data from police UK, prepare and analyse the data to determine if there is a link between population density and certain types of crimes;
- Can we predict when/where a new crime will occur? Using crime data from police UK, prepare and analyse the data to determine if we can predict when/where a new crime will occur; would the predictions be more accurate if we focus on particular types of crime?
- Learning analytics: differences between mock and real tests. Using data from moodle, prepare and analyse the data to investigate links between performance in mock tests and real tests;
- Topic detection from student feedback. Using data from students' unit/course feedback and text mining, investigate the possibility of topic detection to a satisfactory level of accuracy.