Method, system and storage medium for implementing intelligent question answering
Embodiments of the present invention provide a method, system and storage medium for implementing intelligent question answering. The method includes: receiving a query question; performing a semantic analysis of the question; performing corresponding search processing for the question based on a result of the semantic analysis, wherein the search processing includes search processing performed for the question by at least one of a semantic relationship mining system, a text library search system, a knowledge base search system, and a question and answer library search system; and returning an answer based on a result of the search processing. In this way, the accuracy of answers to the questions is effectively improved.
1. A method for implementing intelligent question answering, comprising:
receiving a query question;
performing a semantic analysis of the query question by analyzing the query question to obtain a dependency syntax tree that describes a dependency relationship between various portions of the query question;
performing corresponding search processing for the query question based on a result of the semantic analysis, the corresponding search processing comprising search processing performed for the query question by a text library search system; and
returning an answer based on a result of the corresponding search processing,
wherein performing the search processing for the query question by the text library search system comprises:
performing a search to obtain a document set related to the query question;
calculating document correlation of documents in the document set;
searching the document set for a candidate segment based on the document correlation, the candidate segment comprising a candidate answer; and
extracting the candidate answer from the candidate segment, and performing a first ranking operation;
wherein performing the first ranking operation comprises:
performing ranking based on at least one feature of a uniform resource locator (URL) weight, and an offset weight, the URL weight being a weight of a URL link and a site to which a document where the candidate answer is located belongs, and the offset weight being a distance coefficient of the candidate answer with respect to a keyword in the query question in different candidate segments.
2. The method according to claim 1 , wherein the performing the semantic analysis of the query question comprises:
identifying a focus, an opinion, and a type of the opinion of the query question.
3. The method according to claim 2 , further comprising, after the type of the opinion of the query question is identified:
performing normalization processing on the type of the opinion of the query question.
4. The method according to claim 1 , wherein the returning the answer based on the result of the corresponding search processing comprises:
filtering the result of the corresponding search processing by using intra-domain knowledge; and
returning the answer based on a result of the filtering.
5. The method according to claim 4 , wherein the returning the answer based on the result of the filtering comprises:
performing a second ranking operation of the result of the filtering; and
returning the ranked result of the filtering.
6. The method according to claim 5 , wherein the performing the second ranking operation of the result of the filtering comprises:
ranking the result of the filtering by using a supervised machine learning model.
7. The method according to claim 5 , wherein the returning the answer based on the result of the filtering comprises:
performing a third ranking operation of the result of the filtering; and
returning the ranked result of the filtering.
8. The method according to claim 7 , wherein the performing the third ranking operation of the result of the filtering comprises:
validating a matching degree between the candidate answer and the query question; and
ranking the candidate answer based on the matching degree.
9. A system for implementing intelligent question answering, comprising:
at least one processor; and
a memory storing instructions, which when executed by the at least one processor, cause the at least one processor to perform operations, the operations comprising:
receiving a query question;
performing a semantic analysis of the query question by analyzing the query question to obtain a dependency syntax tree that describes a dependency relationship between various portions of the query question;
performing corresponding search processing for the query question based on a result of the semantic analysis, the corresponding search processing comprising search processing performed for the query question by a text library search system; and
returning an answer based on a result of the corresponding search processing,
wherein the performing the search processing for the query question by the text library search system comprises:
performing a search to obtain a document set related to the query question;
calculating document correlation of documents in the document set;
searching the document set for a candidate segment based on the document correlation, the candidate segment comprising a candidate answer; and
extracting the candidate answer from the candidate segment, and performing a first ranking operation;
wherein performing the first ranking operation comprises:
performing ranking based on at least one feature of a uniform resource locator (URL) weight, and an offset weight, the URL weight being a weight of a URL link and a site to which a document where the candidate answer is located belongs, and the offset weight being a distance coefficient of the candidate answer with respect to a keyword in the query question in different candidate segments.
10. The system according to claim 9 , wherein performing the semantic analysis of the query question analysis comprises:
identifying a focus, an opinion, and a type of the opinion of the query question.
11. The system according to claim 10 , wherein the operations further comprise:
performing normalization processing on the type of the opinion of the query question after the type of the question is identified.
12. A non-volatile computer storage medium storing one or more modules, the one or more modules which when executed by a device, cause the device to perform a method for implementing intelligent question answering, the method comprising:
receiving a query question;
performing a semantic analysis of the query question by analyzing the query question to obtain a dependency syntax tree that describes a dependency relationship between various portions of the query question;
performing corresponding search processing for the query question based on a result of the semantic analysis, the corresponding search processing comprising search processing performed for the query question by a text library search system; and
returning an answer based on a result of the corresponding search processing,
wherein the performing the search processing for the query question by the text library search system comprises:
performing a search to obtain a document set related to the query question;
calculating document correlation of documents in the document set;
searching the document set for a candidate segment based on the document correlation, the candidate segment comprising a candidate answer; and
extracting the candidate answer from the candidate segment, and performing a first ranking operation;
wherein performing the first ranking operation comprises:
performing ranking based on at least one feature of a uniform resource locator (URL) weight, and an offset weight, the URL weight being a weight of a URL link and a site to which a document where the candidate answer is located belongs, and the offset weight being a distance coefficient of the candidate answer with respect to a keyword in the query question in different candidate segments.