CN-115048498-B - Non-discriminant system based on clause library

CN115048498BCN 115048498 BCN115048498 BCN 115048498BCN-115048498-B

Abstract

The invention discloses a non-discrimination system based on a clause library, which belongs to the technical field of electrical knowledge questions and answers and comprises a word library module, a clause library, a standard question and answer module, a combination question and answer module, a correlation module, a discrimination module and a server, wherein the word library module is used for establishing a topic word library, storing corresponding standard clauses through the clause library, the standard question and answer module is used for establishing a corresponding standard question and answer library based on the standard clauses in the clause library, the combination question and answer module is used for establishing a combination question and answer library based on the topic word library, establishing a sentence model, combining topic words in the topic word library through the established sentence model, outputting a topic word set of the combination question and answer, carrying out value conversion of the topic word set to obtain a topic word set value, setting a context value of the corresponding combination question and answer, integrating the context value and the corresponding topic word set value to obtain a matching vector, mapping the matching vector into a vector space, and summarizing the combination question and answer library according to the current combination question and answer.

Inventors

ZHAO CHANGWEI
QIAN YUCHENG
LI JIANLIN
WANG SHUDONG
ZHU TAIYUN
YANG WEI
ZHEN CHAO
PAN CHAO

Assignees

国网安徽省电力有限公司电力科学研究院
国网安徽省电力有限公司

Dates

Publication Date: 20260512
Application Date: 20220607

Claims (7)

1. The term library-based non-discrimination system is characterized by comprising a word library module, a term library, a standard question-answering module, a combined question-answering module, an association module, a discrimination module and a server; The system comprises a term library module, a standard question-answering module, a sentence model, a context value setting module, a matching vector mapping module and a combination question-answering module, wherein the term library module is used for establishing a theme word library, storing corresponding standard terms through the term library, the standard question-answering module is used for establishing a corresponding standard question-answering library based on standard terms in the term library, the combination question-answering module is used for establishing a sentence model based on the theme word library, combining subject words in the theme word library through the established sentence model, outputting a subject word set of the combination question-answering, carrying out value conversion of the subject word set to obtain a subject word set value, setting a context value of the corresponding combination question-answering, integrating the context value and the corresponding subject word set value to obtain a matching vector, mapping the matching vector into a vector space, and summarizing the current combination question-answering to establish the combination question-answering library; The judging module is used for judging whether the identified problem is correct or not, acquiring the identified user problem, converting the identified user problem into a problem vector, inputting the problem vector into a corresponding vector space, calculating the similarity between the problem vector and a corresponding matching vector, matching the corresponding combined question and answer according to the calculated similarity, acquiring a corresponding standard question and answer according to the matched combined question and answer, judging whether the user problem is non-judged according to the standard question and answer and the standard clause, and acquiring a corresponding judging result; the method for setting the context value of the corresponding combined question and answer comprises the following steps: Identifying corresponding subject word set values, labeled ZTi, wherein i represents a subject word, wherein i=1, 2, and i..4, n is a positive integer, identifying a standard question-answer corresponding to the combined question-answer, matching the corresponding standard added value, labeled BZ, setting a context adjustment coefficient of the combined question-answer, labeled α, and formulating according to the context value formula Calculating context values, wherein b1 and b2 are proportionality coefficients, and the value range is 0< b1 less than or equal to 1, and 0< b2 less than or equal to 1.
2. The term-library-based non-discriminant system of claim 1, wherein said means for operating comprises: Setting dictionary establishment rules, performing auditing format conversion to obtain standard limiting conditions, establishing a database and acquisition channels, acquiring data through the established acquisition channels, inputting the acquired data into the database for storage, marking the database as a material library, extracting subject terms from the data in the material library, marking the data as initial terms, checking the initial terms through the set standard limiting conditions, marking the initial terms successfully checked as subject terms, integrating the subject terms and establishing a subject word library.
3. The non-discriminant system of claim 1, wherein said standard question-answering module is operable to: Identifying standard clauses in a clause library, acquiring application directions of the standard clauses, setting corresponding problem search formulas based on the acquired application directions, searching the problems according to the set problem search formulas, obtaining a plurality of question-answer questions, removing duplication of the obtained question-answer questions, obtaining to-be-selected question-answers, screening the to-be-selected question-answers, obtaining standard question-answers, marking corresponding standard clause labels, integrating the obtained standard question-answers, and then establishing a standard question-answer library.
4. The non-discriminant system of claim 1, wherein said means for associating comprises: And identifying the combined questions and the standard questions and answers in the combined question and answer library, performing similarity calculation on the combined questions and answers and the standard questions and answers to obtain corresponding similarity, distributing the combined questions and answers to the corresponding standard questions and answers according to the similarity, marking corresponding subordinate labels, and establishing a subordinate question and answer table.
5. The term-library based non-discriminant system of claim 1, wherein said method of converting identified user questions to question vectors comprises: extracting keywords of a user problem to obtain a problem keyword set, assigning the obtained problem keyword set to obtain a problem keyword set assignment, setting a context value of the user problem, and integrating the problem keyword set into a corresponding problem vector.
6. The term-library-based non-discriminant system of claim 5, wherein said means for assigning said set of problem keywords comprises: matching the question keyword set with the subject words in the subject word stock to obtain corresponding assignments, marking the unmatched question keywords as similar keywords, and integrating the similar assignments corresponding to the similar keywords to the keyword set according to the matching of the similar keywords.
7. The non-discriminant system of claim 6, wherein said means for matching corresponding similarity assignments based on similarity keywords comprises: inputting similar keywords into a topic word library to perform word sense similarity calculation to obtain the highest similarity corresponding to the keywords, and assigning corresponding keywords to be marked as XS and FZ respectively according to a formula Calculating corresponding similar assignment, wherein b3 is a proportionality coefficient, the value range is 0< b3 less than or equal to 1, The coefficients are adjusted for similarity.

Description

Non-discriminant system based on clause library Technical Field The invention belongs to the technical field of electrical knowledge question and answer, and particularly relates to a non-discriminant system based on a clause library. Background Along with the rapid development of the electrical field, related professional knowledge and related standards are continuously updated, which brings a certain trouble to the work of related workers, especially for the workers who just enter the field or other non-field workers, the invention provides a non-discriminating system based on a clause library, which is used for helping corresponding users judge whether the corresponding problems are correct or not, and the corresponding clauses are what, and the great effort is required to search slowly, so that the development of the industry is unfavorable. Disclosure of Invention In order to solve the problems of the scheme, the invention provides a non-discriminant system based on a clause library. The aim of the invention can be achieved by the following technical scheme: The term library-based non-discrimination system comprises a word library module, a term library, a standard question-answering module, a combined question-answering module, a correlation module, a discrimination module and a server; The system comprises a term library module, a standard question-answering module, a sentence model, a context value setting module, a matching vector mapping module and a combination question-answering module, wherein the term library module is used for establishing a theme word library, storing corresponding standard terms through the term library, the standard question-answering module is used for establishing a corresponding standard question-answering library based on standard terms in the term library, the combination question-answering module is used for establishing a sentence model based on the theme word library, combining subject words in the theme word library through the established sentence model, outputting a subject word set of the combination question-answering, carrying out value conversion of the subject word set to obtain a subject word set value, setting a context value of the corresponding combination question-answering, integrating the context value and the corresponding subject word set value to obtain a matching vector, mapping the matching vector into a vector space, and summarizing the current combination question-answering to establish the combination question-answering library; The judging module is used for judging whether the identified problem is correct or not, acquiring the identified user problem, converting the identified user problem into a problem vector, inputting the problem vector into a corresponding vector space, calculating the similarity between the problem vector and a corresponding matching vector, matching the corresponding combined question and answer according to the calculated similarity, acquiring the corresponding standard question and answer according to the matched combined question and answer, judging the user problem in a non-judging mode according to the standard question and answer and the standard clause, and acquiring a corresponding judging result. Further, the working method of the word stock module comprises the following steps: Setting dictionary establishment rules, performing auditing format conversion to obtain standard limiting conditions, establishing a database and acquisition channels, acquiring data through the established acquisition channels, inputting the acquired data into the database for storage, marking the database as a material library, extracting subject terms from the data in the material library, marking the data as initial terms, checking the initial terms through the set standard limiting conditions, marking the initial terms successfully checked as subject terms, integrating the subject terms and establishing a subject word library. Further, the working method of the standard question-answering module comprises the following steps: Identifying standard clauses in a clause library, acquiring application directions of the standard clauses, setting corresponding problem search formulas based on the acquired application directions, searching the problems according to the set problem search formulas, obtaining a plurality of question-answer questions, removing duplication of the obtained question-answer questions, obtaining to-be-selected question-answers, screening the to-be-selected question-answers, obtaining standard question-answers, marking corresponding standard clause labels, integrating the obtained standard question-answers, and then establishing a standard question-answer library. Further, the method for setting the context value of the corresponding combined question and answer comprises the following steps: Identifying corresponding subject word set values, labeled ZTi, wherein i represents a subject word, wherein i=1, 2, and