Dynamic clustering procedures for bibliographic data
Yahiko Kambayashi, Takaki Hayashi, Shuzo Yajima · ACM SIGIR Forum · 1981
Clustering is an important tool for efficient retrieval of documents in bibliographic database systems. It can be also used to find research trend from a set of research papers. This paper discusses new clustering procedures called dynamic ones which seem to be suitable for bibliographic data handling. These procedures are developed to solve the following problems.(1) Depending on the characteristics of data, several different clustering procedures are required to obtain good results.(2) Large clusters are tend to be generated.Dynamic clustering procedures are difined to be procedures which change parameter values according to the characteristics of data. Similarity values and threshold values are dynamically modified to handle the above problems. Furthermore, to treat the latter problem data duplication is considered.