Query expansion
View Patent ↗An expanded queries data structure is described. The data structure is produced on the basis of a set of seed queries, and consists of entries each specifying an expanded query submitted by a user that has been determined to have a high degree of relatedness to at least a plurality of the seed queries of the set. The expanded queries specified by the entries of the expanded queries data structure can be used to define a segment of users expected to have interests characterized by the seed queries.
1. A method in a computing system for expanding a set of seed queries, comprising:
for each of the seed queries:
accessing a behavioral index that maps from actual queries received from users to other actual queries received from the same users during the same time period;
for each of a plurality of candidate queries other than the seed query, determining a score indicating the extent to which the candidate query is mapped-to from the seed query by the behavioral index;
accessing a contextual index that maps from actual queries received from users to other actual queries based upon text provided by multiple publicly-available sources in response to the actual queries;
for each of a plurality of candidate queries other than the seed query, determining a score indicating the extent to which the candidate query is mapped-to from the seed query by the contextual index;
for each of the candidate queries:
combining the scores indicating the extent to which the candidate query is mapped-to from the seed queries by the behavioral index to obtain a behavioral score for the candidate query;
combining the scores indicating the extent to which the candidate query is mapped-to from the seed queries by the contextual index to obtain a contextual score for the candidate query;
combining the behavioral score for the candidate query with the contextual score for the candidate query to obtain an overall score for the candidate query; and
selecting as expanded queries the proper subset of the candidate queries having the highest overall score.
2. The method of claim 1 , further comprising constructing a segment definition based on the selected expanded queries.
3. The method of claim 2 wherein the constructed segment definition is further based on the set of seed queries.
4. A tangible computer-readable medium storing instructions that, when executed by a computing system, cause the computing system to perform operations for expanding a set of seed queries, the operations comprising:
for each of the seed queries:
accessing a behavioral index that maps from actual queries received from users to other actual queries received from the same users during the same time period;
for each of a plurality of candidate queries other than the seed query, determining a score indicating the extent to which the candidate query is mapped-to from the seed query by the behavioral index;
accessing a contextual index that maps from actual queries received from users to other actual queries based upon text provided by multiple publicly-available sources in response to the actual queries;
for each of a plurality of candidate queries other than the seed query, determining a score indicating the extent to which the candidate query is mapped-to from the seed query by the contextual index;
for each of the candidate queries:
combining the scores indicating the extent to which the candidate query is mapped-to from the seed queries by the behavioral index to obtain a behavioral score for the candidate query;
combining the scores indicating the extent to which the candidate query is mapped-to from the seed queries by the contextual index to obtain a contextual score for the candidate query;
combining the behavioral score for the candidate query with the contextual score for the candidate query to obtain an overall score for the candidate query; and
selecting as expanded queries the proper subset of the candidate queries having the highest overall score.
5. The tangible computer-readable medium of claim 4 , the operations further comprising:
constructing a segment definition based on the selected expanded queries.
6. The tangible computer-readable medium of claim 5 wherein the constructed segment definition is further based on the set of seed queries.
7. A computing system for expanding a set of seed queries, comprising:
a first accessing subsystem configured to access a behavioral index that maps from actual queries received from users to other actual queries received from the same users during the same time period;
a first determining subsystem configured to determine, for each of the seed queries, for each of a plurality of candidate queries other than the seed query, a score indicating the extent to which the candidate query is mapped-to from the seed query by the behavioral index, wherein the first determining subsystem comprises a processor;
a second accessing subsystem configured to access a contextual index that maps from actual queries received from users to other actual queries based upon text provided by multiple publicly-available sources in response to the actual queries;
a second determining subsystem configured to determine, for each of the seed queries, for each of a plurality of candidate queries other than the seed query, a score indicating the extent to which the candidate query is mapped-to from the seed query by the contextual index;
a first combining subsystem configured to combine, for each of the candidate queries, the scores indicating the extent to which the candidate query is mapped-to from the seed queries by the behavioral index to obtain a behavioral score for the candidate query;
a second combining subsystem configured to combine, for each of the candidate queries, the scores indicating the extent to which the candidate query is mapped-to from the seed queries by the contextual index to obtain a contextual score for the candidate query;
a third combining subsystem configured to combine, for each of the candidate queries, the behavioral score for the candidate query with the contextual score for the candidate query to obtain an overall score for the candidate query; and
a selecting subsystem configured to select as expanded queries the proper subset of the candidate queries having the highest overall score.
8. The computing system of claim 7 , further comprising:
a constructing subsystem configured to construct a segment definition based on the selected expanded queries.
9. The computing system of claim 8 wherein the constructed segment definition is further based on the set of seed queries.