Building a reputation model for user generated content

Abstract

This paper is research work conducted on reputation model for user generated content on web. I started of the process by trying to understand the various aspects on which a reputation model is based. Then I started looking at what Yahoo! could do differently to engage user and provide relevant targeting and better personalization. Such a system has benefits for both user and network (Yahoo! in this case). A single reputation can provide users with network wide image and identity and also use that recognition on and off network. Having a centralized system can provide better data moderation as user identity is known throughout the network. Then I went on to do industry analysis by speaking to various industry experts and understanding different existing reputation models. Based on the primary research conducted and inputs from secondary research I understood the pillars on which such a reputation system would stand and how users react to reputational scores. In the last section of the report, I have taken a live example in Yahoo! where reputation is very essential to ensure user engagement and abuse control. User generated location is completely crowd-sourced model that needs input from users and moderation and curation also from them. My recommendations for it are based on my research and limitations that such systems carry by itself. I have explained the blocks in detail and concluded my recommendations with my thoughts from the research.