Combining community-based knowledge with association rule mining to alleviate the cold start problem in context-aware recommender systems

Successful Location-Based Services should offer accurate and timely information consumption recommendations to their customers, relevant to their contextual situation. To achieve this and provide the best available recommendations to the user, researchers and developers analyse available data via exploiting data mining techniques. Unfortunately, in some cases, due to lack of available data (e.g. a relatively new member with a limited history) the above technologies and methods are not very effective. In this paper, a novel hybrid approach to alleviate the above-mentioned problem, known as cold start in context-aware recommender systems, is presented. This work aims to help researchers and developers to cope with this problem by combining a) community created knowledge, b) ontologies c) association rule mining and d) an innovative scoring function based on probability metrics.