Personalized content recommendations based on consumption periodicity
Aspects described herein provide techniques for providing content recommendations such as, for example, recommendations for video content provided by a cable network. The content recommendation techniques described herein determine the extent to which previously consumed content should contribute to content considered for recommendation to a user based on the consumption time of the previously consumed content. The consumption habits of users have been observed to occur at regular intervals. These intervals are employed to adjust contribution factors for the previously consumed content when making content recommendations.
1. A method comprising:
determining, by a computing device, a previously outputted content item that has been previously output at a first time;
determining, by the computing device, a content item to be considered for recommendation;
determining, by the computing device, a value indicating a time difference between a reference time associated with the content item and the first time associated with the previously outputted content item;
determining, by the computing device, a contribution factor for the previously outputted content item based on proximity of the value indicating the time difference to each of a first multiple of a first time period and a second multiple of a second time period;
determining, by the computing device, a recommendation score for the content item based, at least in part, on the contribution factor; and
outputting a signal comprising a recommendation based on the determined recommendation score for the content item.
2. The method of claim 1 , wherein:
the first time period is associated with a daily time period; and
the second time period is associated with a weekly time period.
3. The method of claim 2 , wherein determining the contribution factor comprises:
biasing the contribution factor such that the contribution factor is relatively higher based on determining that the value indicating the time difference is relatively closer to the multiple of the daily time period, or
biasing the contribution factor such that the contribution factor is relatively lower based on determining that the value indicating the time difference is relatively further from the multiple of the daily time period.
4. The method of claim 3 , wherein determining the contribution factor further comprises: further biasing the contribution factor such that the contribution factor is relatively higher based on determining that the multiple of the daily time period is also a multiple of the weekly time period, or
biasing the contribution factor such that the contribution factor is relatively lower based on determining that the multiple of the daily time period is not also a multiple of the weekly time period.
5. The method of claim 4 , wherein determining the contribution factor comprises applying a decay factor to the contribution factor that decreases the contribution factor.
6. The method of claim 5 , wherein the decay factor is inversely proportional to the value indicating the time difference.
7. The method of claim 1 , wherein:
the recommendation comprises an indication of the content item based on determining that the recommendation score for the content item is above a score threshold; or
the recommendation does not comprise the indication of the content item based on determining that the recommendation score for the content item is below the score threshold.
8. The method of claim 1 , wherein:
the previously outputted content item is one of a plurality of previously outputted content items;
the contribution factor is one of a plurality of contribution factors determined respectively for each one of the plurality of previously outputted content items; and
determining the recommendation score for the content item comprises aggregating the plurality of contribution factors.
9. The method of claim 1 , wherein determining the recommendation score for the content item comprises determining a similarity score based on a total number of users that previously requested output of both the content item and the previously outputted content item, the similarity score representing a similarity between the content item and the previously outputted content item.
10. The method of claim 1 , wherein the recommendation is a list of recommended content items that is sorted based on the recommendation score.
11. The method of claim 1 , wherein the determining the previously outputted content item comprises determining the previously outputted content item from a user history stored at a data store.
12. The method of claim 1 , wherein the determining the content item to be considered for recommendation comprises determining the content item as part of an automatic recommendation task.
13. The method of claim 1 , wherein determining content item to be considered for recommendation is in response to receipt of a recommendation request.
14. The method of claim 1 , wherein the reference time is a time at which the content item is selected.
15. The method of claim 1 , wherein:
the previously outputted content item is a previously outputted video content item;
the content item to be considered for recommendation is a video content item; and
the first time is a time at which the previously outputted video content item was output for viewing.
16. The method of claim 1 , wherein:
the content item to be considered for recommendation is a prescheduled content item; and
the reference time is a time at which the prescheduled content item is scheduled for broadcast.
17. The method of claim 1 , wherein:
the content item to be considered for recommendation is a recorded content item; and
the reference time is a time at which a set of recorded content items available for playback is requested.
18. The method of claim 1 , wherein:
the content item to be considered for recommendation is an on-demand content item; and
the reference time is a time at which a set of on-demand content items available for selection is requested.
19. The method of claim 1 , wherein:
the previously outputted content item is a previously listened to audio content item;
the content item to be considered for recommendation is an audio content item; and
the first time is a time at which the previously listened to audio content item was output for listening.
20. A method comprising:
determining, by a computing device, a value indicating a time difference between a first time at which first content item was requested and a second time associated with a plurality of second content items to be considered for recommendation;
determining, by the computing device, a contribution factor for the first content item based on proximity of the value indicating the time difference to each of a multiple of a daily time period and a multiple of a weekly time period;
generating, by the computing device, at least one content recommendation for at least one of the plurality of second content items based, at least in part, on the contribution factor for the first content item; and
outputting a signal comprising the at least one content recommendation.
21. A method comprising:
determining, by a computing device, a reference time at which a user action was detected at a gateway device;
retrieving, by the computing device, a history associated with the gateway device wherein the history indicates a previously requested content item requested at a first time;
determining, by the computing device, a recommendation score for a content item based on proximity of a value indicating a time difference between the reference time and the first time to each of a first multiple of a first time period and a second multiple of a second time period; and
outputting, to a display device, a signal comprising a recommendation based on the recommendation score determined for the content item.