Recommending an item based on attention of user group to item

Abstract

A recommendation method includes obtaining a candidate item to be recommended to a social network user, the social network user belonging to a group of users in an online social network, and determining attention of the group to the candidate item in the online social network. The determined attention of the group to the candidate item is based on an importance weight of each user in the group and based on attention of each user in the group to the candidate item. The method further includes determining, according to the determined attention of the group, whether to recommend the candidate item to the social network user through the online social network.

What is claimed is: 
     
         1 . A recommendation method, comprising:
 obtaining a candidate item to be recommended to a social network user, the social network user belonging to a group of users in an online social network;   determining attention of the group to the candidate item in the online social network, wherein the determined attention of the group to the candidate item is based on an importance weight of each user in the group and based on attention of each user in the group to the candidate item; and   determining, according to the determined attention of the group, whether to recommend the candidate item to the social network user through the online social network.   
     
     
         2 . The method according to  claim 1 , wherein the importance weight of each user in the group is based on an activity level of the respective user in the online social network. 
     
     
         3 . The method according to  claim 2 , wherein the determining the attention of the group comprises:
 inputting a vector representation of the group and a vector representation of the candidate item to a first attention model, the first attention model being pre-trained with an attention parameter of the group to the candidate item; and   obtaining attention information of the group to the candidate item that is outputted by the first attention model, the attention information of the group indicating an attention of the group to the candidate item that is generated and outputted by the first attention model according to the attention parameter of the group to the candidate item determined according to the vector representation of the group and the vector representation of the candidate item.   
     
     
         4 . The method according to  claim 3 , wherein the first attention model is pre-trained with the importance weight of each user in the group; and
 the attention information of the group to the candidate item that is outputted by the first attention model is obtained by further performing weighting processing using the importance weight of each user in the group and the attention parameter of the group to the candidate item.   
     
     
         5 . The method according to  claim 2 , wherein
 a second user of the online social network has a friend relationship with the social network user, and   the method further comprises
 determining an attention of the second user to the candidate item, 
 determining, based on the determined attention of the second user and on the determined attention of the group to the candidate item, a comprehensive attention of the second user and the group to the candidate item, and 
 determining whether to recommend the candidate item to the social network user based on the determined comprehensive attention. 
   
     
     
         6 . The method according to  claim 5 , wherein the determining the attention of the second user further comprises:
 inputting a vector representation of the second user and a vector representation of the candidate item to a second attention model, the second attention model being pre-trained with an attention parameter of each friend of the social network user on the online social network to the candidate item; and   obtaining attention information of the second user to the candidate item that is outputted by the second attention model, the attention information of the second user indicating an attention of the second user to the candidate item that is generated by the second attention model according to the attention parameter of the second user to the candidate item determined according to the vector representation of the second user and the vector representation of the candidate item.   
     
     
         7 . The method according to  claim 5 , wherein
 the determining the comprehensive attention includes:
 performing normalization processing on the attention of the second user and the group to the candidate item; and 
 obtaining the comprehensive attention as attentions of social objects having different types of relationships to the social network user to the candidate item; and 
   the determining whether to recommend includes:
 determining a recommendation index for recommending the candidate item based on the comprehensive attention to the candidate item; and 
 determining, according to the recommendation index, whether to recommend the candidate item to the social network user. 
   
     
     
         8 . A recommendation apparatus, comprising:
 processing circuitry configured to:
 obtain a candidate item to be recommended to a social network user, the social network user belonging to a group of users in an online social network; 
 determine attention of the group to the candidate item in the online social network, wherein the determined attention of the group to the candidate item is based on an importance weight of each user in the group and based on attention of each user in the group to the candidate item; and 
 determine, according to the determined attention of the group, whether to recommend the candidate item to the social network user through the online social network. 
   
     
     
         9 . The recommendation apparatus according to  claim 8 , wherein the importance weight of each user in the group is based on an activity level of the respective user in the online social network. 
     
     
         10 . The recommendation apparatus according to  claim 9 , wherein the processing circuitry is configured to:
 input a vector representation of the group and a vector representation of the candidate item to a first attention model, the first attention model being pre-trained with an attention parameter of the group to the candidate item; and   obtain attention information of the group to the candidate item that is outputted by the first attention model, the attention information of the group indicating an attention of the group to the candidate item that is generated and outputted by the first attention model according to the attention parameter of the group to the candidate item determined according to the vector representation of the group and the vector representation of the candidate item.   
     
     
         11 . The recommendation apparatus according to  claim 10 , wherein the first attention model is pre-trained with the importance weight of each user in the group; and
 the attention information of the group to the candidate item that is outputted by the first attention model is obtained by further performing weighting processing using the importance weight of each user in the group and the attention parameter of the group to the candidate item.   
     
     
         12 . The recommendation apparatus according to  claim 9 , wherein a second user of the online social network has a friend relationship with the social network user, and
 the processing circuitry is further configured to
 determine an attention of the second user to the candidate item, 
 determine, based on the determined attention of the second user and on the determined attention of the group to the candidate item, a comprehensive attention of the second user and the group to the candidate item, and 
 determine whether to recommend the candidate item to the social network user based on the determined comprehensive attention. 
   
     
     
         13 . The recommendation apparatus according to  claim 12 , wherein the processing circuitry is configured to:
 input a vector representation of the second user and a vector representation of the candidate item to a second attention model, the second attention model being pre-trained with an attention parameter of each friend of the social network user on the online social network to the candidate item; and   obtain attention information of the second user to the candidate item that is outputted by the second attention model, the attention information of the second user indicating an attention of the second user to the candidate item that is generated by the second attention model according to the attention parameter of the second user to the candidate item determined according to the vector representation of the second user and the vector representation of the candidate item.   
     
     
         14 . The recommendation apparatus according to  claim 12 , wherein the processing circuitry is configured to:
 perform normalization processing on the attention of the second user and the group to the candidate item;   obtain the comprehensive attention as attentions of social objects having different types of relationships to the social network user to the candidate item;   determine a recommendation index for recommending the candidate item based on the comprehensive attention to the candidate item; and   determine, according to the recommendation index, whether to recommend the candidate item to the social network user.   
     
     
         15 . A non-transitory computer-readable storage medium storing instructions which when executed by at least one processor cause the at least one processor to perform a method comprising:
 obtaining a candidate item to be recommended to a social network user, the social network user belonging to a group of users in an online social network;   determining attention of the group to the candidate item in the online social network, wherein the determined attention of the group to the candidate item is based on an importance weight of each user in the group and based on attention of each user in the group to the candidate item;   determining, according to the determined attention of the group, whether to recommend the candidate item to the social network user through the online social network.   
     
     
         16 . The non-transitory computer-readable storage medium according to  claim 15 , wherein importance weight of each user in the group is based on an activity level of the respective user in the online social network. 
     
     
         17 . The non-transitory computer-readable storage medium according to  claim 16 , wherein the determining the attention of the group comprises:
 inputting a vector representation of the group and a vector representation of the candidate item to a first attention model, the first attention model being pre-trained with an attention parameter of the group to the candidate item; and   obtaining attention information of the group to the candidate item that is outputted by the first attention model, the attention information of the group indicating an attention of the group to the candidate item that is generated and outputted by the first attention model according to the attention parameter of the group to the candidate item determined according to the vector representation of the group and the vector representation of the candidate item.   
     
     
         18 . The non-transitory computer-readable storage medium according to  claim 17 , wherein the first attention model is pre-trained with the importance weight of each user in the group; and
 the attention information of the group to the candidate item that is outputted by the first attention model is obtained by further performing weighting processing using the importance weight of each user in the group and the attention parameter of the group to the candidate item.   
     
     
         19 . The non-transitory computer-readable storage medium according to  claim 16 , wherein a second user of the online social network has a friend relationship with the social network user, and
 the method further comprises
 determining an attention of the second user to the candidate item,
 determining, based on the determined attention of the second user and on the determined attention of the group to the candidate item, a comprehensive attention of the second user and the group to the candidate item, and 
 
   
       determining whether to recommend the candidate item to the social network user based on the determined comprehensive attention. 
     
     
         20 . The non-transitory computer-readable storage medium according to  claim 19 , wherein the determining the attention of the second user further comprises:
 inputting a vector representation of the second user and a vector representation of the candidate item to a second attention model, the second attention model being pre-trained with an attention parameter of each friend of the social network user on the online social network to the candidate item; and   obtaining attention information of the second user to the candidate item that is outputted by the second attention model, the attention information of the second user indicating an attention of the second user to the candidate item that is generated by the second attention model according to the attention parameter of the second user to the candidate item determined according to the vector representation of the second user and the vector representation of the candidate item.