Dynamic cluster maintenance

摘要

Partitioning of very large document databases is a necessity to reduce the space/time complexity of retrieval operations. Modern information retrieval (IR) environments demand dynamic clustering to constantly achieve this goal. In this article, a new strategy is proposed for dynamic cluster maintenance. The strategy is based on the cover coefficient (CC) concept. The maintenance performance and behavior are tested on a database consisting of 214 ACM Transactions on Database Systems abstracts, titles, and keywords. The similarity/stability characteristics, cost analysis, and retrieval effectiveness of both unclustered and reclustered documents are among the problems studied.