A Rough-Apriori Technique in Mining Linguistic Association Rules

This paper has proposed a rough-Apriori based mining technique in mining linguistic association rules focusing on the problem of capturing the numerical interval with linguistic terms in quanitative association rules mining. It uses the rough membership function to capture the linguistic interval before implementing the Apriori algorithm to mine interesting association rules. The performance of conventional quantitative association rules mining algorithm with Boolean reasoning as the discretization method was compared to the proposed techniques and the fuzzy-based technique. Five UCI datasets were tested in the 10-fold cross validation experiment settings. The frequent itemsets discovery in the Apriori algorithm was constrained to five iterations comparing to maximum iterations. Results show that the proposed technique has performed comparatively well by generating more specific rules as compared to the other techniques.