Reverse question answering
View Patent ↗Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for associating each of a plurality of different terms to documents that contain the term to create first associations; associating each of the associated documents to one or more queries to create second associations, wherein search results for each of the queries include a reference to the associated document; determining a particular query and a particular term are associated with a same document based on the first and second associations; in response to the determination, creating a mapping between the particular query and the particular term when both are associated with the same document; and selecting a respective answer from a plurality of mapped terms for each the mapped queries, wherein the respective answer is selected based on a count the respective answer occurs in documents contained in search results for the mapped queries.
1. A computer-implemented method, the method comprising:
for each of a plurality of categories and for a collection of documents, wherein each category is associated with one or more respective text patterns and one or more keywords or phrases specific to the category:
applying the text patterns specific to the category to the collection of documents to identify a plurality of fact terms in the documents, each fact term being a segment of text that matches one of the text patterns specific to the category one or more times in the documents of the collection;
creating and storing in a computer readable memory a plurality of first associations, each first association associating a respective fact term of the plurality of fact terms to a respective plurality of documents that each contain the respective fact term;
creating and storing in a computer readable memory a plurality of second associations between documents and previously-submitted queries, each document being one of the associated documents in the first associations, each previously-submitted query including a keyword or phrase specific to the category, each second association associating a respective document from the first associations to the one or more previously-submitted queries that had a search result that referenced the respective document; and
generating from the first associations and the second associations and storing in a computer readable memory a plurality of mappings, each of the mappings associating a query with a single fact term, wherein the query and the fact term are both associated with at least a threshold number of the same documents by the second and first associations respectively.
2. The method of claim 1 , wherein, for each of the plurality of categories, the plurality of fact terms comprises fact terms that occur in more than a threshold number of documents in the collection of documents.
3. The method of claim 1 , wherein associating a respective document to the one or more previously-submitted queries comprises:
determining that the respective document is referenced by a highly ranked search result for each of the one or more previously-submitted queries.
4. The method of claim 1 , wherein associating a respective document to the one or more previously-submitted queries comprises:
determining that a percentage of clicks of the search result that referenced the respective document exceeds a threshold percentage of total clicks of search results for each of the one or more previously-submitted queries.
5. The method of claim 1 , wherein:
the plurality of categories comprise a category of measurements, a category of dates, and a category of names of persons.
6. The method of claim 1 , further comprising:
receiving a user query;
identifying in the mappings a first mapping that associates the user query with a first fact term; and
providing the first fact term in a response to the user query.
7. A system comprising:
a storage medium having instructions stored thereon; and
data processing apparatus operable to execute the instructions to perform operations comprising:
for each of a plurality of categories and for a collection of documents, wherein each category is associated with one or more respective text patterns and one or more keywords or phrases specific to the category:
applying the text patterns specific to the category to the collection of documents to identify a plurality of fact terms in the documents, each fact term being a segment of text that matches one of the text patterns specific to the category one or more times in the documents of the collection;
creating and storing in a computer readable memory a plurality of first associations, each first association associating a respective fact term of the plurality of fact terms to a respective plurality of documents that each contain the respective fact term;
creating and storing in a computer readable memory a plurality of second associations between documents and previously-submitted queries, each document being one of the associated documents in the first associations, each previously-submitted query including a keyword or phrase specific to the category, each second association associating a respective document from the first associations to the one or more previously-submitted queries that had a search result that referenced the respective document; and
generating from the first associations and the second associations and storing in a computer readable memory a plurality of mappings, each of the mappings associating a query with a single fact term, wherein the query and the fact term are both associated with at least a threshold number of the same documents by the second and first associations respectively.
8. The system of claim 7 , wherein, for each of the plurality of categories, the plurality of fact terms comprises fact terms that occur in more than a threshold number of documents in the collection of documents.
9. The system of claim 7 , wherein associating a respective document to the one or more previously-submitted queries comprises:
determining that the respective document is referenced by a highly ranked search result for each of the one or more previously-submitted queries.
10. The system of claim 7 , wherein associating a respective document to the one or more previously-submitted queries comprises:
determining that a percentage of clicks of the search result that referenced the respective document exceeds a threshold percentage of total clicks of search results for each of the one or more previously-submitted queries.
11. The system of claim 7 , wherein:
the plurality of categories comprise a category of measurements, a category of dates, and a category of names of persons.
12. The system of claim 7 , wherein the operations further comprise:
receiving a user query;
identifying in the mappings a first mapping that associates the user query with a first fact term; and
providing the first fact term in a response to the user query.
13. A computer storage medium having instructions stored thereon which, when executed by a data processing apparatus, cause the data processing apparatus to perform operations comprising:
for each of a plurality of categories and for a collection of documents, wherein each category is associated with one or more respective text patterns and one or more keywords or phrases specific to the category:
applying the text patterns specific to the category to the collection of documents to identify a plurality of fact terms in the documents, each fact term being a segment of text that matches one of the text patterns specific to the category one or more times in the documents of the collection;
creating and storing in a computer readable memory a plurality of first associations, each first association associating a respective fact term of the plurality of fact terms to a respective plurality of documents that each contain the respective fact term;
creating and storing in a computer readable memory a plurality of second associations between documents and previously-submitted queries, each document being one of the associated documents in the first associations, each previously-submitted query including a keyword or phrase associated with the category, each second association associating a respective document from the first associations to the one or more previously-submitted queries that had a search result that referenced the respective document; and
generating from the first associations and the second associations and storing in a computer readable memory a plurality of mappings, each of the mappings associating a query with a single fact term, wherein the query and the fact term are both associated with at least a threshold number of the same documents by the second and first associations respectively.
14. The computer storage medium of claim 13 , wherein, for each of the plurality of categories, the plurality of fact terms comprises fact terms that occur in more than a threshold number of documents in the collection of documents.
15. The computer storage medium of claim 13 , wherein associating a respective document to the one or more previously-submitted queries comprises:
determining that the respective document is referenced by a highly ranked search result for each of the one or more previously-submitted queries.
16. The computer storage medium of claim 13 , wherein associating a respective document to the one or more previously-submitted queries comprises:
determining that a percentage of clicks of the search result that referenced the respective document exceeds a threshold percentage of total clicks of search results for each of the one or more previously-submitted queries.
17. The computer storage medium of claim 13 , wherein:
the plurality of categories comprise a category of measurements, a category of dates, and a category of names of persons.
18. The computer storage medium of claim 13 , wherein the operations further comprise:
receiving a user query;
identifying in the mappings a first mapping that associates the user query with a first fact term; and
providing the first fact term in a response to the user query.