TFM 2016/17 Exploiting subsequence matching in Recommender Systems
Pablo Sánchez Pérez · TFM · 2018
Recommender Systems (RS) are software tools that allow users nding the items andinformation they need in a simple and direct way. These items are related to the specicdomain of the recommender system, although movies, books, and music are some of themost studied domains in the scientic community. In order to predict the most interestingitems for each user in the system, these methods analyze the tastes and interests of theusers to make personalized recommendations [36].Although the rise of the Internet dates back to the early 1970s, the research in thesesystems has taken place especially in the last 20 years, due to the global spread of informationand communication technologies. The antecedents of these systems can be foundin the early 1990s, in the Tapestry [20] and Grouplens [35] projects, co-occurring withthe rise of the Internet. However, they are especially relevant today as they have nowbecome essential to lter the large amount of data available in the cloud. A large numberof important companies oering online services make use of recommendation algorithmsto expand their economic activity and improve the user experience. Some examples areAmazon (online store), Youtube (videos) or Net ix (streaming audiovisual content). Thislast company became very popular in 2006 for a three-year contest with a prize of 1 milliondollars to the research group who managed to improve its prediction algorithm by 10% [31].The team \BellKor's Pragmatic Chaos ended up winning this contest and putting thesetechnologies in the spotlight.