User-tailored recommendations

Abstract

A facility for producing an item recommendation for a selected user is described. The facility accesses an information resource that, for each of a number of buckets that each correspond to a different collection of personal information, identifies users for a members of the bucket into each of whom the entire collection of personal information applies. The facility selects a bucket among the plurality of buckets of which the selected user is a member. The facility accesses a number of item ratings that were each contributed by a member of the selected bucket other than the selected user. For each item rated among the accessed item ratings, the facility aggregates the ratings of the item. On the basis of items' aggregated ratings, the facility selects one or more rated items for recommendation to the selected user.

We claim: 
     
         1 . A computer-readable hardware device storing a data structure for making user-tailored recommendations, the data structure comprising:
 a tree-structure comprising a plurality of entries, each entry corresponding to a different collection of at least two pieces of personal information, each entry identifying users to each of whom the entire collection of personal information of the entry applies, and a number of item ratings contributed by the users, such that:
 evaluations by a first set of members of a bucket are predictive of an experience that a second set of members of the bucket will have when the second set of members make a same choice as the first set of members, and 
 for a selected user identified by a selected entry, the contents of the selected entry are usable to recommend items to the selected user that have been rated by users identified by the selected entry other than the selected user. 
   
     
     
         2 . The computer-readable hardware device of  claim 1  wherein the data structure further comprises, for each of the plurality of entries, item ratings contributed by users identified by the entry. 
     
     
         3 . The computer-readable hardware device of  claim 1  wherein the data structure is a decision tree having a root in which each entry is a node connected by an edge to its parent, and wherein each entry corresponds to a proper superset of the collection of personal information to which its parent corresponds. 
     
     
         4 . The computer-readable hardware device of  claim 3  wherein, for each of at least one distinguished entry, the portion of the distinguished entry's collection of personal information not among the distinguished entry's parent's collection of personal information has been determined to be of less predictive value than the portion of the distinguished entry's parent's collection of personal information not among the distinguished entry's grandparent's collection of personal information. 
     
     
         5 . The computer-readable hardware device of  claim 3  wherein, for each of at least one distinguished entry, the portion of the distinguished entry's collection of personal information not among the distinguished entry's parent's collection of personal information comprises demographic information, personality type information, preference information, subjectivity measure information, information about actions performed by users, information about at least three regions on a gender continuum, or any combination thereof. 
     
     
         6 . The computer-readable hardware device of  claim 1  wherein, for each of the plurality of entries, the entry's corresponding collection of personal information comprises at least one observed user behavior. 
     
     
         7 . The computer-readable hardware device of  claim 1  wherein, for each of the plurality of entries, the entry's corresponding collection of personal information comprises at least one item of user demographic information. 
     
     
         8 . A method in a computing system for organizing a plurality of users, comprising:
 for each of the plurality of users, posing a plurality of questions and receiving responses thereto;   establishing a tree-structure including a plurality of buckets each representing a different combination of particular responses to particular questions among the plurality of questions; and   assigning each user to a bucket representing a distinct combination of responses to questions that apply to all members of the bucket,
 wherein evaluations by a first set of members of the bucket are predictive of an experience that a second set of members of the bucket will have when the second set of members make a same choice as the first set of members, 
   such that the assignments of users to buckets are usable to generate item recommendations tailored to any particular user.   
     
     
         9 . The method of  claim 8 , further comprising, upon receiving a response to question among the plurality of questions from a distinguished user among the plurality of users, awarding a reward to the distinguished user. 
     
     
         10 . The method of  claim 9  wherein the reward awarded to the distinguished user is a virtual reward. 
     
     
         11 . The method of  claim 9  wherein the reward awarded to the distinguished user is one of a plurality of badges. 
     
     
         12 . The method of  claim 9 , further comprising, in response to an action by the distinguished user selected from among (1) making a referral of the facility to another prospective user, (2) using the facility at at least a threshold frequency, (3) using the facility with respect to a particular geographic location, and (4) discovering new features of the facility, awarding an additional award to the distinguished user. 
     
     
         13 . The method of  claim 8 , further comprising:
 for a selected user, using responses to posed questions received from the selected to user to infer for the selected user a response to a question for which no response was received from the selected user,   wherein assigning each user to a bucket assigns a selected user to a bucket representing a combination of responses to questions received from the selected user and the response inferred for the selected user.   
     
     
         14 . The method of  claim 8 , further comprising:
 for a selected user, where a first response is received from the selected user for a selected question, using actions taken by the selected to user as a basis for imputing to the selected user a second response to the selected question that is different from the first response,   wherein assigning each user to a bucket assigns a selected user to a bucket representing a combination of responses to questions for the selected user that include the second response to the selected question and excludes the first response to the selected question.   
     
     
         15 . The method of  claim 8 , further comprising organizing the established buckets into a first decision tree, and wherein the first decision tree is used in assigning each user to a bucket. 
     
     
         16 . The method of  claim 8 , further comprising organizing the established buckets into a first decision tree, and wherein the first decision tree is used in assigning each user to a bucket, the method further comprising:
 during a first period of time:
 generating a plurality of first item recommendations for a set of users among the plurality of users using the assignments of users to buckets performed using the first decision tree; 
 causing the plurality of first item recommendations to be presented to the users for which they were generated; 
   for each of one or more alternate decision trees that are distinct from the first decision tree, generating a plurality of second item recommendations for the set of users;   identifying an alternate decision tree whose second item recommendations have proven more apt than the first item recommendations;   during a second period of time following the first period of time:
 generating a plurality of third item recommendations for a group of users among the plurality of users using the assignments of users to buckets performed using the identified alternate decision tree; and 
 causing the plurality of third item recommendations to be presented to the users for which they were generated. 
   
     
     
         17 . A computer-readable hardware device having contents adapted to cause a computing system to perform a method for organizing a plurality of users, the method comprising:
 for each of the plurality of users, posing a plurality of questions and receiving responses thereto;   establishing a tree-structure including a plurality of buckets each representing a different combination of particular responses to particular questions among the plurality of questions; and   assigning each user to a bucket representing a distinct combination of responses to questions that apply to all members of the bucket,
 wherein evaluations by a first set of members of the bucket are predictive of an experience that a second set of members of the bucket will have when the second set of members make a same choice as the first set of members, 
   such that the assignments of users to buckets are usable to generate item recommendations tailored to any particular user.   
     
     
         18 . The computer-readable hardware device of  claim 17 , wherein the method further comprises:
 upon receiving a response to question among the plurality of questions from a distinguished user among the plurality of users, awarding a reward to the distinguished user,
 wherein the reward awarded to the distinguished user is a virtual reward; and 
   in response to an action by the distinguished user selected from among (1) making a referral of the facility to another prospective user, (2) using the facility at at least a threshold frequency, (3) using the facility with respect to a particular geographic location, and (4) discovering new features of the facility, awarding an additional award to the distinguished user.   
     
     
         19 . The computer-readable hardware device of  claim 17 , wherein the method further comprises:
 for a selected user, using responses to posed questions received from the selected to user to infer for the selected user a response to a question for which no response was received from the selected user,   wherein assigning each user to a bucket assigns a selected user to a bucket representing a combination of responses to questions received from the selected user and the response inferred for the selected user.   
     
     
         20 . The computer-readable hardware device of  claim 17 , wherein the method further comprises:
 for a selected user, where a first response is received from the selected user for a selected question, using actions taken by the selected to user as a basis for imputing to the selected user a second response to the selected question that is different from the first response,   wherein assigning each user to a bucket assigns a selected user to a bucket representing a combination of responses to questions for the selected user that include the second response to the selected question and excludes the first response to the selected question.