Personalization of Product Novelty Assessment via Bayesian Surprise

• Activity Number: 81
• Type: Contributed
• Date/Time: Sunday, August 3, 2014 : 4:00 PM to 5:50 PM
• Sponsor: Section on Statistics in Marketing
• Author(s): Nan Shao*+ and Kush R. Varshney and Lav R. Varshney and Florian Pinel and Anshul Sheopuri and Pavankumar Murali
• Companies: IBM Research and IBM and University of Illinois at Urbana-Champaign and IBM Research and IBM and IBM Research
• Keywords: Bayesian surprise ; personalization ; novelty assessment

Abstract

Personalization in product development is important to industries such as consumer products. In this work, we focus on novelty assessment aspect of the problem. In general, the same product may be novel to one observer, and not novel to another observer. We propose to use social networks, shared content, purchase history and internet activity history to construct a personalized set of artifacts known to a given user, and then use Bayesian surprise to calculate a personalized surprise score of the artifact. The results can be used in either serendipitous product recommendation or new product creation for targeted markets or demographics.