International Symposium on Chinese Spoken Language Processing (ISCSLP 2002)

Taipei, Taiwan
August 23-24, 2002

Improvements on a Belief Network Framework for Natural Language Understanding of Domain-Specific Chinese Queries

Bonnie Mok, Helen M. Meng

The Chinese University of Hong Kong, China

This paper extends our work on natural language understanding (NLU) using Belief Networks, as proposed in [1]. We have previously devised a method for identifying the userís communicative goal(s) out of a finite set of domain-specific goals. The problem was formulated as making N binary decisions, each performed by a Belief Network (BN). This formulation allows for the identification of queries with multiple goals, as well as queries with out-of-domain (OOD) goals. Our current work presents two extensions: (i) migrating our investigation from English to Chinese; and (ii) exploring the alternate formulation of goal identification as making one N-ary decision by a single BN. Experiments with the AITS (Air Travel Information System) corpus showed that the N-ary formulation improved over the N binary formulation in terms of single/multiple goal identification accuracies and OOD rejection.

Full Paper

Bibliographic reference.  Mok, Bonnie / Meng, Helen M. (2002): "Improvements on a belief network framework for natural language understanding of domain-specific Chinese queries", In ISCSLP 2002, paper 107.