Reducing query ambiguity using graph matching
Embodiments of the present invention provide systems, methods, and computer storage media directed at interpreting a query related to a dataset to reduce ambiguity of the query. In embodiments, a method includes determining potential meanings of words of a natural language query related to a particular dataset. A matching graph can be generated for the words of the query and the potential meanings for the words including similarity scores for the potential meanings. The method further includes reducing ambiguity of the matching graph by assigning selected potential meanings to the words. Upon reducing ambiguity of the matching graph, the matching graph can be used to provide a response to the query. Other embodiments may be described and/or claimed herein.
1. A computer-implemented method for understanding queries, comprising:
determining potential meanings of portions of a query using a dataset, wherein the dataset includes metadata, event data, and specialized data;
generating a query graph, wherein the query graph comprises similarity scores for the potential meanings matched with the portions of the query, wherein the query graph comprises:
a first similarity score of a first meaning of a first portion of the query, wherein the first similarity score is based upon a first similarity of the first portion of the query to one or more portions of the dataset;
a second similarity score of a second meaning of the first portion of the query, wherein the second similarity score is based upon the first similarity of the first portion of the query to one or more portions of the dataset;
a third similarity score of a third meaning of a second portion of the query, wherein the third similarity score is based upon a second similarity of the second portion of the query to one or more portions of the dataset; and
a fourth similarity score of a fourth meaning of the second portion of the query, wherein the fourth similarity score is based upon the second similarity of the second portion of the query to one or more portions of the dataset;
reducing ambiguity of the query by assigning a single meaning to each of the portions of the query based upon the similarity scores, wherein the assigning comprises assigning the first meaning to the first portion of the query based upon the first similarity score of the first meaning and assigning the third meaning to the second portion of the query based upon the third similarity score of the third meaning; and
upon reducing ambiguity of the query, generating an answer to the query using the first meaning of the first portion of the query and the third meaning of the second portion of the query.
2. The computer-implemented method of claim 1 , wherein the similarity scores are generating using string matching based upon string comparison.
3. The computer-implemented method of claim 1 , wherein reducing ambiguity further comprises:
designating at least one of the portions of the query as a key word by assigning a potential meaning associated with the key word, wherein the key word indicates a main element being asked in the query.
4. The computer-implemented method of claim 1 , wherein reducing ambiguity further comprises:
selecting a potential meaning for a word when the potential meaning scores above a predetermined threshold.
5. The computer-implemented method of claim 1 , wherein reducing ambiguity further comprises:
selecting a first potential meaning for a word when the first potential meaning for the word is equivalent to a second potential meaning for the word.
6. The computer-implemented method of claim 1 , wherein reducing ambiguity further comprises:
analyzing an order of words to select a potential meaning for a word, the order of words indicating a relationship between the order of words with the potential meaning for the word.
7. The computer-implemented method of claim 1 , wherein reducing ambiguity further comprises:
presenting a clarification question to a user based upon failure to select one of the potential meanings for a word.
8. The computer-implemented method of claim 7 , wherein the clarification question is presented using one or more of a selectable multiple-choice list and asking to restate the query.
9. A non-transitory machine-readable storage medium including instructions which, when executed by a machine, cause the machine to:
determine potential meanings of words of a query using a dataset, wherein the dataset includes metadata, event data, and specialized data;
generate a query graph, wherein the query graph includes similarity scores for the potential meanings matched with the words of the query, wherein the query graph comprises:
a first similarity score of a first meaning of a first word of the query, wherein the first similarity score is based upon a first similarity of the first word of the query to one or more portions of the dataset;
a second similarity score of a second meaning of the first word of the query, wherein the second similarity score is based upon the first similarity of the first word of the query to one or more portions of the dataset;
a third similarity score of a third meaning of a second word of the query, wherein the third similarity score is based upon a second similarity of the second word of the query to one or more portions of the dataset; and
a fourth similarity score of a fourth meaning of the second word of the query, wherein the fourth similarity score is based upon the second similarity of the second word of the query to one or more portions of the dataset;
reduce ambiguity of the query by assigning a single meaning to at least one of the words of the query; and
upon reducing ambiguity of the query, generate an answer to the query using the query graph.
10. The non-transitory machine-readable storage medium of claim 9 , wherein the instructions further cause the machine to:
designate at least one of the words of the query as a key word; and
assign a potential meaning associated with the key word, wherein the key word indicates a main element being asked in the query.
11. The non-transitory machine-readable storage medium of claim 9 , wherein the instructions further cause the machine to:
select a potential meaning for a word when the potential meaning scores above a predetermined threshold.
12. The non-transitory machine-readable storage medium of claim 9 , wherein the instructions further cause the machine to:
select a first potential meaning for a word when the first potential meaning for the word is equivalent to a second potential meaning for the word.
13. The non-transitory machine-readable storage medium of claim 9 , wherein the instructions further cause the machine to:
analyze an order of words to select a potential meaning for a word, the order of words indicating a relationship between the order of words with the potential meaning for the word.
14. The non-transitory machine-readable storage medium of claim 9 , wherein the instructions further cause the machine to:
present a clarification question to a user based upon failure to select one of the potential meanings for a word.
15. A computer system comprising:
one or more processors; and
one or more computer storage media storing instructions that, when used by the one or more processors, cause the one or more processors to run an advertising system of one or more components configured to:
determine potential meanings of words of a query using a dataset, wherein the dataset includes metadata, event data, and specialized data;
generate a query graph, wherein the query graph includes similarity scores for the potential meanings matched with the words of the query;
reduce ambiguity of the query by assigning a single meaning to each of the words of the query based upon the similarity scores, wherein the assigning comprises assigning a first single meaning to a first word of the query based upon a first similarity score of the first single meaning and assigning a second single meaning to a second word of the query based upon a second similarity score of the second single meaning; and
upon reducing ambiguity of the query, generate an answer to the query using the first single meaning of the first word of the query and the second single meaning of the second word of the query.
16. The computer system of claim 15 , wherein the instructions further cause the one or more processors to:
designate at least one of the words of the query as a key word; and
assign a potential meaning associated with the key word, wherein the key word indicates a main element being asked in the query.
17. The computer system of claim 15 , wherein the instructions further cause the one or more processors to:
select a potential meaning for a word when the potential meaning scores above a predetermined threshold.
18. The computer system of claim 15 , wherein the instructions further cause the one or more processors to:
select a first potential meaning for a word when the first potential meaning for the word is equivalent to a second potential meaning for the word.
19. The computer system of claim 15 , wherein the instructions further cause the one or more processors to:
analyze an order of words to select a potential meaning for a word, the order of words indicating a relationship between the order of words with the potential meaning for the word.
20. The computer system of claim 15 , wherein the instructions further cause the one or more processors to:
present a clarification question to a user based upon failure to select one of a first potential meaning for a word and a second potential meaning for the word.