Personalized content recommendations based on consumption periodicity
摘要
說明書
For recommendation techniques that are based on content similarity, the similarity may be a value quantifying the similarity between candidate content and previously consumed content and expressed by the variable, S. The contribution factor may function to weight the contribution previously consumed content makes to a recommendation score and may be expressed by the variable, W. As described above, the contribution factor, W, in this example, is based on the proximity of the time difference to an integer multiple of 24 hours. The value of the contribution factor may be highest for programs where the time difference is an integer multiple of 168 hours (a week) and 24 hours (a day), and may be progressively lower as the time difference departs from the 24 hour and 168 hour multiples.
The recommendation score, R, for candidate content, C, may be expressed as a sum of the respective contribution factors, W, and the similarity scores, S, for pairings of the candidate content, C, and each previously-consumed content, V1 . . . Vn, in the consumption history. The recommendation score, R, may be expressed as the following: R=1+W1S1+W2S2+ . . . +WnSn
