Abstract: Retrieval is an important way of information acquisition.The classic means of retrieval is still dwelling on logical similarity or difference between key words without consideration of the semantic relations.Based on knowledge organization system of ontology for the purpose of retrieval，a way of ontology-oriented document and query representation named semantic vector is proposed；further a way of similarity calculation is founded to create condition for semantic retrieval and to make the query result care more about semantic matching.At last an experimental study is presented under the guidance.

摘要： 检索是获取信息的重要方式。传统检索只停留在关键字异同的逻辑层面，忽略了语义层面的信息。以本体的知识组织体系为基础，以检索应用为目标，提出面向本体的文档和查询的语义向量表示方法，进而建立面向本体的相似度计算方法，为语义检索创造条件，检索结果关注语义层面的匹配。并在理论的指导下，进行实验和分析。