Interactive Topic Search System Based on Topic Cluster Technology

In this paper, we develop an interactive hierarchical topic search system. In our system, the generation of topic names is mainly based on the N-gram statistical language model. The construction of hierarchical tree relationships between topics is mainly based on the concept of mathematical sets. In this study, the concept of mathematical sets not only helps the system to construct a topic hierarchy tree quickly, but also allows different users to use different binary operations to generate different interactive search results. In general, this study has the following three advantages. First, the generated topic names are presented in a hierarchical form rather than a flat form. Secondly, the interactive search for this study was achieved by non-stored user search and click history. Therefore, our approach can avoid personal privacy and large storage space issues. Finally, the concept of mathematical sets not only allows us to generate topic trees in linear time, but also allows users to run all possible binary operations to meet various interactive search needs.