Context-adaption based on ontologies and spreading activation

Hussein, T., Westheide, D., and Ziegler, J. (2007). LWA 2007: Lernen – Wissen – Adaption. Martin-Luther-Universität Halle-Wittenberg.

Abstract

Ontologies and spreading activation are known terms within the scope of information retrieval. In this paper we introduce SPREADR, an integrated adaptation mechanism for web applications that uses ontologies for representing the application domain as well as context information like location, user history and local time. Those context factors can be modeled in an ontology and be linked to certain domain nodes. In each session a Spreading Activation Network is build based on those ontologies and recognized context factors or user actions can trigger an activation flow through this network. A node’s resulting activation value then represents its importance according to the current circumstances. While identically in structure, the Spreading Activation Networks are personalized by automatically modifying link weights and activation levels of nodes. As a result the system learns about the user preferences and can adjust its adaptation mechanism for future runs through implicit feedback.

Related focus areas

Resources

Related publications

Adapting web sites by spreading activation in ontologies

Hussein, T. and Ziegler, J. (2008). ReColl ’08: Int. Workshop on Recommendation and Collaboration (in conjunction with IUI 2008). Gran Canaria.

Situationsgerechtes Recommending – Kontextadaptive, hybride Empfehlungsgenerierung

Hussein, T. and Ziegler, J. (2011). Informatik Spektrum, 34(2), 143–152.

Context Modeling for Adaptive Collaboration

Haake, J., Hussein, T., Joop, B., Lukosch, S., Veiel, D., and Ziegler, J. (2009). Essen: Abteilung für Informatik und Angewandte Kognitionswissenschaft, Fakultät für Ingenieurwissenschaften, Universität Duisburg-Essen.

VIEW MORE »