My research project focuses on Conditional Preference Networks (CP-nets). These are graphical models for representing and reasoning with a person’s preferences over a set of discrete variables. The natural applications of these networks are in Artificial Intelligence as preference modelling is vital for applications such as recommender systems, automated decision making, and product configuration.
In my first year I created a new quantification for user preferences, given a CP-net representation, which makes reasoning with preference information easier. In particular, these rank values can improve the efficiency of answering dominance queries. A dominance query asks, given a pair of outcomes, which will the user prefer? On the surface, this appears to be a simple question and one we would naturally like to be able to answer. However, dominance testing turns out to be a complex problem. Having a quantitative measure of preference allows us to enforce a numerical bound on the query, enabling us to answer it more efficiently. The full details of these contributions can be found in my JAIR paper (see below). After this, I looked into another method of improving dominance query efficiency via preprocessing the CP-net in order to simplify the query.
Currently I am developing a new technique for learning CP-nets from observed data. The existing methods usually assume that the observed data is in the form of pairwise preferences between outcomes. However, it is not clear in most contexts how you could obtain such preferences without querying the user. Instead, I am looking into how we can learn CP-nets from the historical count data of which outcome is chosen by the user - for example which products they have previously purchased- which can be obtained without user input.
- MMath, BSc Mathematics - University of Leeds
Research groups and institutes