Exploiting Answer Set Programming for Building explainable Recommendations

摘要

The capability of a recommendation system to justify its proposals becomes an ever more important aspect in light of recent legislation and skeptic users. Answer Set Programming (ASP) is a logic programming paradigm aiming at expressing complex problems in a succinct and declarative manner. Due to its rich set of high level language constructs it turns out that ASP is also perfectly suitable for realizing knowledge and/or utility-based recommendation applications, since every aspect of such a utility-based recommendation capable of producing explanations can be specified within ASP. In this paper we give an introduction to the concepts of ASP and how they can be applied in the domain of recommender systems. Based on a small excerpt of a real life recommender database we exemplify how utility based recommendation engines can be implemented with just some few lines of code and show how meaningful explanations can be derived out of the box.