Presentation is loading. Please wait.

Presentation is loading. Please wait.

Ranking Potentially Popular Items from Early Votes Peifeng Yin +

Similar presentations


Presentation on theme: "Ranking Potentially Popular Items from Early Votes Peifeng Yin +"— Presentation transcript:

1 Ranking Potentially Popular Items from Early Votes Peifeng Yin +
Ranking Potentially Popular Items from Early Votes Peifeng Yin + Ping Luo # Min Wang * Wang-Chien Lee § + § Pervasive Data Access Group, Pennsylvania State University {+pzy102, # * Hewlett Packard Labs in China {#ping.luo, Why rank potential popularity? Conformer-Maverick Model For common people Find interesting items as early as possible Avoid “rich-get-richer” awkwardness Popularity means obsolete, e.g., hot deals Conformer Maverick Popular Item Unpopular Item For merchandise Find proper advertisement slot Problem Formulation Item 1 Item 2 Item n Item 3 Item 1’ Item 2’ Item k’ Item 3’ Popularity Evaluation Metrics Root mean squared error Normalized discount cumulative gain Observation Experiment Early vote vs. popularity Individual prediction General Comparison No. of Early Votes Conclusion Future Work CM model assumes a binary latent personality, Conformer & Maverick for each person. Potential popularity can be ranked based on the early voters’ personality distribution Extend the CM model to support multi-value rating, e.g., 1 - 5 Explore the topic-sensitive CM personality distribution, e.g., movie vote References G. Szabo and B. o. A. Huberman. Predicting the popularity of online content. Communications of the Acm, 53(8):80-88, 2010. K. Lerman and T. Hogg. Using a model of social dynamics to predict popularity of news. In WWW, pages , 2010. J. L. Herlocker, J. A. Konstan, A. Borchers, and J. Riedl. An algorithmic framework for performing collaborative filtering. In SIGIR, pages , 1999. Y. Koren. Factorization meets the neighborhood: a multifaceted collaborative filtering model. In KDD, pages , 2008.


Download ppt "Ranking Potentially Popular Items from Early Votes Peifeng Yin +"

Similar presentations


Ads by Google