CAJ | 학술논문

检索词自动扩展词库构建方法的基本思路是：根据语料是否规范化处理进行词库分类建设，优化了系统的检索性能；结合学科类别，对词库语料进行领域划分，引导科技人员对技术领域的准确把握；建设以本体库为基础，将与规范词具有关联性、相似性的语料通过关系表与关联库关联，把科技文献中的关键词组成一个有序的关系网，解决了传统检索系统中检索词无关联的不足；通过对检索词出现频率进行统计分析，进而更新词库，保证本体库、关联库语料的时效性，突破了人工对词库更新管理的受限性。
검색사자동확전사고구건방법적기본사로시：근거어료시부규범화처리진행사고분류건설，우화료계통적검색성능；결합학과유별，대사고어료진행영역화분，인도과기인원대기술영역적준학파악；건설이본체고위기출，장여규범사구유관련성、상사성적어료통과관계표여관련고관련，파과기문헌중적관건사조성일개유서적관계망，해결료전통검색계통중검색사무관련적불족；통과대검색사출현빈솔진행통계분석，진이경신사고，보증본체고、관련고어료적시효성，돌파료인공대사고경신관리적수한성。
For high retrieval precision and recall rate, a lexicon construction solution is introduced for query expansion in document retrieval. According to specific technology domain, an ontology based is built on basis of authoritative lexicons by the China national committee for terms and Wiktionary. Synonyms, hypernyms and hyponyms are acquired on basis of template matching and hierarchy structure reasoning from natural language contexts and Wikipedia. For better query expansion performance, a relationship network with statistical link strength is founded on basis of mutual information of related query terms. hTe above query term network enables a powerful knowledge management tool for document retrieval together with user logs and intermediate retrieval results.