A SAT-Based Approach for Enumerating Interesting Patterns from Uncertain Data

Discovering useful patterns plays an essential role in data management and data mining. Frequent itemset mining in uncertain transaction databases semantically and computationally differs from traditional techniques applied on (standard) precise transaction databases. Uncertain transaction databases consist of sets of existentially uncertain items. The uncertainty of items in transactions makes traditional techniques in applicable. Recent works propose interesting SAT-based encodings for the problem of discovering frequent itemsets in deterministic transaction databases. Our aim in this work is to extend the SAT-based encoding of frequent itemset mining to uncertain databases. Then, we propose a novel declarative mining frame-work for extracting uncertain frequent patterns from uncertain transaction databases. It makes an original use of constraints relaxation to obtain upper bounds to the expected support of frequent patterns, while guaranteeing the enumeration of all frequent itemsets with no false negatives. We experimentally evaluated our approach. The experimental results on real and synthetic data sets demonstrate the effectiveness of our proposal in mining frequent patterns.