Mining interestingness measures for string pattern mining

摘要: In this paper we present a novel method to detect interesting patterns in strings. A common way refine results of pattern mining algorithms is using interestingness measures. But the set appropiate measures different each domain and problem. The aim our research obtain model that classify by interest. based on application machine learning generated dataset from factors features. Each row associated factor string contains values contextual information. We also propose new measure an entropy principle which improves obtained classification results. proposed avoids experts having configure parameters order patterns. demonstrated utility giving example real data. datasets scripts reproduce experiments are available on-line.