Generation of refinement terms for search queries

Abstract

A computer-implemented method is provided, which includes receiving from a user a first search query consisting of one or more first query terms, and receiving from the user an indication of a desired level of personalization of refinement options for the search query. Responsively to the search query, a set of one or more refinement terms is generated at least in part responsively to the indication. The set of refinement terms is presented to the user. Responsively to a selection of at least one of the refinement terms by the user, the selected at least one refinement term is added to the first search query to generate a second search query. Search results are presented to the user responsively to the second search query. Other embodiments are also described.

1 - 25 . (canceled) 
     
     
         26 . A computer-implemented method comprising:
 receiving from a plurality of first users a respective plurality of first search queries, each of which consists of one or more first query terms;   presenting to the plurality of first users respective first search result listings, each of which includes one or more first snippets associated with one or more respective search result documents;   constructing a plurality of document profiles, each of which (a) is associated with one of the first snippets selected by one or more of the first users from the respective first search result listings, and (b) represents an interaction of one or more of the first query terms with the first snippet;   receiving from a second user a second search query, which consists of one or more second query terms;   responsively to the second search query, generating a second search result listing, which includes at least a portion of the one or more first snippets;   performing a comparison between (a) the second search query and (b) one or more of the document profiles associated with the first snippets included in the second search result listing;   ranking at least a portion of the second search result listing at least in part responsively to comparison; and   presenting at least a portion of the ranked second search result listing to the second user.   
     
     
         27 . The method according to  claim 26 , wherein constructing the document profiles comprises constructing the document profiles such that at least one of the document profiles includes one or more of the first query terms that do not appear in the first snippet associated with the document profile. 
     
     
         28 . (canceled) 
     
     
         29 . The method according to  claim 26 , wherein the document profiles include respective document association graphs (DAGs), each of which includes a plurality of the first query terms as DAG vertices. 
     
     
         30 . The method according to  claim 29 , wherein constructing the document profiles comprises constructing the DAGs such that at least one of the DAGs includes as vertices one or more of the first query terms that do not appear in the first snippet associated with the document profile. 
     
     
         31 . The method according to  claim 29 ,
 wherein receiving the second search query comprises receiving the second search query, which consists of the one or more second query terms organized linearly, and transforming the second search query into a query association graph including the one or more second query terms as query association graph vertices, and   performing the comparison comprises performing the comparison between (a) the query association graph and (b) the one or more of the DAGs associated with the first snippets included in the second search result listing.   
     
     
         32 . The method according to  claim 29 , wherein performing the comparison comprises constructing a personal association graph (PAG) for the second user that represents interactions of a plurality of previous searches conducted by the second user with a respective plurality of previous search result documents presented to the second user during the previous searches, and includes as vertices at least a portion of one or more previous query terms used in the plurality of previous searches, wherein the comparison includes a correlation between (a) a subgraph of the PAG consisting of: (i) the one or more second query terms, and (ii) at least one additional previous query term directly linked to all of the one or more second query terms in the PAG and (b) the one or more DAGs associated with the first snippets included in the second search result listing, and wherein ranking comprises ranking the at least a portion of the second search result listing at least in part responsively to the correlation. 
     
     
         33 . The method according to  claim 32 , comprising:
 receiving a selection made by the second user of one of the first snippets included in the ranked second search result listing;   calculating a query score for the second search query; and   responsively to the query score:   incrementing edge scores of the PAG between vertices of the PAG respectively including the second query terms; and   incrementing edge scores of the DAG associated with the selected first snippet between vertices of the DAG respectively including the second query terms.   
     
     
         34 . (canceled) 
     
     
         35 . The method according to  claim 33 , wherein incrementing the edge scores of the DAG comprises adding one of the second query terms to the DAG as a new vertex upon finding the second query term is not already included in the DAG. 
     
     
         36 - 44 . (canceled) 
     
     
         45 . The method according to  claim 29 , wherein performing the comparison comprises constructing a topic association graph (TAG) that represents interactions of a plurality of previous searches conducted within a single topic by a plurality of previous users including the second user, and includes as vertices at least a portion of one or more previous query terms used in the plurality of previous searches, wherein the comparison includes a correlation between (a) a subgraph of the TAG consisting of: (i) the one or more second query terms, and (ii) at least one additional previous query term directly linked to all of the one or more second query terms in the PAG and (b) the one or more DAGs associated with the first snippets included in the second search result listing, and wherein ranking comprises ranking the at least a portion of the second search result listing at least in part responsively to the correlation. 
     
     
         46 . The method according to  claim 29 , wherein performing the comparison comprises constructing a global association graph (GAG) that represents interactions of a plurality of searches conducted by a plurality of previous users including the second user with a respective plurality of previous search result documents presented to the previous users during the searches, and includes as vertices at least a portion of one or more previous query terms used in the plurality of previous searches, wherein the comparison includes a correlation between (a) a subgraph of the GAG consisting of: (i) the one or more second query terms, and (ii) at least one additional previous query term directly linked to all of the one or more second query terms in the GAG and (b) the one or more DAGs associated with the first snippets included in the second search result listing, and wherein ranking comprises ranking the at least a portion of the second search result listing at least in part responsively to the correlation. 
     
     
         47 - 101 . (canceled) 
     
     
         102 . Apparatus comprising:
 an interface for communicating with a plurality of first users and with a second user; and   a processor, which is configured to receive from the plurality of first users, via the interface, a respective plurality of first search queries, each of which consists of one or more first query terms; present to the plurality of first users, via the interface, respective first search result listings, each of which includes one or more first snippets associated with one or more respective search result documents; construct a plurality of document profiles, each of which (a) is associated with one of the first snippets selected by one or more of the first users from the respective first search result listings, and (b) represents an interaction of one or more of the first query terms with the first snippet; receiving from the second user, via the interface, a second search query, which consists of one or more second query terms; responsively to the second search query, generate a second search result listing, which includes at least a portion of the one or more first snippets; perform a comparison between (a) the second search query and (b) one or more of the document profiles associated with the first snippets included in the second search result listing; rank at least a portion of the second search result listing at least in part responsively to comparison; and present at least a portion of the ranked second search result listing to the second user via the interface.   
     
     
         103 . The apparatus according to  claim 102 , wherein the processor is configured to construct the document profiles such that at least one of the document profiles includes one or more of the first query terms that do not appear in the first snippet associated with the document profile. 
     
     
         104 . (canceled) 
     
     
         105 . The apparatus according to  claim 102 , wherein the document profiles include respective document association graphs (DAGs), each of which includes a plurality of the first query terms as DAG vertices. 
     
     
         106 . The apparatus according to  claim 105 , wherein the processor is configured to construct the document profiles by constructing the DAGs such that at least one of the DAGs includes as vertices one or more of the first query terms that do not appear in the first snippet associated with the document profile. 
     
     
         107 . The apparatus according to  claim 105 , wherein the processor is configured to receive the second search query, which consists of the one or more second query terms organized linearly; transform the second search query into a query association graph including the one or more second query terms as query association graph vertices; and perform the comparison between (a) the query association graph and (b) the one or more of the DAGs associated with the first snippets included in the second search result listing. 
     
     
         108 . The apparatus according to  claim 105 , wherein the processor is configured to perform the comparison by constructing a personal association graph (PAG) for the second user that represents interactions of a plurality of previous searches conducted by the second user with a respective plurality of previous search result documents presented to the second user during the previous searches, and includes as vertices at least a portion of one or more previous query terms used in the plurality of previous searches, and wherein the comparison includes a correlation between (a) a subgraph of the PAG consisting of: (i) the one or more second query terms, and (ii) at least one additional previous query term directly linked to all of the one or more second query terms in the PAG and (b) the one or more DAGs associated with the first snippets included in the second search result listing, and wherein ranking comprises ranking the at least a portion of the second search result listing at least in part responsively to the correlation. 
     
     
         109 . The apparatus according to  claim 108 , wherein the processor is configured to receive a selection made by the second user of one of the first snippets included in the ranked second search result listing; calculate a query score for the second search query; and responsively to the query score: increment edge scores of the PAG between vertices of the PAG respectively including the second query terms, and increment edge scores of the DAG associated with the selected first snippet between vertices of the DAG respectively including the second query terms. 
     
     
         110 . (canceled) 
     
     
         111 . The apparatus according to  claim 109 , wherein the processor is configured to increment the edge scores of the DAG by adding one of the second query terms to the DAG as a new vertex upon finding the second query term is not already included in the DAG. 
     
     
         112 - 120 . (canceled) 
     
     
         121 . The apparatus according to  claim 105 , wherein the processor is configured to perform the comparison by constructing a topic association graph (TAG) that represents interactions of a plurality of previous searches conducted within a single topic by a plurality of previous users including the second user, and includes as vertices at least a portion of one or more previous query terms used in the plurality of previous searches, and wherein the comparison includes a correlation between (a) a subgraph of the TAG consisting of: (i) the one or more second query terms, and (ii) at least one additional previous query term directly linked to all of the one or more second query terms in the PAG and (b) the one or more DAGs associated with the first snippets included in the second search result listing, and wherein ranking comprises ranking the at least a portion of the second search result listing at least in part responsively to the correlation. 
     
     
         122 - 177 . (canceled) 
     
     
         178 . A computer software product, comprising a tangible computer-readable medium in which program instructions are stored, which instructions, when read by a computer, cause the computer to receive from a plurality of first users a respective plurality of first search queries, each of which consists of one or more first query terms; present to the plurality of first users respective first search result listings, each of which includes one or more first snippets associated with one or more respective search result documents; construct a plurality of document profiles, each of which (a) is associated with one of the first snippets selected by one or more of the first users from the respective first search result listings, and (b) represents an interaction of one or more of the first query terms with the first snippet; receiving from a second user a second search query, which consists of one or more second query terms; responsively to the second search query, generate a second search result listing, which includes at least a portion of the one or more first snippets; perform a comparison between (a) the second search query and (b) one or more of the document profiles associated with the first snippets included in the second search result listing; rank at least a portion of the second search result listing at least in part responsively to comparison; and present at least a portion of the ranked second search result listing to the second user. 
     
     
         179 - 315 . (canceled)