Online stores or Web page bring information about a myriad of items such as books, CDs, restaurants or movies at the user’s fingertips. Although, the Web reduces the barrier to the information, the user is overwhelmed by the number of available items. Therefore, online stores provide recommender systems that aim to guide the user to relevant items. However, recommender systems are generally limited to the Web page’s content and the explicit or implicit ratings provided by the users on the particular Web page. User are lazy when it comes to repeat providing rating information to recommender systems on other Web pages. That is a typical lock-in situation based on high transaction costs such that people are addicted to one or at least a limited number of Web pages.
People are required to have an account on a particular Web page before being provided with interesting recommendations. People may have concerns about providing explicit or implicit ratings on items that may expose some delicate details about their privacy. But many rather small online stores do not even provide recommendations.
Thus, we need a recommender system that (1) recognizes items over various Web pages as the same and remembers the ratings for those, (2) applies algorithms to provide recommendations and (3) smoothly integrates the rating and recommendation functionality directly in the Web site. We found the basic infrastructure for such a recommender system in the Firefox Add-on API and WEKA, a common data mining library.
We formulated these requirements as a master thesis. We were very happy to engage Tobias, an excellent master student.
The described recommender system implemented as Firefox Add-on can be downloaded at s.e.a.l. group site.
Online stores and Web portals bring information about a myriad of items such as books, CDs, restaurants or movies at the user’s fingertips. Although, the Web reduces the barrier to the information, the user is overwhelmed by the number of available items. Therefore, recommender systems aim to guide the user to relevant items. Current recommender systems store user ratings on the server side. This way the scope of the recommendations is limited to this server only. In addition, the user entrusts the operator of the server with valuable information about his preferences.
In this thesis, we introduce our recommender system OMORE, a private, personal movie recommender, which learns the user model based on the user’s movie ratings. To preserve privacy, OMORE is implemented as a Mozilla Firefox add-on, which stores the user’s ratings and the learned user model locally at the client side. Although OMORE makes use of the movie features, which are provided by the different movie pages on the Amazon.com, Blockbuster, Netflix and Rotten Tomatoes.
Tobias Bannwart: “OMORE – Private, Personal Movie Recommendations implemented in a Mozilla Firefox Add-on“, ed. by Amancio Bouza, Gerald Reif and Harald C. Gall, University of Zurich, July 2009. (master thesis)