블로그 구조적인 특징정보를 활용한 클러스터 레이블 선정에 관한 연구

Abstract

Document clustering is used in various clusters as the amount of data in a document type is increased by grouping similar documents into clusters through appearance frequency or characteristics of the keywords included in the document. Labels on these clusters help users to understand the meaning of the document cluster and help to understand the relationship between each document cluster. Therefore, it is necessary to have a label that covers the meaning of the cluster and can express the characteristics. Also, it is used in various fields such as blogs as well as documents. However, blogs have a problem that it is difficult to select a representative label that can express the meaning of each cluster generated as a result of clustering because the information is widened as a lot of information is produced in real time due to its characteristics. Therefore, in this paper, we select the representative label that can cover the whole contents by expressing the characteristics of the cluster through the problem that it is difficult to select the representative label of the cluster due to a large amount of data due to the nature of the blog. First, we collect the title, body, and tags of a blog, extract only nouns, and generate a candidate keyword set through keyword normalization and specific location keyword weights. The generated candidate keyword set is generated by selecting the semantic association label with the cluster using the FP-growth algorithm. In this way, the performance of the representative label selection method which does not utilize the existing specific weight is improved, and the representative label of the blogs cluster for the specific search is determined and proved to be provided to the user.