System and method of geo-based prediction in search result selection
A system and method is disclosed for determining a prediction measurement, or measure, using geo-spatial information which can be used to determine whether or not to include type of information in search results. The prediction measurement comprises a measure of the likelihood that an item of the type of information for which the prediction measure is determined will be selected, or clicked on, by a user, if the item of the type of information is included in the search result. Without limitation, one such information type is news.
1. A method implemented by at least one processing device, comprising:
collecting, by the processing device, prediction information using a received query, the prediction information comprises geo-spatial information including a location name identified in the received query and a newsworthiness click probability value for the location name, the newsworthiness click probability value is obtained from a probability table that specifies, for each place name, a probability that a query comprising the place name will result in a click on a news item;
calculating, by the processing device, a prediction measure using the collected prediction information prior to executing the received query for generating a search result set, the prediction measure being a measure of the likelihood that a user will select a news item type of information returned in response to the query;
transmitting, by the processing device, the prediction measure for generating content of the search result set.
2. The method of claim 1 , further comprising:
using, by a processor, the prediction measure to determine whether or not to include one or more items of the news item type of information in the set of search results generated using the query.
3. The method of claim 1 , wherein the geo-spatial information comprises a location of a user that submitted the query, and a distance between the user's location and the location identified in the query.
4. The method of claim 3 , wherein the information collected further comprises a location type, the method further comprising:
identifying, by the processing device, a location type for the location identified in the query using a location classification scheme.
5. The method of claim 3 , wherein the information collected further comprises a confidence score, the confidence score identifying a level of confidence that a term used in the query is a place name.
6. The method of claim 3 , wherein the information collected further comprises a population measure of at least one of the user's location and the location identified in the query.
7. The method of claim 3 , wherein the information collected further comprises a population density measure of at least one of the user's location and the location identified in the query.
8. The method of claim 1 , wherein the information collected further comprises non-geo-spatial information.
9. The method of claim 1 , wherein the information collected further comprises a query location confidence score.
10. The method of claim 1 , wherein calculating a prediction measure is performed using a model definition generated using training data comprising query logs, the query logs identifying a number of queries, for each query the query logs identifying the search results generated for the query and the search terms used to generate the search results, and for each query for which the search results included at least one item of the type of information the query log identifies whether or not the at least one item was selected.
11. A system comprising:
one or more processors and at least a processor readable storage device having stored thereon:
a feature extractor configured to collect prediction information using a received query, the information comprises geo-spatial information including at least a location name identified in the received query and a newsworthiness click probability value determined for the location name, the newsworthiness click probability value is obtained from a probability table that specifies for each place name, a probability that a query comprising the place name will result in a click on a news item; and
a prediction engine configured to calculate and transmit a prediction measure using the collected information prior to executing the received query for generating a set of search results, the prediction measure being a measure of the likelihood that a user will select a news item type of information returned in response to the query.
12. The system of claim 11 , further comprising:
a search engine configured to use the prediction measure to determine whether or not to include one or more items of the news item type of information in a set of search results generated using the query.
13. The system of claim 11 , wherein the geo-spatial information comprises a location of a user that submitted the query, and a distance between the user's location and the location identified in the query.
14. The system of claim 13 , wherein the information collected further comprises a location type, the method further comprising: identifying a location type for the location identified in the query using a location classification scheme.
15. The system of claim 13 , wherein the information collected further comprises a confidence score, the confidence score identifying a level of confidence that a term used in the query is a place name.
16. The system of claim 13 , wherein the information collected further comprises a population measure of at least one of the user's location and the location identified in the query.
17. The system of claim 13 , wherein the information collected further comprises a population density measure of at least one of the user's location and the location identified in the query.
18. The system of claim 11 , wherein the information collected further comprises non-geo-spatial information.
19. The system of claim 11 , wherein the information collected further comprises a query location confidence score.
20. The system of claim 11 , further comprising a trainer configured to generate a model definition that is used by the prediction engine to calculate the prediction measure, the trainer is configured to generate the model definition using training data comprising query logs, the query logs identifying a number of queries, for each query the query logs identifying the search results generated for the query and the search terms used to generate the search results, and for each query for which the search results included at least one item of the type of information the query log identifies whether or not the at least one item was selected.
21. A non-transitory computer-readable medium tangibly storing program code thereon, the program code comprising:
code to collect prediction information using a received query, the prediction information comprises geo-spatial information including a location name identified in the received query, a newsworthiness click probability value determined for the location name, the newsworthiness click probability value is obtained from a probability table that specifies for each place name, a probability that a query comprising the place name will result in a click on a news item;
code to calculate a prediction measure using the collected information prior to execution of the received query for generation of search results, the prediction measure being a measure of the likelihood that a user will select a news item type of information returned in response to the query; and
code to transmit the prediction measure for the generation of the search results.
22. The medium of claim 21 , the program code further comprising: code to use the prediction measure to determine whether or not to include one or more items of the news item type of information in a set of search results generated using the query.
23. The medium of claim 21 , wherein the geo-spatial information comprises a location of a user that submitted the query, and a distance between the user's location and the location identified in the query.
24. The medium of claim 23 , wherein the information collected further comprises a location type, the method further comprising: identifying a location type for the location identified in the query using a location classification scheme.
25. The medium of claim 23 , wherein the information collected further comprises a confidence score, the confidence score identifying a level of confidence that a term used in the query is a place name.
26. The medium of claim 23 , wherein the information collected further comprises a population measure of at least one of the user's location and the location identified in the query.
27. The medium of claim 23 , wherein the information collected further comprises a population density measure of at least one of the user's location and the location identified in the query.
28. The medium of claim 21 , wherein the information collected further comprises non-geo-spatial information.
29. The medium of claim 21 , wherein the information collected further comprises a query location confidence score.
30. The medium of claim 21 , wherein code to calculate a prediction measure is performed using a model definition generated using training data comprising query logs, the query logs identifying a number of queries, for each query the query logs identifying the search results generated for the query and the search terms used to generate the search results, and for each query for which the search results included at least one item of the type of information the query log identifies whether or not the at least one item was selected.