Augmenting Points of Interest Recommendations with Music
msra(2009)
摘要
Recommender systems are personalized information search and decision support tools that can selectively retrieve from large information sources personalized set of items (e.g. news, CD descriptions, or services) that suit the particular preferences of a user. Recommender systems are normally limited in providing recommendations for just one type of information items. But, recently the notion of crossdomain recommender systems has been introduced to denote applications that can reuse knowledge about the users, which is derived in one domain, to provide recommendations in another different domain. A particular kind of cross domain recommendation task consists of selecting simultaneously two items in two different domains and recommending them together because they fit the user preferences and also they fit well together. In this work we show that given a personalized recommendation of points of interests (POIs), the user satisfaction for the itinerary composed by these POIs can be increased by enriching the itinerary presentation with music tracks that match the user’s profile and are coherent with the POIs. We present the results of an online experiment where alternative approaches for matching POIs and music, based on tagging and text matching, have been tested with users.
更多查看译文
关键词
recommender system,mobile service,travel planning,consumer behavior
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络