User Modeling and Adaptivity

Contact: Till Plumbaum, Akram Alkouz

 

The "User Modeling and Adaptivity"-Cluster works on advanced methods and tools that allow collecting, aggregating and understanding user data and user behavior. This user information will then be processed and enriched with semantic information. This enriched information serves as a basis for personalization of adaptive systems and recommender systems as well as for the Smart Layouting of news articles.

As part of this research, the CC IRML concentrates on the intelligent collection of user behavioral data from Web applications and 3D worlds. Another core theme is the research on how to aggregate distributed user information. The selected approaches are focused on the application of semantic methods.

Semantic User Tracking:  User Behavior is a valuable information source for adaptive systems and to support people using a system. With the advent of the Web 2.0 Web applications become more and more dynamic, and the way users can interact with them change. Therefore, the techniques to track the user behavior have to cope with these new challenges and have to be extended with semantic techniques to collect fine-grained data from user interactions to provide better information for adaptive systems.

Expertise Modeling in Virtual Worlds: The objective of this research is to build an expert avatar in 3D virtual worlds to answer a given query and recommend services or resident avatars related to that query. The research includes collecting and learning of the avatar’s  dynamic behavior and interaction in virtual worlds. Build the social network model of the avatar. Analyze content and context of dynamic contents in virtual world for ontology enrichment. Build and manage user model in virtual worlds. These models can then be used to recommend services and persons in virtual worlds.

Aggregation of User Information Applications typically store their user information in a proprietary format in their own application depended databases. This leads to a distributed web model of a user with several partial UMs in different applications potentially duplicating information. Therefore, the challenge is to solve the heterogeneity of the user models and create one big user model.