Towards User-Aware Rule Discovery (workshop paper)
Published in the 11th International Workshop on Information Search, Integration and Personalization, ISIP, 2016
Abstract
Rule discovery is a challenging but inevitable process in several data centric applications. The main challenges arise from the huge search space that needs to be explored, and from the noise in the data, which makes the mining results hardly useful. While existing state-of-the-art systems pose the users at the beginning and the end of the mining process, we argue that this paradigm must be revised and new rule mining algorithms should be developed to let the domain experts interact during the discovery process. We discuss how new systems that embrace this approach overcome current limitations and ultimately result in shorter time and smaller user effort for rule discovery.