Systems and methods for grouping users based on metadata tag relevance ratings

Abstract

A content item may be associated with metadata comprising one or more tags. A user may indicate a relevance rating associated with a tag. The relevance rating may indicate whether the user feels the tag is relevant to the associated content item. Using a plurality of user-provided relevance ratings, a tag relevance model may be established. A tag relevance model may comprise a weighted or un-weighted average and/or median relevance rating of the tag and/or a cohesiveness of the of the relevance ratings. Rating cohesiveness may be used to identify controversial tags. Tag relevance information may be used to order search results. Tag ratings may also be used to aggregate users into groups comprising users having a similar point of view relative to one or more tag ratings. In addition, users may be grouped according to content access and/or tags rated regardless of the relevance rating applied.

1 . A method for grouping users in a user community, comprising:
 receiving a plurality of relevance ratings at a network-accessible service, each relevance rating submitted by a respective user in the user community and rating the relevance of a respective user-submitted metadata tag to a content item;   associating each relevance rating with a respective user of the user community that submitted the relevance rating;   comparing, by a computing device, the relevance ratings of users in the user community to identify users that have submitted similar relevance ratings;   selecting two or more users of the user community for inclusion in a group based upon identified similarities in the relevance ratings submitted by the users; and   providing for displaying indications of one or more users included in the group on a display.   
     
     
         2 . The method of  claim 1 , further comprising:
 calculating a distribution of relevance ratings of a metadata tag; and   identifying similarities between user relevance ratings of the metadata tag using the distribution.   
     
     
         3 . The method of  claim 2 , further comprising:
 determining a rating threshold using the distribution; and   identifying a similarity between user relevance ratings of the metadata tag when the relevance ratings are within the rating threshold.   
     
     
         4 . The method of  claim 3 , wherein the rating threshold is a standard deviation of the distribution. 
     
     
         5 . The method of  claim 1 , further comprising:
 comparing relevance ratings of metadata tags submitted by a first user of the user community to relevance ratings of other users of the user community; and   selecting two or more users of the user community for inclusion in the group based upon similarities between the relevance ratings submitted by the first user and relevance ratings submitted by the other users.   
     
     
         6 . The method of  claim 5 , further comprising determining a correlation between the relevance ratings submitted by the first user and relevance ratings submitted by the selected users. 
     
     
         7 . The method of  claim 1 , further comprising:
 determining a similarity between relevance ratings submitted by a first user of the user community and relevance ratings submitted by a second user of the user community, by
 identifying a plurality of metadata tags having relevance ratings submitted by the first user and the second user, 
 for each of the identified metadata tags, comparing a relevance rating submitted by the first user to a relevance rating submitted by the second user, and 
 determining the similarity between the relevance ratings of the first user and the second user based upon the comparisons. 
   
     
     
         8 . The method of  claim 7 , further comprising:
 calculating a distribution of the relevance ratings of each of the metadata tags, the distribution comprising a mean and a deviation,   wherein comparing a relevance rating submitted by the first user to a relevance rating submitted by the second user comprises comparing the relevance rating of the first user and the relevance rating of the second user to the mean and the deviation.   
     
     
         9 . The method of  claim 1 , further comprising selecting the two or more users of the user community for inclusion in the group by identifying similarities in the content items accessed by the users in the user community. 
     
     
         10 . The method of  claim 9 , further comprising
 calculating a cross-over between a first user and a second user as a ratio of content items having metadata tags relevance ratings submitted by both the first user and the second user to content items having metadata tags rated by either the first user or the second user.   
     
     
         11 . A non-transitory computer-readable storage medium comprising instructions to cause a computing device to perform a method for grouping users in a user community, the method comprising:
 receiving a plurality of relevance ratings, each relevance rating submitted by a respective user in the user community and rating the relevance of a respective metadata tag to a content item;   associating each relevance rating with a respective user of the user community that submitted the relevance rating;   comparing the relevance ratings of users in the user community to identify users that have submitted similar relevance ratings;   selecting two or more users of the user community for inclusion in a group based upon identified similarities in the relevance ratings submitted by the users; and   providing for displaying indications of the one or more users included in to the group to a user.   
     
     
         12 . The non-transitory computer-readable storage medium of  claim 11 , the method further comprising:
 calculating a distribution of the relevance ratings of each of the metadata tags; and   identifying similarities in the relevance ratings of the selected users using the relevance rating distributions.   
     
     
         13 . The non-transitory computer-readable storage medium of  claim 12 , wherein each distribution comprises a relevance rating mean and a relevance rating deviation, the method further comprising:
 determining a similarity between relevance ratings of a first user and a second user when the relevance ratings of the first user and the second user both differ from the mean of the distribution by at least the deviation of the distribution.   
     
     
         14 . The non-transitory computer-readable storage medium of  claim 11 , the method further comprising comparing relevance ratings submitted by a first user of the user community to relevance ratings submitted by other users of the user community; and
 selecting two or more users of the user community for inclusion in the group based upon similarities between the relevance ratings submitted by the first user and the relevance ratings submitted by the other users.   
     
     
         15 . The non-transitory computer-readable storage medium of  claim 11 , the method further comprising:
 determining a similarity between relevance ratings submitted by a first user of the user community and relevance ratings submitted by a second user of the user community, by
 identifying a plurality of metadata tags having relevance ratings submitted by the first user and the second user, 
 for each of the identified metadata tags, comparing a relevance rating submitted by the first user to a relevance rating submitted by the second user, and 
 determining the similarity between the relevance ratings of the first user and the second user based upon the comparisons. 
   
     
     
         16 . The non-transitory computer-readable storage medium of  claim 11 , the method further comprising selecting the two or more users of the user community for inclusion in the group by identifying similarities in content items accessed by the users in the user community. 
     
     
         17 . The non-transitory computer-readable storage medium of  claim 11 , the method further comprising:
 calculating a cross-over between a first user and a second user as a ratio of content items having metadata tags relevance ratings submitted by both the first user and the second user to content items having metadata tags rated by either the first user or the second user.   
     
     
         18 . A system for grouping users in a user community based upon relevance ratings of metadata tags, comprising:
 a computer-readable storage medium comprising a plurality of content items, each content item being associated with respective, user-submitted metadata tags, each metadata tag having a plurality of relevance ratings, each rating the relevance of the metadata tag to the associated content item, wherein each relevance rating is associated with a respective user in the user community that submitted the relevance rating; and   a server configured to calculate a distribution of the relevance ratings of each of the metadata tags, to compare the user-submitted relevance ratings of each of the metadata tags using the respective metadata tag distributions to identify similarities between the relevance ratings of the users in the user community, and to select two or more users of the user community for inclusion in a group based upon identified similarities in the relevance ratings of submitted by the selected users.   
     
     
         19 . The system of  claim 18 , wherein each distribution comprises a deviation of the relevance ratings of a respective metadata tag, and wherein a first relevance rating is similar to a second relevance rating when the first and the second relevance ratings differ by less than the deviation. 
     
     
         20 . The system of  claim 18 , wherein each distribution comprises a deviation of the relevance ratings of a respective metadata tag, and wherein a first relevance rating is similar to a second relevance rating when both the first and the second relevance ratings differ from the mean by at least the deviation. 
     
     
         21 . The system of  claim 18 , wherein the server is configured to display indications of the selected users included in the group to a user.