Classification and Retrieval through Semantic Kernels

Abstract. This work proposes a family of language-independent semantic kernel functions defined for individuals in an ontology. This allows exploiting wellfounded kernel methods for several mining applications related to OWL knowledge bases. Namely, our method integrates the novel kernel functions with a support vector machine that can be set up to work with these representations. In particular, we present preliminary experiments where statistical classifiers are induced to perform the tasks of instance classification and retrieval. 1 Ontology Mining Many application domains require operating on large repositories made up of structured data. In the field of the Semantic Web (SW) [2], knowledge intensive manipulations on complex relational descriptions to be performed by machines are foreseen. In this context, expressive languages borrowed from Description Logics (DLs) [1] have been adopted for representing ontological knowledge. Unfortunately, machine learning through logic-based metho...

• Type: Journal
• Year: 2008
• Where: KES
• Authors: Claudia d'Amato, Nicola Fanizzi, Floriana Esposito
• Keywords: Information Technology, Kernel Function, KES 2008, Language-independent Semantic Kernel, Support Vector Machines