Abstract: In text retrieve area, popular methods either considered word frequency or semantic information between retrieve terms and text corpus. These methods ignore the semantic structure information of retrieve terms and text corpus,and then the good result limits to some domains. This paper analyzed the subject verb-object structure information of text, and then computed the similarity of the words where the words lie in the subject-verb-object structure, and final1y implemented semantic information retrieving of texts. The experiment shows that the approach could improve the precision effectively.

Key words: Text information retrieve, Semantic similarity, Syntax structure information, Retroeve algorithm