Detecting noise in recommender system databases

摘要: In this paper, we propose a framework that enables the detection of noise in recommender system databases. We consider two classes noise: natural and malicious noise. The issue arises from imperfect user behaviour (e.g. erroneous/careless preference selection) various rating collection processes are employed. Malicious concerns deliberate attempt to bias output some particular manner. argue both important can adversely effect recommendation performance. Our objective is devise techniques enable administrators identify remove process any such present data. provide an empirical evaluation our approach demonstrate it successful with respect key performance indicators.