CN-121524314-B - Knowledge question-answering method, device, medium, equipment and product based on large model

CN121524314BCN 121524314 BCN121524314 BCN 121524314BCN-121524314-B

Abstract

A knowledge question-answering method, a device, a medium, a device and a product based on a large model relate to the technical field of computers, the method determines the query category of the query intention corresponding to second content through a first large model, when the query category is the first query category, the first content and the second content are input into the second large model to obtain a first query result corresponding to a query request output by the second large model, when the first query type is the second query type, the first content and the second content are input into the first large model to obtain third content output by the first large model, and then a second query result corresponding to the query request is obtained according to the third content, so that not only can knowledge question and answer by a user through using the multi-mode query request be supported, but also the user can be enabled to be free from configuring a specific question and answer strategy. For complex questions, the first content can be effectively converted into retrievable text questions, so that the question asking efficiency of the user is greatly improved, and the answering accuracy is improved.

Inventors

TIAN XINYU
FENG JIAYU
Xiu Zongda
SHI XING
LI HAO
ZHAO SHUO
SHEN YUQI
JIA HUI
ZHANG LU
WANG HAO
LIN SEN
YU GUOSEN

Assignees

北京飞书科技有限公司

Dates

Publication Date: 20260505
Application Date: 20260114

Claims (10)

1. A knowledge question-answering method based on a large model, comprising: obtaining a query request, wherein the query request comprises first content and second content, the first content and the second content belong to different content types, and the second content is used for describing query intention; Determining a query category to which the query intention corresponding to the second content belongs through a first large model, wherein the query category comprises a first query category and a second query category, the first query category represents a query result corresponding to the second content generated through the first content, and the second query category represents a query result corresponding to the second content generated through knowledge retrieval; Responding to the query category corresponding to the second content as the first query category, and inputting the first content and the second content into a second large model to obtain a first query result corresponding to the query request output by the second large model; Responding to the query category corresponding to the second content as the second query category, inputting the first content and the second content into the first large model to obtain third content output by the first large model, wherein the first large model is used for adjusting the second content according to information identified from the first content to obtain the third content, and the content type of the third content is consistent with the content type of the second content; And obtaining a second query result corresponding to the query request according to the third content.
2. The method of claim 1, wherein determining, by the first large model, a query category to which the query intent corresponding to the second content belongs, comprises: Determining the query intention corresponding to the second content through a first large model; determining, in response to the query intent being of a search type, that a query category to which the second content belongs is the second query category; responsive to the query intent being a non-search type of intent, determining that the query category to which the second content belongs is the first query category.
3. The method of claim 2, wherein the determining that the query category to which the second content belongs is the second query category in response to the query intent being a search type intent comprises: Responding to the query intention as the search type intention, and determining a query object for which the query intention corresponding to the second content is aimed in the first content according to the second content; And determining that the query category to which the second content belongs is the first query category in response to the query object containing non-text information which cannot be converted into text.
4. A method according to any one of claims 1-3, wherein the obtaining, according to the third content, a second query result corresponding to the query request includes: checking whether the third content accords with a preset rule; responding to the third content to accord with a preset rule, and obtaining a second query result corresponding to the query request according to the third content; And responding to the third content which does not accord with the preset rule, inputting the first content and the second content into a second large model to obtain a third query result corresponding to the query request output by the second large model.
5. A method according to any one of claims 1-3, wherein the second large model is determined by: in response to a selection operation for a target large model of a plurality of candidate large models, the selected target large model is determined as the second large model.
6. A method according to any one of claims 1-3, wherein the determining, by the first large model, a query category to which the query intent corresponding to the second content belongs, comprises: Determining the request length corresponding to the query request; Responding to the request length being smaller than or equal to a preset threshold value, and determining the query category to which the query intention corresponding to the second content belongs through a first large model; Inputting the first content and the second content into a third large model to obtain fourth content output by the third large model in response to the request length being greater than the preset threshold, wherein the third large model is used for adjusting the second content according to information identified from the first content to obtain the fourth content, and the content type of the fourth content is consistent with the content type of the second content; and obtaining a fourth query result corresponding to the query request according to the fourth content.
7. A knowledge question-answering device based on a large model, comprising: The system comprises an acquisition module, a query module and a query module, wherein the acquisition module is used for acquiring a query request, the query request comprises first content and second content, the first content and the second content belong to different content types, and the second content is used for describing query intention; The first determining module is used for determining a query category to which the query intention corresponding to the second content belongs through a first large model, wherein the query category comprises a first query category and a second query category, the first query category represents a query result corresponding to the second content generated through the first content, and the second query category represents a query result corresponding to the second content generated through knowledge retrieval; The first response module is used for responding to the query category corresponding to the second content as the first query category, inputting the first content and the second content into a second large model, and obtaining a first query result corresponding to the query request output by the second large model; The second response module is used for responding to the query category corresponding to the second content as the second query category, inputting the first content and the second content into the first large model to obtain third content output by the first large model, wherein the first large model is used for adjusting the second content according to the information identified by the first content to obtain the third content, and the content type of the third content is consistent with the content type of the second content; and the second determining module is used for obtaining a second query result corresponding to the query request according to the third content.
8. A computer readable medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processing device, implements the method of any of claims 1-6.
9. An electronic device, comprising: A storage device having a computer program stored thereon; processing means for executing said computer program in said storage means to implement the method of any one of claims 1-6.
10. A computer program product comprising a computer program, characterized in that the computer program, when executed by a processor, implements the method of any of claims 1-6.

Description

Knowledge question-answering method, device, medium, equipment and product based on large model Technical Field The disclosure relates to the technical field of computers, in particular to a knowledge question-answering method, a knowledge question-answering device, a knowledge question-answering medium, knowledge question-answering equipment and knowledge question answering products based on a large model. Background In a knowledge question-answering scene, plain text input is generally used as the main principle, if a user needs to ask questions in combination with picture information, the user needs to manually describe the picture content, and therefore the problems of inaccurate description and complicated operation of the user exist. Although as technology evolves, question-and-answer systems are increasingly capable of supporting complex query requests containing multimodal data, the lack of ability to dynamically adjust answer strategies makes it difficult to guarantee relevance and accuracy of answer content. Disclosure of Invention This summary is provided to introduce a selection of concepts in a simplified form that are further described below in the detailed description. This summary is not intended to identify key features or essential features of the claimed subject matter, nor is it intended to be used to limit the scope of the claimed subject matter. In a first aspect, the present disclosure provides a knowledge question-answering method based on a large model, including: obtaining a query request, wherein the query request comprises first content and second content, the first content and the second content belong to different content types, and the second content is used for describing query intention; Determining a query category to which the query intention corresponding to the second content belongs through a first large model, wherein the query category comprises a first query category and a second query category, the first query category represents a query result corresponding to the second content generated through the first content, and the second query category represents a query result corresponding to the second content generated through knowledge retrieval; Responding to the query category corresponding to the second content as the first query category, and inputting the first content and the second content into a second large model to obtain a first query result corresponding to the query request output by the second large model; Responding to the query category corresponding to the second content as the second query category, inputting the first content and the second content into the first large model to obtain third content output by the first large model, wherein the first large model is used for adjusting the second content according to information identified from the first content to obtain the third content, and the content type of the third content is consistent with the content type of the second content; And obtaining a second query result corresponding to the query request according to the third content. In a second aspect, the present disclosure provides a knowledge question-answering apparatus based on a large model, including: The system comprises an acquisition module, a query module and a query module, wherein the acquisition module is used for acquiring a query request, the query request comprises first content and second content, the first content and the second content belong to different content types, and the second content is used for describing query intention; The first determining module is used for determining a query category to which the query intention corresponding to the second content belongs through a first large model, wherein the query category comprises a first query category and a second query category, the first query category represents a query result corresponding to the second content generated through the first content, and the second query category represents a query result corresponding to the second content generated through knowledge retrieval; The first response module is used for responding to the query category corresponding to the second content as the first query category, inputting the first content and the second content into a second large model, and obtaining a first query result corresponding to the query request output by the second large model; The second response module is used for responding to the query category corresponding to the second content as the second query category, inputting the first content and the second content into the first large model to obtain third content output by the first large model, wherein the first large model is used for adjusting the second content according to the information identified by the first content to obtain the third content, and the content type of the third content is consistent with the content type of the second content; and the second determining module is used for obtain