Managing feature distribution
A method of managing feature distribution in content of a content catalog for a content distribution system having a plurality of users is provided. The method comprises taking corrective action responsive to determining that a number of content items in a content catalog tagged with a feature exceeds a threshold.
1 . A method of managing feature distribution in content of a content catalog for a content distribution system having a plurality of users, the method comprising determining a number of content items in the content catalog tagged with a feature; and taking corrective action responsive to determining that the number of content items in the content catalog tagged with the feature exceeds a threshold, wherein taking corrective action comprises ignoring the feature in a content scoring function, and wherein determining comprises:
determining that a first number of tagged content items is greater than a second number of tagged content items, the first number of tagged content items tagged with a first feature and the second number of tagged content items tagged with a second feature.
2 . The method of claim 1 , wherein the feature is present in a user profile.
3 . The method of claim 1 , further comprising determining the one or more features in the user profile stored in a second storage resource, wherein the user profile is also stored in a first storage resource, the first storage resource is less rapidly readable than the second storage resource.
4 . The method of claim 3 , wherein taking corrective action comprises modifying a weight of a user feature vector of the user profile stored in the second storage resource, the user feature vector corresponding to the feature.
5 . The method of claim 1 , wherein determining comprises:
determining a distribution of the feature in the tagged content items and untagged content items; and determining if the number of tagged content items in the distribution exceeds the threshold.
6 . The method of claim 1 , wherein taking corrective action comprises: taking corrective action in in response to determining that the first number of tagged content items is greater than the second number of tagged content items.
7 . The method of claim 6 , wherein taking corrective action comprises:
de-tagging the first feature from the first number of tagged content items and maintaining the second feature from the second number of tagged content items.
8 . The method of claim 1 , further comprising, prior to the determining, tagging one or more content items with the feature.
9 . The method of claim 1 , wherein determining the number of content items tagged with the feature occurs in response to a trigger event.
10 . The method of claim 1 , wherein taking corrective action comprises at least one of: de-tagging the feature from the content items, ignoring the feature in a content scoring function, or communicating the feature to a content recommendation engine (CRE), the CRE adapted to generate one or more content recommendations responsive to one or more content requests from a user of a plurality of users.
11 . The method of claim 1 , wherein taking corrective action comprises notifying a client that the threshold has been exceeded.
12 . The method of claim 1 , wherein taking corrective action comprises removing the feature from the tagged content items.
13 . The method of claim 12 , wherein removing the feature comprises de-tagging the feature from the tagged content items.
14 . The method of claim 1 , wherein the threshold is a percentage of a total number of content items.
15 . The method of claim 1 , wherein taking corrective action comprises modifying a scoring function stored in a second storage resource, the scoring function also stored in a first storage resource.
16 . The method of claim 15 , wherein modifying the scoring function comprises modifying a weight of a user feature vector.
17 . A non-transitory computer-readable medium having computer program code stored thereon, the program code executable by a processor to perform the method of claim 1 .
18 . A content recommendation system comprising a content recommendation engine (CRE) for generating one or more content recommendations responsive
to one or more content requests from a user of a plurality of users and a feature management module, the feature management module adapted to:
determine that a number of tagged content items exceeds a threshold; and
take corrective action in response to determining the number of tagged content items exceeds the threshold, wherein the feature management module is adapted to ignore the feature in a content scoring function, and determine that a first number of tagged content items is greater than a second number of tagged content items, the first number of tagged content items tagged with a first feature and the second number of tagged content items tagged with a second feature.