Explaining Ontology-Based Information Extraction in the NEPOMUK Semantic Desktop

Benjamin Adrian, Björn Forcher, Thomas Roth-Berghofer, Andreas Dengel

In: Thomas Roth-Berghofer , Nava Tintarev , David B. Leake (editor). Workshop 10@IJCAI-09: Explanation-aware Computing (ExaCt 2009). International Joint Conference on Artificial Intelligence (IJCAI-09) July 11-12 Pasadena California United States Pages 94-101 AAAI 7/2009.


Ontology-based information extraction uses symbolic background knowledge for extracting information from text. Thus, extraction results depend on existing knowledge fragments. Addressing the user's comprehension, trust and acceptance in results of such systems, it is necessary to explain and thus provide transparency. This paper describes explanations generated for the ontology-based information extraction system iDocument as a component of the NEPOMUK Semantic Desktop. Here, the user provides a formalized personal information model for extracting personal categories from text resources. The need for explanations in NEPOMUK's UI is quite obvious due to its complexity. The impact of different kinds of explanations on user acceptance is shown in a user evaluation and first conclusions are drawn.


German Research Center for Artificial Intelligence
Deutsches Forschungszentrum für Künstliche Intelligenz