Ontology-based inference for information-seeking in natural language dialog system

Abstract

Many natural language dialog systems have been developed with relational database (RDB) as a machine-readable knowledge source. However, RDB has some problems for answering the questions which need complex domain-specific information. In addition to the typical problems of RDB such as dependency and redundancy problems, limitations of meaning representation and storing various domain knowledge problems also exist. To solve the problems of RDB, we adopted ontology concepts as a knowledge representation method. Ontology knowledge has some advantages about representing and querying information. In this paper, we developed ontology-based approach for information-seeking to improve traditional RDB-based natural language dialog systems. This is more flexible than RDB-based dialog system for answering to complex questions. To implement our system, we designed hand-crafted ontology schemas for an electronic program guide (EPG) application and we populated ontology instances from Web pages semi-automatically. We believe that our preliminary evaluations show the possibility of our new system.