| Summary: | 碩士 === 中華大學 === 資訊工程學系碩士班 === 100 === The traditional recommendation system is mostly done by similarity discriminate
when the data is sufficient, and use the high correlation as recommendation items. It will become very difficult to produce accurate advertisements recommendation when the data is not enough. Therefore, some scholars explore the Cold-starting Problem: To resolve the recommendation effectively when too small amount of data is provided.
In this paper, we address the Cold-starting Problem by using location and time as the initial condition to produce accurate advertisements recommendation. Also, in the paper, we use fuzzy theory to obtain the objective value by the classification results with time, location and advertising attribution calculation, we call it as LTRS (Location-Time based Recommendation System). From the experimental results, the proposed LTRS can improve the accuracy of advertising, not only to improve the traditional recommendation system but also improve the recommended advertising accuracy of the Cold-starting Problem.
|
|---|
