Personalized content recommendations based on consumption periodicity
摘要
說明書
In the example formula above, S1 represents the similarity score for the pairing of content, C, and previously viewed content, V1—S(C, V1); S2 represents the similarity score for the pairing of content, C, and previously viewed content, V2—S(C, V2); and Sn represents the similarity score for the pairing of content, C, and previously viewed content, V2—S(C, V,2). Other formulas for obtaining the recommendation score, R, may be selectively employed. In some example implementations, obtaining the recommendation score may include determining a popularity score for the candidate content and incorporating the popularity score in the formula for the recommendation score. For convenience the function for obtaining a popularity score, O, for candidate content, C, may be expressed as O(C). The popularity score, O, may be, for example, the percentage of all users that consumed the content, C. In these other implementations, a recommendations score, R′, may be expressed as: R′=O×(1+W1S1+W2S2+ . . . +WnSn)
The pseudocode below may be employed to obtain a contribution factor for previously consumed video content based on the relationship between content similarity and consumption time observed from the graph 400 in
