Integration of concept-driven semantic interpretation with speech recognition

This paper discusses issues concerning incorporating speech recognition with semantic interpretation based on concept. In our approach, a concept is a unit of semantic interpretation and an utterance is regarded as a sequence of concepts with an intention to attain both linguistic robustness and constraints for speech recognition. First, we propose a basic search method for detecting concepts from a phrase lattice by island-driven search evaluating the linguistic likelihood of concept hypotheses. Second, an improved method to search efficiently for N-best meaning hypotheses is proposed. Experimental results show that the basic search method has achieved a 92% understanding rate, and that the improved search technique reduced search processing time to approximately one-tenth in comparison with the basic search method without reducing performance. A scheme of understanding complex sentences based on the proposed method is also described