(2007) Folksonomies, the Semantic Web, and Movie Recommendation. In Proceedings of 4th European Semantic Web Conference, Bridging the Gap between Semantic Web and Web 2.0 (in press), Innsbruck, Austria. Retrieved on August, 2, 2007 from the University of Southampton School of Electronics and Computer Science.
Szomszor et al. provide a thorough and compelling article exploring future opportunities in leveraging tools of the Semantic Web in recommendation engines, such as is seen on Netflix and Amazon. Offering equations and examples on how such integration might be done, the group of authors presents a design for recommendation systems that might aid in offering a rich experience toward discoverability on the Web.
Recommendation systems, most commonly used by Amazon and Netflix, offer a significant ability for customers to discover new items that lay close to their taste and desire. Current trends in such systems have grown to be quite sophisticated in its ability to understand items of interest based on user behavior (what one viewed, bought, didn’t buy, left the site from, searched for, etc.) and content he or she has provided (ratings, comments, etc.).
While these system have been wildly effective in offering good solutions (I never knew I’d love the movie “Amelie!”), these recommendations are blind to anything that happens outside of their site. Netflix recommendations don’t know that I tagged last uploaded Flickr photos as Tofino, vacation, British Columbia, Canada, family, beach, and surfing, and use them to offer “Blue Crush” or “Canadian Bacon” the next time I enter their site.
Incorporating the data provided by users—how they categorize movies, actors, etc.—into recommendation systems can offer a richer experience in allowing for discoverability on the Web.
August 9, 2007 at 6:30 am
You might want to point Joytsna to your article.