Considering Rating as Probability Distribution of Attitude in Recommender System

Recommender systems play increasingly significant roles in solving the information explosion problem. Generally, the user ratings are treated as ground truth of their tastes, and used as index for later predict unknown ratings. However, researchers have found that users are inconsistent in giving their feedbacks, which can be considered as rating noise. Some researchers focus on improving recommendation quality by de-noising user feedbacks. In this paper, we try to improve recommendation quality in a different way. The rating inconsistency is considered as an inherent characteristic of user feedbacks. User rating is described by the probability distribution of user attitude instead of the exact attitude towards the current item. According to it, we propose a recommendation approach based on conventional user-based collaborative filtering using the Manhattan Distance to measure user similarities. Experiments on MovieLens dataset show the effectiveness of the proposed approach on both accuracy and diversity.

[1]  Greg Linden,et al.  Amazon . com Recommendations Item-to-Item Collaborative Filtering , 2001 .

[2]  Yehuda Koren,et al.  Factorization meets the neighborhood: a multifaceted collaborative filtering model , 2008, KDD.

[3]  Nava Tintarev,et al.  Rate it again: increasing recommendation accuracy by user re-rating , 2009, RecSys '09.

[4]  Jaana Kekäläinen,et al.  Cumulated gain-based evaluation of IR techniques , 2002, TOIS.

[5]  David Heckerman,et al.  Empirical Analysis of Predictive Algorithms for Collaborative Filtering , 1998, UAI.

[6]  John Riedl,et al.  Is seeing believing?: how recommender system interfaces affect users' opinions , 2003, CHI '03.

[7]  Min Zhao,et al.  Probabilistic latent preference analysis for collaborative filtering , 2009, CIKM.

[8]  Nuria Oliver,et al.  I Like It... I Like It Not: Evaluating User Ratings Noise in Recommender Systems , 2009, UMAP.

[9]  Martha Larson,et al.  CLiMF: learning to maximize reciprocal rank with collaborative less-is-more filtering , 2012, RecSys.

[10]  Neil J. Hurley,et al.  Detecting noise in recommender system databases , 2006, IUI '06.

[11]  I. Berlin.  I like it. , 1921 .

[12]  Gediminas Adomavicius,et al.  Improving Aggregate Recommendation Diversity Using Ranking-Based Techniques , 2012, IEEE Transactions on Knowledge and Data Engineering.

[13]  Wei Chen,et al.  Opinion-Based Collaborative Filtering to Solve Popularity Bias in Recommender Systems , 2013, DEXA.

[14]  Yi Li,et al.  A hybrid recommendation algorithm adapted in e-learning environments , 2012, World Wide Web.

[15]  Wenfei Fan,et al.  Keys with Upward Wildcards for XML , 2001, DEXA.

[16]  Mark Rosenstein,et al.  Recommending and evaluating choices in a virtual community of use , 1995, CHI '95.

[17]  Thomas Hofmann,et al.  Collaborative filtering via gaussian probabilistic latent semantic analysis , 2003, SIGIR.

[18]  Francesco Ricci,et al.  User Modeling, Adaptation, and Personalization , 2014, Lecture Notes in Computer Science.