A Text Association Rules Mining Method Based on Concept Algebra

The Concept Algebra (CA) based text representation method which can be auto-constructed and used in ordinary texts´ includes more semantic information compared to Keyword methods. This method provides a new way of thinking of more accurately mining meaningful association rules. In this paper concepts association rules mining is based on CA represented texts, and use the relations among concepts to optimize the mining results with definition and calculation of association weight and rule strength. Compared with traditional method regarding keyword as independent to each other lacks semantic information while domain-oriented method would be applicable to specific areas with specialists construction. Method in this paper use CA´s superiority on utilizing Ri and Ro of concepts to remedy the two deficiencies effectively which is proved by the theoretical and experiment.