Semantics enhanced browsing for intelligent information systems

This paper provides an approach to construct an information retrieval system combining search and browsing. The proposed retrieval mechanism is based on semantics of objects. Objects are e.g. documents, pictures, music files, software modules, system components, failure descriptions in a maintenance system, etc. The approach is based on description logics [1]: objects are represented as instances in an ABox whereas their semantics is represented via a combination of a TBox and an ABox. Search is implemented by a retrieval of all instances belonging to a certain (complex) concept - query (retrieval problem in description logics). The proposed approach enables to apply browsing to the query result. Browsing is performed by providing hints - concepts adding which to the current query would make the query more specific, i.e. concepts that allow to distinguish certain individuals in the original query result from other individuals in the query result. Application areas and further work directions are discussed.