A Unification-Based Approach for Chinese Inquiry Sentences Processing

In this paper, we present a unification-based approach for processing natural inquiry sentences on the relational database. We also propose a method to process the proper nouns in the queries. The semantics of some words in the inquiry sentences may have close relationship with the relational schema; they therefore can be expressed in the relation algebra. We design four types of feature structures, value specification type, attribute specification type, relation specification type and function specification type. as semantic representations to describe the corresponding algebraic content of the signs of the inquiry sentence. Unification is used as the primary information-combining operation to construct the syntactic and semantic content. In this system, we perform the syntactic parsing and the semantic interpretation in an integrated way. As a result, when the parser unifies two constituents. their algebraic contents are formed. Lastly, according to the algebraic content of the inquiry sentence, we design transformation rules to trans1ate the feature structure to database retrieval commands and then get the desired data from the database.