Re-ranking device
Provided is a re-ranking device that enables a re-ranking process with less possibility of decrease in accuracy due to shortage of data. A re-ranking device 100 includes an input unit 101 that receives a search query from one user, a re-ranking model storage unit 106 that stores a plurality of re-ranking models prepared in accordance with common information of a plurality of users, a search unit 102 that performs a search on the basis of the search query and obtains a search result, and a re-ranking processing unit 107 that selects one re-ranking model on the basis of common information of the one user and performs a re-ranking process on the search result using the one re-ranking model.
1. A re-ranking device comprising:
an input unit, implemented by circuitry, configured to receive a search query from one user;
a plurality of re-ranking models prepared in accordance with common information of a plurality of users;
a search unit, implemented by the circuitry, configured to perform a search on the basis of the search query and obtain a search result;
a re-ranking processing unit, implemented by the circuitry, configured to select one re-ranking model on the basis of common information of the one user, and perform a re-ranking process on the search result using the one re-ranking model;
a log database configured to store log information containing a search query, a search result list by the search query, and a click log selected by a user;
a clustering unit, implemented by the circuitry, configured to classify the log information on the basis of common information of a plurality of users; and
a re-ranking model generation unit, implemented by the circuitry, configured to generate the plurality of re-ranking models on the basis of the log information classified on the basis of the common information, wherein
the re-ranking processing unit selects one re-ranking model from the plurality of re-ranking models generated by the re-ranking model generation unit, and performs the re-ranking process.
2. The re-ranking device according to claim 1 , wherein the common information is information indicating a search tendency of a user.
3. The re-ranking device according to claim 2 , wherein the re-ranking processing unit derives a search tendency from a search query received by the input unit and a search history of the one user.
4. The re-ranking device according to claim 3 , wherein
the search tendency includes an association between a search query and a click log, and
the re-ranking processing unit derives the search tendency by deriving the association from the search history of the one user.
5. The re-ranking device according to claim 4 , wherein the association is at least one of the number of character strings in each of the search query and the click log, the degree of matching in character strings between the search query and the click log, and a difference in character strings between the search query and the click log.
6. The re-ranking device according to claim 5 , wherein the search tendency is based on at least one of the type of characters forming the search query, a character string in the search query, and the number of characters of a character string not used for search.
7. The re-ranking device according to claim 3 , wherein the search tendency is based on at least one of the type of characters forming the search query, a character string in the search query, and the number of characters of a character string not used for search.
8. The re-ranking device according to claim 3 , wherein the search tendency is based on a means of inputting the search query.
9. The re-ranking device according to claim 1 , wherein the common information is information indicating user attributes.
10. The re-ranking device according to claim 4 , wherein the search tendency is based on a means of inputting the search query.
11. The re-ranking device according to claim 5 , wherein the search tendency is based on a means of inputting the search query.
12. The re-ranking device according to claim 4 , wherein the search tendency is based on at least one of the type of characters forming the search query, a character string in the search query, and the number of characters of a character string not used for search.
13. The re-ranking device according to claim 1 , wherein the re-ranking model is generated by machine learning using a search query and a search result list as explanatory variables and a clicked POI as a response variable.
14. A re-ranking device comprising:
an input unit, implemented by circuitry, configured to receive a search query from one user;
a plurality of re-ranking models prepared in accordance with common information of a plurality of users;
a search unit, implemented by the circuitry, configured to perform a search on the basis of the search query and obtain a search result; and
a re-ranking processing unit, implemented by the circuitry, configured to select one re-ranking model on the basis of common information of the one user, and perform a re-ranking process on the search result using the one re-ranking model,
wherein
the common information is information indicating a search tendency of a user,
the re-ranking processing unit derives a search tendency from a search query received by the input unit and a search history of the one user;
the search tendency includes an association between a search query and a click log,
the re-ranking processing unit derives the search tendency by deriving the association from the search history of the one user,
the association is at least one of the number of character strings in each of the search query and the click log, the degree of matching in character strings between the search query and the click log, and a difference in character strings between the search query and the click log, and
the re-ranking model generation unit derives a search tendency based on at least one of a degree of coincidence or a difference in the number of characters of a search query and a click log, clusters log information including the search query, the search result list and the click log based on the search tendency, and generates the re-ranking models using the search query and the search result list as an explanatory variable and the click log as a response variable.
15. A method, implemented by circuitry of a re-ranking device, the method comprising:
receiving, by an input unit, a search query from one user;
storing plurality of re-ranking models prepared in accordance with common information of a plurality of users;
searching, by a search unit, a search on the basis of the search query and obtain a search result;
selecting, by a re-ranking processing unit, one re-ranking model on the basis of common information of the one user, and performing a re-ranking process on the search result using the one re-ranking model;
wherein the method further includes
storing a log database containing a search query, a search result list by the search query, and a click log selected by a user;
classifying, by a clustering unit, the log information on the basis of common information of a plurality of users; and
generating, by a re-ranking model generation unit, the plurality of re-ranking models on the basis of the log information classified on the basis of the common information, wherein
selecting, by the re-ranking processing unit, one re-ranking model from the plurality of re-ranking models generated by the re-ranking model generation unit, and performs the re-ranking process.