Amit Goyal
Laks V.S. Lakshmanan
ABSTRACT
In recent times, collaborative filtering based Recommender Systems (RS) have become extremely popular. While research in recommender systems has mostly focused on improving the accuracy of recommendations, in this paper, we look at the "flip" side of a RS. That is, instead of improving existing recommender algorithms, we ask whether we can use an existing operational RS to launch a targeted marketing campaign. To this end, we propose a novel problem called RecMax that aims to select a set of "seed" users for a marketing campaign for a new product, such that if they endorse the product by providing relatively high ratings, the number of other users to whom the product is recommended by the underlying RS algorithm is maximum. We motivate RecMax with real world applications. We show that seeding can make a substantial difference, if done carefully. We prove that RecMax is not only NP-hard to solve optimally, it is NP-hard to even approximate within any reasonable factor. Given this hardness, we explore several natural heuristics on 3 real world datasets - Movielens, Yahoo! Music and Jester Joke and report our findings. We show that even though RecMax is hard to approximate, simple natural heuristics may provide impressive gains, for targeted marketing using RS.
Supplemental Material
- G. Adomavicius and A. Tuzhilin. Toward the next generation of recommender systems: A survey of the state-of-the-art and possible extensions. IEEE Trans. on Knowl. and Data Eng., 17, 2005. Google Scholar
Digital Library
- S. S. Anand and N. Griffiths. A market-based approach to address the new item problem. In RecSys, 2011. Google ScholarDigital Library
- S. Bhagat, A. Goyal, and L. V. S. Lakshmanan. Maximizing product adoption in social networks. In WSDM, 2012. Google ScholarDigital Library
- W. Chen, Y. Wang, and S. Yang. Efficient influence maximization in social networks. In KDD, 2009. Google ScholarDigital Library
- P. Domingos and M. Richardson. Mining the network value of customers. In KDD, 2001. Google ScholarDigital Library
- K. Y. Goldberg et al. Eigentaste: A constant time collaborative filtering algorithm. Inf. Retr., 4(2), 2001. Google ScholarDigital Library
- J. D. Hartline et al. Optimal marketing strategies over social networks. In WWW, 2008. Google ScholarDigital Library
- J. Hastad. Clique is hard to approximate within n1∈. In FOCS, 1996. Google ScholarDigital Library
- J. L. Herlocker et al. An algorithmic framework for performing collaborative filtering. In SIGIR, 1999. Google ScholarDigital Library
- D. Kempe, J. M. Kleinberg, and É. Tardos. Maximizing the spread of influence through a social network. In KDD, 2003. Google ScholarDigital Library
- X. N. Lam et al. Addressing cold-start problem in recommendation systems. In ICUIMC, 2008. Google ScholarDigital Library
- B. Mobasher et al. Toward trustworthy recommender systems: An analysis of attack models and algorithm robustness. ACM Trans. Internet Techn., 7(4), 2007. Google ScholarDigital Library
- A. Rashid, G. Karypis, and J. Riedl. Influence in ratings-based recommender systems: An algorithm-independent approach. In SDM, 2005.Google ScholarCross Ref
- F. Ricci, L. Rokach, B. Shapira, and P. B. Kantor, editors. Recommender Systems Handbook. Springer, 2011. Google ScholarDigital Library
- B. Sarwar, G. Karypis, J. Konstan, and J. Reidl. Item-based collaborative filtering recommendation algorithms. In WWW, 2001. Google ScholarDigital Library
- G. Takács et al. Matrix factorization and neighbor based algorithms for the netflix prize problem. In RecSys, 2008.Google ScholarDigital Library
- J. Wang, A. P. de Vries, and M. J. T. Reinders. Unifying user-based and item-based collaborative filtering approaches by similarity fusion. In SIGIR, 2006. Google ScholarDigital Library
Index Terms
- RecMax: exploiting recommender systems for fun and profit
Recommendations
A Scalable, Accurate Hybrid Recommender System
WKDD '10: Proceedings of the 2010 Third International Conference on Knowledge Discovery and Data MiningRecommender systems apply machine learning techniques for filtering unseen information and can predict whether a user would like a given resource. There are three main types of recommender systems: collaborative filtering, content-based filtering, and ...
A framework for diversifying recommendation lists by user interest expansion
Recommender systems have been widely used to discover users' preferences and recommend interesting items to users during this age of information overload. Researchers in the field of recommender systems have realized that the quality of a top-N ...
Cross-representation mediation of user models
Personalization is considered a powerful methodology for improving the effectiveness of information search and decision making. It has led to the dissemination of systems capable of suggesting relevant and personalized information (or items) to the users,...
Comments