Keyword-Based TV Program Recommendation
msra(2011)
摘要
Notwithstanding the success of collaborative ltering algorithms for item recommendation there are still situations in which there is a need for content-based recommendation, especially in new-item scenarios, e.g. in streaming broad- casting. Since video content is hard to analyze we use documents describing the videos to com- pute item similarities. We do not use the de- scriptions directly, but use their keywords as an intermediate level of representation. We argue that a nearest-neighbor approach relying on un- restricted keywords deserves a special denition of similarity that also takes word similarities into account. We dene such a similarity mea- sure as a divergence measure of smoothed key- word distributions. The smoothing is done on the basis of co-occurrence probabilities of the present keywords. Thus co-occurrence similar- ity of words is also taken into account. We have evaluated keyboard-based recommenda- tions with a dataset collected by the BBC and on a subset of the MovieLens dataset aug- mented with plot descriptions from IMDB. Our main conclusions are (1) that keyword-based rating predictions can be very eective for some types of items, and (2) that rating predictions are signicantly better if we do not only take into account the overlap of keywords between two documents, but also the mutual similarities between keywords.
更多查看译文
关键词
nearest neighbor
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络