Supplementing Case-based Recommenders with Context Data

We propose that traditional case-based recommender systems can be improved by informing them with context data describing the user’s environment. We outline existing applications with similar objectives and describe an application of our own — Ticketyboo — which uses music listening preferences and context information from users’ calendars to recommend tickets for music concerts. This data is gathered by virtual sensors that monitor each user’s music player and calendar applications. The novelty of this approach is that context data is provided to Ticketyboo via a dedicated context infrastructure. This results in a clear separation between the providers and consumers of context data. By utilising context data in this way, minimal user input/feedback is required to guide the system since the need for explicit user feedback is negated.