摘要 |
PURPOSE:To improve the removing capacity of ambiguity in syntax analysis by collating partial tree structure data expressing a node or an arc as a variable with pending tree structure data and judging the validity/invalidity of the latter data. CONSTITUTION:In a device for applying morpheme analysis and syntax analysis to a Japanese document, forming tree structure data consisting of nodes indicating words and arcs indicating relation and analyzing meaning by using the tree structure, a knowledge data storing means 2 stores knowledge 1 for specifying tree structure data. The knowledge 1 includes partial tree structure data 3 expressing any one of nodes or arcs as a variable. When plural tree structure data 4 corresponding to the pending document are obtained, a tree structure data specidying means 6 is driven and specifies optimum data 5 from the plural tree structure data 4 in accordance with the knowledge 1 and a collating/judging means 7 collates pending tree structure data 4-1 with the data 3 to judge the validity/invalidity of the data 4-1. |