Hypothetical Recommendation: A Study of Interactive Profile Manipulation Behavior for Recommender Systems

Explanation and dynamic feedback given to a user during the recommendation process can influence user experience. Despite this, many real-world recommender systems separate profile updates and feedback, obfuscating the relationship between them. This paper studies the effects of what we call hypothetical recommendations. These are recommendations generated by low-cost, exploratory profile manipulations, or "what-if" scenarios. In particular, we evaluate the effects of dynamic feedback from the recommender system on profile manipulations, the resulting recommendations and the user's overall experience. Results from a user experiment (N=129) suggest that (i) dynamic feedback improves the effectiveness of profile updates, (ii) when dynamic feedback is present, users can identify and remove items that contribute to poor recommendations, (iii) profile update tasks improve perceived accuracy of recommendations and trust in the recommender, regardless of actual recommendation accuracy.

[1]  John Riedl,et al.  GroupLens: an open architecture for collaborative filtering of netnews , 1994, CSCW '94.

[2]  John Riedl,et al.  Explaining collaborative filtering recommendations , 2000, CSCW '00.

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

[4]  Patrick Seemann,et al.  Matrix Factorization Techniques for Recommender Systems , 2014 .

[5]  Alfred Kobsa,et al.  Inspectability and control in social recommenders , 2012, RecSys.

[6]  Michael D. Buhrmester,et al.  Amazon's Mechanical Turk , 2011, Perspectives on psychological science : a journal of the Association for Psychological Science.

[7]  Jonathan L. Herlocker,et al.  Evaluating collaborative filtering recommender systems , 2004, TOIS.

[8]  Yehuda Koren,et al.  Matrix Factorization Techniques for Recommender Systems , 2009, Computer.

[9]  Craig Boutilier,et al.  Active Collaborative Filtering , 2002, UAI.

[10]  Panagiotis G. Ipeirotis,et al.  Running Experiments on Amazon Mechanical Turk , 2010, Judgment and Decision Making.

[11]  Tobias Höllerer,et al.  SmallWorlds: Visualizing Social Recommendations , 2010, Comput. Graph. Forum.

[12]  Barry Smyth,et al.  Passive Profiling and Collaborative Recommendation , 1999 .

[13]  Bamshad Mobasher,et al.  Towards Trustworthy Recommender Systems : An Analysis of Attack Models and Algorithm Robustness , 2007 .

[14]  Barry Smyth,et al.  Experiments in dynamic critiquing , 2005, IUI.

[15]  Bradley N. Miller,et al.  Using filtering agents to improve prediction quality in the GroupLens research collaborative filtering system , 1998, CSCW '98.

[16]  Barry Smyth,et al.  PeerChooser: visual interactive recommendation , 2008, CHI.

[17]  Rashmi R. Sinha,et al.  The role of transparency in recommender systems , 2002, CHI Extended Abstracts.

[18]  Erik Duval,et al.  Visualizing recommendations to support exploration, transparency and controllability , 2013, IUI '13.

[19]  Tobias Höllerer,et al.  TasteWeights: a visual interactive hybrid recommender system , 2012, RecSys.