Training a search query intent classifier using wiki article titles and a search click log
Techniques are described herein for training a search query intent classifier using wiki article titles and a search click log. Titles of wiki articles that correspond to links that are associated with a specified wiki article and/or titles of wiki articles that are included in a category that includes the specified wiki article are extracted and included with the title of the specified wiki article in an initial set. Each title in the initial set is correlated with respective clicked URI(s) using a search click log. The initial set is expanded to include search terms that are correlated to the clicked URIs based on the search click log to provide an expanded set. The search query intent classifier is trained to classify search queries with respect to a query intent that is associated with the title of the specified wiki article based on the expanded set.
1. A method comprising:
obtaining a designated query intent from a search query;
in response to obtaining the designated query intent from the search query, obtaining a second wiki article corresponding to the designated query intent, wherein the second wiki article includes a second title, a second body, and a plurality of respective links;
obtaining a plurality of first wiki articles that correspond to the plurality of respective links included in the second wiki article, wherein each of the first wiki articles includes a different title and a different body;
extracting a plurality of first titles of the plurality of first wiki articles;
extracting the second title of the second wiki article;
generating an initial key term set including the plurality of first titles and the second title;
correlating each of the plurality of first titles and the second title with at least one respective clicked uniform resource identifier based on a search click log, wherein the search click log comprises a record of historical search queries provided by users and documents clicked by the users from search results retrieved in response to the historical search queries, and wherein each of the documents has a respective uniform resource identifier;
generating an expanded key term set to include the initial key term set expanded to further include search terms correlated to the at least one respective clicked uniform resource identifier;
training a search query intent classifier to classify search queries with respect to the designated query intent based on the expanded key term set;
obtaining the designated query intent from a subsequent search query; and
in response to obtaining the subsequent search query corresponding to the designated query intent, controlling the trained search query intent classifier to select advertisement data based on the expanded key term set.
2. The method of claim 1 , further comprising:
determining a category to which the second wiki article is assigned; and
extracting a plurality of third titles of a plurality of respective third wiki articles that are assigned to the category to be included in the initial key term set;
wherein correlating each of the plurality of first titles and the second title with at least one respective clicked uniform resource identifier comprises:
correlating each of the plurality of first titles, the second title, and the plurality of third titles with at least one respective clicked uniform resource identifier based on the search click log; and
wherein generating the expanded key term set comprises:
expanding the initial key term set to include search terms, in addition to the plurality of first titles, the second title, and the plurality of third titles, that are correlated to the at least one respective clicked uniform resource identifier based on the search click log to provide the expanded key term set.
3. The method of claim 2 , further comprising:
extracting a plurality of fourth titles of a plurality of respective fourth wiki articles that corresponds to a plurality of respective second links to be included in the initial key term set, each of the plurality of second links being associated with at least one of the plurality of third wiki articles;
wherein correlating each of the plurality of first titles and the second title with at least one respective clicked uniform resource identifier comprises:
correlating each of the plurality of first titles, the second title, the plurality of third titles, and the plurality of fourth titles with at least one respective clicked uniform resource identifier based on the search click log; and
wherein generating the expanded key term set comprises:
expanding the initial key term set to include search terms, in addition to the plurality of first titles, the second title, the plurality of third titles, and the plurality of fourth titles, that are correlated to the at least one respective clicked uniform resource identifier based on the search click log to provide the expanded key term set.
4. The method of claim 1 , further comprising:
extracting a plurality of third titles of a plurality of respective third wiki articles that corresponds to a plurality of respective second links to be included in the initial key term set, each of the plurality of second links being associated with at least one of the plurality of first wiki articles;
wherein correlating each of the plurality of first titles and the second title with at least one respective clicked uniform resource identifier comprises:
correlating each of the plurality of first titles, the second title, and the plurality of third titles with at least one respective clicked uniform resource identifier based on the search click log; and
wherein generating the expanded key term set comprises:
expanding the initial key term set to include search terms, in addition to the plurality of first titles, the second title, and the plurality of third titles, that are correlated to the at least one respective clicked uniform resource identifier based on the search click log to provide the expanded key term set.
5. The method of claim 1 , wherein correlating each of the plurality of first titles and the second title with at least one respective clicked uniform resource identifier using a search click log comprises:
generating a bipartite graph that correlates each of the plurality of first titles and the second title to the at least one respective clicked uniform resource identifier based on the search click log.
6. The method of claim 1 , further comprising:
assigning a first probability to the plurality of first titles and to the second title; and
assigning a respective second probability to each of the search terms, each second probability being less than the first probability,
wherein generating the expanded key term set comprises:
expanding the initial key term set to include a first subset of the search terms and to not include a second subset of the search terms to provide the expanded key term set, the first subset including search terms to which respective second probabilities that are greater than a threshold probability are assigned, the second subset including search terms to which respective second probabilities that are less than the threshold probability are assigned.
7. The method of claim 1 , further comprising:
randomly selecting second search terms that are not included in the expanded key term set to provide a negative sample set,
wherein training the search query intent classifier comprises:
training the search query intent classifier to classify the search queries with respect to the designated query intent based on the expanded key term set and the negative sample set.
8. The method of claim 1 , wherein training the search query intent classifier comprises:
training a maxentropy classifier to classify the search queries with respect to the designated query intent based on the expanded key term set.
9. The method of claim 1 , wherein training the search query intent classifier comprises:
training the search query intent classifier to classify the search queries with respect to the designated query intent based on the expanded key term set and further based on text that is included in the second wiki article.
10. The method of claim 1 , wherein extracting the plurality of first titles comprises:
extracting the plurality of first titles of the plurality of respective first wiki articles that corresponds to a plurality of respective article links that are associated with the second wiki article to provide the initial key term set.
11. The method of claim 1 , wherein extracting the plurality of first titles comprises:
extracting the plurality of first titles of the plurality of respective first wiki articles that corresponds to a plurality of respective category links that are associated with the second wiki article to provide the initial key term set.
12. The method of claim 1 , wherein extracting the plurality of first titles comprises:
extracting the plurality of first titles of the plurality of respective first wiki articles that corresponds to a plurality of respective redirection links that are associated with the second wiki article to provide the initial key term set.
13. A method comprising:
determining a category to which a first wiki article is assigned, the first wiki article including a first title associated with a designated query intent and a first body, wherein the first wiki article is included in an initial key term set;
retrieving a plurality of second wiki articles that are assigned to the category;
extracting a plurality of second titles of the plurality of respective second wiki articles, wherein each of the second wiki articles includes a different title and a different body, and wherein the initial key term set includes the first title and the plurality of second titles;
correlating each of the first title and the plurality of second titles with at least one respective clicked uniform resource identifier based on a search click log, wherein the search click log comprises a record of historical search queries provided by users and documents clicked by the users from search results retrieved in response to the search queries, and wherein each of the documents has a respective uniform resource identifier;
generating an expanded key term set to include the initial key term set expanded to further include search terms correlated to the at least one respective clicked uniform resource identifier;
training a search query intent classifier to classify search queries with respect to the designated query intent based on the expanded key term set;
obtaining the designated query intent from a subsequent search query; and
in response to obtaining the subsequent search query corresponding to the designated query intent, controlling the trained search query intent classifier to select advertisement data based on the expanded key term set.
14. The method of claim 13 , further comprising:
extracting a plurality of third titles of a plurality of respective third wiki articles that correspond to a plurality of respective links to be included in the initial key term set, each of the plurality of links being associated with at least one of the plurality of second wiki articles;
wherein correlating each of the first title and the plurality of second titles with at least one respective clicked uniform resource identifier comprises:
correlating each of the first title, the plurality of second titles, and the plurality of third titles with at least one respective clicked uniform resource identifier based on the search click log; and
wherein generating the expanded key term set comprises:
expanding the initial key term set to include search terms, in addition to the first title, the plurality of second titles, and the plurality of third titles, that are correlated to the at least one respective clicked uniform resource identifier based on the search click log to provide the expanded key term set.
15. The method of claim 13 , wherein correlating each of the first title and the plurality of second titles with at least one respective clicked uniform resource identifier using a search click log comprises:
generating a bipartite graph that correlates each of the first title and the plurality of second titles to the at least one respective clicked uniform resource identifier based on the search click log.
16. The method of claim 13 , further comprising:
assigning a first probability to the first title and to the plurality of second titles; and
assigning a respective second probability to each of the search terms, each second probability being less than the first probability,
wherein generating the expanded key term set comprises:
expanding the initial key term set to include a first subset of the search terms and to not include a second subset of the search terms to provide the expanded key term set, the first subset including search terms to which respective second probabilities that are greater than a threshold probability are assigned, the second subset including search terms to which respective second probabilities that are less than the threshold probability are assigned.
17. The method of claim 13 , further comprising:
randomly selecting second search terms that are not included in the expanded key term set to provide a negative sample set,
wherein training the search query intent classifier comprises:
training the search query intent classifier to classify the search queries with respect to the designated query intent based on the expanded key term set and the negative sample set.
18. The method of claim 13 , wherein training the search query intent classifier comprises: training a maxentropy classifier to classify the search queries with respect to the designated query intent based on the expanded key term set.
19. The method of claim 13 , wherein training the search query intent classifier comprises:
training the search query intent classifier to classify the search queries with respect to the designated query intent based on the expanded key term set and further based on text that is included in the first wild article.
20. A system comprising:
a processor coupled to memory and configured to obtain a designated query intent from a search query;
wherein the processor is configured to control an extraction module to:
in response to obtaining the designated query intent from the search query, obtain a second wiki article corresponding to the designated query intent, wherein the second wiki article includes a second title, a second body, and a plurality of respective links;
obtain a plurality of first wiki articles that correspond to the plurality of respective links included in the second wiki article;
configured to extract a plurality of first titles of the plurality of respective first wiki articles and a second title of the second wiki article, wherein each of the first wiki articles has a different title and a different body; and
generate an initial key term including the plurality of first titles and the second title;
wherein the processor is configured to control a correlation module to correlate each of the plurality of first titles and the second title with at least one respective clicked uniform resource identifier using a search click log, wherein the search click log comprises a record of historical search queries provided by users and documents clicked by the users from search results retrieved in response to the historical search queries, and wherein each of the documents has a respective uniform resource identifier;
wherein the processor is configured to further control the expansion module to generate an expanded key term set to include the initial key term set expanded to include search terms correlated to the at least one respective clicked uniform resource identifier;
wherein the processor is configured to control a training module to:
train a search query intent classifier to classify search queries with respect to the designated query intent based on the expanded key term set; and
control the trained search query intent classifier to select advertisement data based on the expanded key term set for a subsequent search query determined to correspond to the designated query intent.
21. A system comprising:
a determination module configured to determine a category to which a first wiki article is assigned, the first wiki article including a first title associated with a designated query intent and a first body, wherein the first wiki article is included in an initial key term set;
an extraction module configured to:
retrieve a plurality of second wiki articles that are assigned to the category,
extract a plurality of second titles of the plurality of respective second wiki articles, and
obtain an initial key term set including the first title and the plurality of second titles wherein each of the second wiki articles includes a different title and a different body, and wherein the initial key term set includes the first title and the plurality of second titles;
a correlation module configured to correlate each of the first title and the plurality of second titles with at least one respective clicked uniform resource identifier based on a search click log, wherein the search click log comprises a record of historical search queries provided by users and documents clicked by the users from search results retrieved in response to the historical search queries, and wherein each of the documents has a respective uniform resource identifier;
an expansion module configured to generate an expanded key term set to include the initial key term set expanded to further include search terms correlated to the at least one respective clicked uniform resource identifier;
a training module configured to:
train a search query intent classifier to classify search queries with respect to the designated query intent based on the expanded key term set; and
control the trained search query intent classifier to select advertisement data based on the expanded key term set for a subsequent search query determined to correspond to the designated query intent.