Inductive Concept Retrieval and Query Answering with Semantic Knowledge Bases Through Kernel Methods

This work deals with the application of kernel methods to structured relational settings such as semantic knowledge bases expressed in Description Logics. Our method integrates a novel kernel function for the ALC logic in a support vector machine that could be set up to work with these representations. In particular, we present experiments where our method is applied to the tasks of concept retrieval and query answering on existing ontologies. Key words: Inductive Concept Retrieval, Query Answering, Kernel Methods, Kernel Function, Description Logics, Semantic Web 1 Learning in Multi-Relational Settings Many application domains, spanning from computational biology and chemistry to natural language processing, require operating on structured data representations. A new emerging domain is represented by the Semantic Web (SW) [1] where knowledge intensive manipulations on complex relational descriptions are foreseen to be performed by machines. In this context, Description Logics (DLs) [2...

• Type: Conference
• Year: 2007
• Where: KES
• Authors: Nicola Fanizzi, Claudia d'Amato
• Keywords: Description Logics, Kernel Function, Kernel Methods, KES 2007