Topic-Based Recommendations for Enterprise 2.0 Resource Sharing Platforms

Companies increasingly often deploy social media technologies to foster the knowledge transfer between employees. As the amount of resources in such systems is usually large there is a need for recommender systems that provide personalized information access. Traditional recommender systems suffer from sparsity issues in such environments and do not take the users' different topics of interest into account. We propose a topic-based recommender system tackling these issues. Our approach applies algorithms from the domain of topic detection and tracking on the metadata profiles of the users' preferred resources to identify their interest topics. Every topic is represented as a weighted term vector that can be used to retrieve unknown, relevant resources matching the users' topics of interest. An evaluation of the approach has shown that our method retrieves on-topic resources with a high precision.