Affiliation of Author(s):

(1) The School of Computer, Wuhan University, Wuhan, Hubei, 430079, China

Journal:

Journal of Computational Information Systems

Abstract:

Customers who have priori knowledge would feel more interest in the relation between domain knowledge of known and unknown when mining rules in domain documents. But traditional knowledge discovering methods such as frequent itemsets mining are not suitable for finding those semantic relations between terms in documents. In order to handle the mining task, this paper firstly proposes a concept fusion framework based on formal concept analysis, and then give an algorithm called Acom, which constructs priori concept space from an information formal context defined by customers and those concepts in it is utilized to annotates each document in corpus. Documents clustered by unknown feature terms would be considered as similar data sources about common unknown information. Finally the algorithm fusion all annotated concepts of documents in same cluster together with above concept fusion framework to generate association concepts which represents the relation between known concepts by customers and new concepts mined from corpus. Experiment shows that the algorithm can effectively mine rich semantic relations between concepts implied by association concepts than the relation represented by traditional frequent itemsets. Copyright © 2010 Binary Information Press May, 2010.

Co-author:

Li, Shijun(1), Zhuo(1),Zhang

Translation or Not:

no

Date of Publication:

2010-01-01