Link Prediction in Affiliation Networks
Applications of Link Prediction
1) Retail Recommender Systems
Used by e-commerce and marketing companies to recommend potential products to customers.
Amazon.com: recommends books, music and retail products to potential customers.
SkyMall.com: recommends retail products to potential customers.
Netflix.com and MovieLens.org: recommend movies to potential viewers.
[Riedl & Konstan 2002]
Applications of Link Prediction
2) Information Filtering
Used by news forums, academic publication repositories and web search engines.
GroupLens: recommends current event news articles to subscribers.
ReferallWeb: recommends authors and information to people based on a trusted chain of referrals from other users.
CiteSeer: recommends scientific publications to researchers.
[Resnick & Varian 1997]
Applications of Link Prediction
3) Biological Systems – Protein Interactions
Due to the fact that protein-protein interaction networks contain hub proteins and contain triads and tetrads at a higher frequency than random networks, it is possible to predict future protein interactions. “We show that we can leverage the information encoded in consensus interaction patterns to generate high relevance predictions for new