This paper proposes an algorithm combining statistics with rules,called the priority merger algorithm,which merges phrases by calculating the relationship of words,checks phrases by using phrase rules and a semantic dictionary in grammar and semantics,and recognizes Chinese phrases by the means of hierarchic analysis.

Through building it′ "semantic dictionary" module and " primitive-feature base" module in retrieval system, have finished the organic integration of two kinds of retrieval methods and improved the performance of image retrieval system on existing technical level.

To compute the semantic relatedness of words,a frequently-used method is to use the classified semantic dictionary.

Automatic query classification via constructing semantic lexicon;

The article analyzes the limitations and problems of corpus-based dictionaries from the angle of application of corpus linguistics.

This paper makes an introduction of the construction of a Chinese semantic dictionary made use of two machine readable resource(MRD)based on the 《Modem Chinese Dictionary》 (MRD) and the (Tong Yi Ci Ci Lin) (MRD).