Concept Hierarchy In Web Mining: Construction, Application And Representation

Concept Hierarchy is the hierarchical organization of the large quantity of concepts. In this form, concepts in the lower level are more specified than their superiors and thus can be subsumed by higher concepts. In the hierarchical concept model, distance between the concepts is different depending on the symantic relationships between them, which makes such model different from the flat concept model. Such kind of model is more realastic and closer to the human perspective, makes classifing, clustering, and matching work based on it more reasonable. The construction of the concept hierarchy defines the rule of hierarchical organization. The symatnic relationships between the conepts can be measured by the concept contents which is mapped from other information sources(such as words, querylogs, documents). The hierarchically organzation of the concepts is no doubt one of the common problem in web mining, and many applications are stepping on it.The work in the thesis can be approximated divided into three parts. We first concentrate on developing the method of mining well organized and meaningful concept hierarchy from web contents. Our work is based on the growing social annotation system. We designed a system to extract concepts and build relationships from social annotation data. Then web based application, such as suggestion work is proposed based on concept hierarchy. We proposed a novel solution to meet the need of keyword suggestion in the search engine advertising field. Our approach tries to improve the coverage and accuracy of suggestion with the help of concept hierarch. Moreover, consider about the challenges in studying the large scale concept hierarchies, we developed some visualization techniques to help user browse and learn the concept hierarchies conveniently. Our methods get promising effect in both visualizing the internal relationships between the concepts and the structure of the hierarchy itself.