构件的合理分类是实现构件高效检索的基础和关键。针对目前应用广泛的刻面分类方法存在主观性因素的弊端,采用刻面分类和全文检索相结合的方法来描述构件。在此构件描述的基础上,利用聚类分析技术和语义分析技术提出一种基于语义的构件聚类索引树。并通过实验验证,该聚类索引树是可行的,有效地克服刻面分类方法的缺点,在一定程度上实现对构件的语义检索,而且具有较高的构件查全率和查准率。此外,用户在描述检索条件时,不再局限于限定的术语,更方便于普通用户。
The reasonable classification of components is the basis and key of component efficient retrieval. In order to overcome the disadvantages of faceted classification method widely used, adopts a method combining faceted classification with full-text retrieval to describe compo-nents. Based on that description, proposes a component cluster index tree in which uses cluster analysis technique and semantic analysis technique. And the experiments prove that the index tree is feasible, which can effectively overcome the disadvantages of faceted classifi-cation method. Meanwhile to some extent, it can achieve the component semantic retrieval and has higher component recall ratio and pre-cision ratio. Moreover, the description of retrieval conditions is no longer limited by restrictive terms so as to be convenient for general users.