Search result ranking method and system

Abstract

Search result ranking includes recording user action information on displayed objects in search results obtained using one or more query words, upon receiving a switch-page request or switch-screen request, determining a commonality level of one or more attribute characteristics in a set of objects subjected to user actions, the determining of the commonality level being based on the user action information on the displayed objects, selecting attribute characteristics that comply with predetermined requirements to serve as reference norms for ranking objects that are to be displayed or ranked, the selecting of the attribute characteristics being based on the commonality level, and adjusting rank of objects that are to be displayed or to be ranked and whose attribute characteristics comply with the reference norms.

What is claimed is: 
     
         1 . (canceled) 
     
     
         2 . A method, comprising:
 recording user action information on displayed objects in search results obtained using one or more query words, wherein the user action information includes a sequence in which a user browsed or clicked the objects;   upon receiving a switch-page request or a switch-screen request, determining a commonality level of one or more attribute characteristics in objects subjected to user actions, wherein the determining of the commonality level is based on the user action information on the displayed objects;   selecting attribute characteristics that comply with predetermined requirements to serve as reference norms for ranking objects that are to be displayed or ranked, wherein the selecting of the attribute characteristics is based on the commonality level;   calculating, based on the user action information, ranking scores for the displayed objects; and   adjusting, based on the ranking scores, rank of the objects that are to be displayed or to be ranked, and whose attribute characteristics comply with the reference norms.   
     
     
         3 . The method as described in  claim 2 , wherein the action information includes objects of user action, information on relative positions of user action objects in the search results, or both. 
     
     
         4 . The method as described in  claim 2 , wherein the objects are products or product information. 
     
     
         5 . The method as described in  claim 2 , wherein the determining of the commonality level of the one or more attribute characteristics in the objects subjected to user actions based on the user action information on displayed objects comprises:
 acquiring the attribute characteristics of the displayed objects;   calculating a commonality level of an attribute characteristic of a displayed object that has been clicked based on the recorded user action information on the displayed objects; and   calculating a commonality level of an attribute characteristic of a displayed object that has not been clicked.   
     
     
         6 . The method as described in  claim 2 , wherein:
 the determining of the commonality level of the one or more attribute characteristics in the objects subjected to user actions based on the user action information on displayed objects comprises:
 acquiring the attribute characteristics of the displayed objects; 
 calculating a commonality level of an attribute characteristic of a displayed object that has been clicked based on the recorded user action information on the displayed objects; and 
 calculating a commonality level of an attribute characteristic of a displayed object that has not been clicked; and 
   the commonality level of the attribute characteristic of the displayed object that has been clicked or the commonality level of the attribute characteristic of the displayed object that has not been clicked corresponds to a ratio of a number of objects that have the same or similar attribute characteristics in displayed objects that have been clicked or displayed objects that have not been clicked to a total number of objects in the displayed objects that have been clicked or the displayed objects that have not been clicked.   
     
     
         7 . The method as described in  claim 2 , wherein:
 the determining of the commonality level of the one or more attribute characteristics in the objects subjected to user actions based on the user action information on displayed objects comprises:
 acquiring the attribute characteristics of the displayed objects; 
 calculating a commonality level of an attribute characteristic of a displayed object that has been clicked based on the recorded user action information on the displayed objects; and 
 calculating a commonality level of an attribute characteristic of a displayed object that has not been clicked; and 
   the selecting of the attribute characteristics that comply with the predetermined requirements to serve as the reference norms for ranking objects that are to be displayed or ranked based on the commonality level comprises:
 ranking various attribute characteristics in displayed objects that have been clicked and various attribute characteristics in displayed objects that have not been clicked in order of high to low commonality level, and selecting a predetermined quantity of top-ranked attribute characteristics to serve as the reference norms; or 
 regarding attribute characteristics with commonality levels greater than a threshold value as the reference norms. 
   
     
     
         8 . The method as described in  claim 2 , wherein:
 the determining of the commonality level of the one or more attribute characteristics in the objects subjected to user actions based on the user action information on displayed objects comprises:
 acquiring the attribute characteristics of the displayed objects; 
 calculating a commonality level of an attribute characteristic of a displayed object that has been clicked based on the recorded user action information on the displayed objects; and 
 calculating a commonality level of an attribute characteristic of a displayed object that has not been clicked; and 
   the selecting of the attribute characteristics that comply with the predetermined requirements to serve as the reference norms for ranking objects that are to be displayed or ranked based on the commonality level comprises:
 calculating differences in commonality levels of attribute characteristics of displayed objects that have been clicked and commonality levels of attribute characteristics of the displayed object that have not been clicked, ranking the various attribute characteristics in order of large to small differences in the commonality levels, and selecting a predetermined quantity of top-ranked attribute characteristics to serve as the reference norms; or 
 regarding attribute characteristics whose differences in commonality levels are greater than a set threshold value as the reference norms. 
   
     
     
         9 . The method as described in  claim 2 , wherein the objects of user actions are objects that were clicked from among the search results. 
     
     
         10 . The method as described in  claim 2 , wherein:
 the objects of user actions are objects that were clicked from among the search results; and   the adjusting of the rank of the objects that are to be displayed or to be ranked, and whose attribute characteristics comply with the reference norms comprises:
 calculating a commonality level of an attribute characteristic of an object that has been clicked based on the user action information on objects in displayed objects that have been clicked; 
 selecting attribute characteristics whose commonality levels are greater than a preset threshold value as reference norms; and 
 raising the rank of the objects that are to be displayed or to be ranked and whose attribute characteristics comply with the reference norms. 
   
     
     
         11 . The search result ranking method as described in  claim 2 , wherein:
 the objects of user actions are objects that were clicked from among the search results; and   the adjusting of the rank of the objects that are to be displayed or to be ranked and whose attribute characteristics comply with the reference norms comprises:
 calculating a commonality level of an attribute characteristic of an object that has been clicked based on the user action information on objects that have been clicked; 
 ranking various attribute characteristics on the objects that have been clicked and various attribute characteristics on objects that have not been clicked in order of highest to lowest commonality levels; 
 selecting a pre-established quantity of top-ranked attribute characteristics to serve as the reference norms; and 
 raising the rank of the objects that are to be displayed or to be ranked and whose attribute characteristics comply with the reference norms. 
   
     
     
         12 . The search result ranking method as described in  claim 2 , wherein the adjusting of the rank of the objects that are to be displayed or to be ranked and whose attribute characteristics comply with the reference norms comprises:
 calculating a commonality level of an attribute characteristic of an object that has not been clicked;   selecting attribute characteristics whose commonality levels are greater than a preset threshold value as the reference norms; and   lowering the rank of objects that are to be displayed or to be ranked and whose attribute characteristics comply with the reference norms.   
     
     
         13 . The search result ranking method as described in  claim 2 , wherein the adjusting of the rank of the objects that are to be displayed or to be ranked and whose attribute characteristics comply with the reference norms comprises:
 calculating a commonality level of an attribute characteristic of an object that has not been clicked based on the user action information on objects that have not been clicked;   ranking various attribute characteristics in the objects that have been clicked and various attribute characteristics in the objects that have not been clicked in order of highest to lowest commonality levels;   selecting a pre-established quantity of top-ranked attribute characteristics to serve as the reference norms; and   lowering ranks of objects that are to be displayed or to be ranked and whose attribute characteristics comply with the reference norms.   
     
     
         14 . The search result ranking method as described in  claim 2 , wherein:
 the adjusting of the ranks of the objects that are to be displayed or to be ranked, and whose attribute characteristics comply with the reference norms comprises re-ranking objects that are to be displayed or to be ranked according to the ranking scores; and   the calculating of the ranking score for each object that is to be displayed or to be ranked based on a corresponding reference norm comprises:
 assigning first weights according to the sequence in which displayed objects are clicked by the user; 
 calculating a weight of a reference norm based on the first weights of the objects that are among objects that have been click and that comply with the reference norms; 
 calculating comprehensive scores of degrees of influence of the reference norms on the object rankings based on the weights of the reference norms with which the attribute characteristics of the objects that are to be displayed or to be ranked comply; and 
 adjusting the ranking scores of the objects according to the comprehensive scores. 
   
     
     
         15 . A search result ranking device, comprising:
 a processor; and   a memory coupled with the processor, wherein the memory is configured to provide the processor with instructions which when executed cause the processor to:
 record user action information on displayed objects in search results obtained using one or more query words, wherein the user action information includes a sequence in which a user browsed or clicked the objects; 
 upon receiving a switch-page request or a switch-screen request, determine a commonality level of one or more attribute characteristics in objects subjected to user actions, wherein the determining of the commonality level is based on the user action information on the displayed objects; 
 select attribute characteristics that comply with predetermined requirements to serve as reference norms for ranking objects that are to be displayed or ranked, wherein the selecting of the attribute characteristics is based on the commonality level; 
 calculate, based on the user action information, ranking scores for the displayed objects; and 
 adjust, based on the ranking scores, rank of the objects that are to be displayed or to be ranked and whose attribute characteristics comply with the reference norms. 
   
     
     
         16 . The search result ranking device as described in  claim 15 , wherein the action information includes objects of user action, information on relative positions of user action objects in the search results, or both. 
     
     
         17 . The search result ranking device as described in  claim 15 , wherein the objects are products or product information. 
     
     
         18 . The search result ranking device as described in  claim 15 , wherein the determining of the commonality level of the one or more attribute characteristics in the objects subjected to user actions based on the user action information on the displayed objects comprises to:
 acquire the attribute characteristics of the displayed objects;   calculate a commonality level of an attribute characteristic of a displayed object that has been clicked based on the recorded user action information on the displayed objects; and   calculate a commonality level of an attribute characteristic of a displayed object that has not been clicked.   
     
     
         19 . The search result ranking device as described in  claim 15 , wherein:
 the determining of the commonality level of the one or more attribute characteristics in the objects subjected to user actions based on the user action information on the displayed objects comprises to:
 acquire the attribute characteristics of the displayed objects; 
 calculate a commonality level of an attribute characteristic of a displayed object that has been clicked based on the recorded user action information on the displayed objects; and 
 calculate a commonality level of an attribute characteristic of a displayed object that has not been clicked; and 
   the commonality level of the attribute characteristic of the displayed object that has been clicked or the commonality level of the attribute characteristic of the displayed object that has not been clicked is a ratio of a number of objects that have the same or similar attribute characteristics in displayed objects that have been clicked or displayed objects that have not been clicked to a total number of objects in the displayed objects that have been clicked or the displayed objects that have not been clicked.   
     
     
         20 . The search result ranking device as described in  claim 15 , wherein:
 the determining of the commonality level of the one or more attribute characteristics in the objects subjected to user actions based on the user action information on the displayed objects comprises to:
 acquire the attribute characteristics of the displayed objects; 
 calculate a commonality level of an attribute characteristic of a displayed object that has been clicked based on the recorded user action information on the displayed objects; and 
 calculate a commonality level of an attribute characteristic of a displayed object that has not been clicked; and 
   the selecting of the attribute characteristics that comply with predetermined requirements to serve as the reference norms for ranking objects that are to be displayed or ranked based on the commonality level comprises to:
 rank various attribute characteristics in displayed objects that have been clicked and various attribute characteristics in displayed objects that have not been clicked in order of high to low commonality level, and select a predetermined quantity of top-ranked attribute characteristics to serve as the reference norms; or 
 regard attribute characteristics with commonality levels greater than a threshold value as the reference norms. 
   
     
     
         21 . The search result ranking device as described in  claim 15 , wherein:
 the determining of the commonality level of the one or more attribute characteristics in the objects subjected to user actions based on the user action information on the displayed objects comprises to:
 acquire the attribute characteristics of the displayed objects; 
 calculate a commonality level of an attribute characteristic of a displayed object that has been clicked based on the recorded user action information on the displayed objects; and 
 calculate a commonality level of an attribute characteristic of a displayed object that has not been clicked; and 
   the selecting of the attribute characteristics that comply with the predetermined requirements to serve as the reference norms for ranking objects that are to be displayed or ranked based on the commonality level comprises to:
 calculate differences in commonality levels of attribute characteristics of displayed objects that have been clicked and commonality levels of attribute characteristics of the displayed object that have not been clicked, rank the various attribute characteristics in order of large to small differences in the commonality levels, and select a predetermined quantity of top-ranked attribute characteristics to serve as the reference norms; or 
 regard attribute characteristics whose differences in commonality level are greater than a set threshold value as the reference norms. 
   
     
     
         22 . The search result ranking device as described in  claim 15 , wherein the objects of user actions are objects that were clicked from among the search results. 
     
     
         23 . The search result ranking device as described in  claim 15 , wherein:
 the objects of user actions are objects that were clicked from among the search results; and   the adjusting of the rank of the objects that are to be displayed or to be ranked, and whose attribute characteristics comply with the reference norms comprises to:
 calculate the commonality level of an attribute characteristics in an object that has been clicked based on the user action information on objects in displayed objects that have been clicked; 
 select attribute characteristics whose commonality levels are greater than a preset threshold value as the reference norms; and 
 raise the rank of the objects that are to be displayed or to be ranked and whose attribute characteristics comply with the reference norms. 
   
     
     
         24 . The search result ranking device as described in  claim 15 , wherein:
 the objects of user actions are objects that were clicked from among the search results; and   the adjusting of the rank of the objects that are to be displayed or to be ranked and whose attribute characteristics comply with the reference norms comprises to:
 calculate a commonality level of an attribute characteristic of an object that has been clicked based on the user action information on objects that have been clicked; 
 rank various attribute characteristics on the objects that have been clicked and various attribute characteristics on objects that have not been clicked in order of highest to lowest commonality levels; 
 select a pre-established quantity of top-ranked attribute characteristics to serve as the reference norms; and 
 raise the rank of the objects that are to be displayed or to be ranked and whose attribute characteristics comply with the reference norms. 
   
     
     
         25 . The search result ranking device as described in  claim 15 , wherein the adjusting of the rank of the objects that are to be displayed or to be ranked and whose attribute characteristics comply with the reference norms comprises to:
 calculate a commonality level of an attribute characteristic of an object that has not been clicked;   select attribute characteristics whose commonality levels are greater than a preset threshold value as the reference norms; and   lower the rank of objects that are to be displayed or to be ranked and whose attribute characteristics comply with the reference norms.   
     
     
         26 . The search result ranking device as described in  claim 15 , wherein the adjusting of the rank of the objects that are to be displayed or to be ranked and whose attribute characteristics comply with the reference norms comprises to:
 calculate a commonality level of an attribute characteristic of an object that has not been clicked based on the user action information on objects that have not been clicked;   rank various attribute characteristics in the objects that have been clicked and various attribute characteristics in the objects that have not been clicked in order of highest to lowest commonality levels;   select a pre-established quantity of top-ranked attribute characteristics to serve as the reference norms; and   lower ranks of objects that are to be displayed or to be ranked and whose attribute characteristics comply with the reference norms.   
     
     
         27 . The search result ranking device as described in  claim 15 , wherein:
 the adjusting of the ranks of the objects that are to be displayed or to be ranked, and whose attribute characteristics comply with the reference norms comprises to re-rank objects that are to be displayed or to be ranked according to the ranking scores; and   the calculating of the ranking scores for the displayed objects comprises to:
 assign first weights according to the sequence in which displayed objects are clicked by the user; 
 calculate a weight of a reference norm based on the first weights of the objects that are among objects that have been click and that comply with the reference norms; 
 calculate comprehensive scores of degrees of influence of the reference norms on the object rankings based on the weights of the reference norms with which the attribute characteristics of the objects that are to be displayed or to be ranked comply; and 
 adjust the ranking scores of the objects according to the comprehensive scores. 
   
     
     
         28 . A computer program product for search result ranking, the computer program product being embodied in a tangible non-transitory computer readable storage medium and comprising computer instructions for:
 recording user action information on displayed objects in search results obtained using one or more query words, wherein the user action information includes a sequence in which a user browsed or clicked the objects;   upon receiving a switch-page request or a switch-screen request, determining a commonality level of one or more attribute characteristics in objects subjected to user actions, wherein the determining of the commonality level is based on the user action information on the displayed objects;   selecting attribute characteristics that comply with predetermined requirements to serve as reference norms for ranking objects that are to be displayed or ranked, wherein the selecting of the attribute characteristics is based on the commonality level;   calculating, based on the user action information, ranking scores for the displayed objects; and   adjusting, based on the ranking scores, rank of the objects that are to be displayed or to be ranked and whose attribute characteristics comply with the reference norms.