The central idea of ontology engineering in semantic Web is extensive reuse of existing ontology. Ontology selection mechanism should offer support for modularization and automatic knowledge reuse. Because the chance of finding a completely covering ontology is extremely low, the present ontology selection mechanism frequently returns a collection of ontology that jointly cover the query vocabulary and the returned ontology should be extracted and combined. The efficiency of most existing ontology extraction algorithm and the decomposition algorithm is low when they try to get a suitable ontology module from a large ontology at run time. In order to solve this problem, this paper proposed a kind of ontology module selection method without the use of top ontology: Decomposes the big ontology into ontology modules in advance, inquires and chooses the corresponding ontology modules from the ontology module base in application stage, applies or composes the selected ontology modules in the next step. In this paper, relevant concepts and criterions of ontology modules selection are studied, including basic notions and conventions. Method of ontology module selection is described. Candidate ontology module identification and evaluation method are introduced; preliminary experimental results and the corresponding analysis are shown at the last section.