content classifier

Knowledge-oriented content-based filtering techniques are among the most effective ways to search, filter, and manage information resources. In this paper, a novel filtering framework for (textual) information resource management purposes is introduced. The proposed method uses the collective knowledge of ontology and structured knowledge bases for developing semantic similarity methods. The semantic similarity methods are used to filter and classify documents in accordance with user preferences. Also, the knowledge-based semantic similarity methods are integrated in a “Mixture of Experts” model so that information resources and documents are filtered and managed based on the collective knowledge of these methods (experts). The integration of knowledge-based methods in the learning "Mixture of Experts" model is a novel idea and one of the main contribution of this paper. The evaluation results suggest that the integration of knowledge-based semantic similarity measures in "Mixture of Experts" model improves system performance and leads to the accurate classification of documents.