CN-121998074-A - Dual-path generation method, medium and system based on elastic difficulty threshold decision

CN121998074ACN 121998074 ACN121998074 ACN 121998074ACN-121998074-A

Abstract

The invention relates to a dual-path generating method, medium and system based on elastic difficulty threshold decision, which are characterized in that the difficulty coefficient of a query problem is determined according to the characteristic parameter by acquiring the current query problem of a user and extracting the characteristic parameter of the query problem, the difficulty threshold is dynamically updated according to the characteristic index of the characteristic parameter or/and the operation index of a retrieval thinking module and an reasoning thinking module, whether the difficulty coefficient exceeds the difficulty threshold is judged, the retrieval thinking module or the reasoning thinking module is selectively called according to the judgment result, and the query result is output. The problems of unreasonable resource allocation, low query efficiency, insufficient accuracy and the like in the prior art in a complex task scene are solved.

Inventors

NIE XIAOYI
ZHANG HAITAO
ZHANG JIAYI
Shu Jiajin
YU JIAN

Assignees

湖南农业大学

Dates

Publication Date: 20260508
Application Date: 20251111

Claims (10)

1. A dual path generation method based on elastic difficulty threshold decision, comprising: acquiring a current query problem of a user, and extracting characteristic parameters of the query problem; determining a difficulty coefficient of the query problem according to the characteristic parameters; Dynamically updating the difficulty threshold according to the characteristic indexes of the characteristic parameters or/and the operation indexes of the searching thinking module and the reasoning thinking module; Judging whether the difficulty coefficient exceeds a difficulty threshold, selectively calling a retrieval thinking module or an inference thinking module according to the judging result, and outputting a query result.
2. The method according to claim 1, wherein the step of dynamically updating the difficulty threshold according to the characteristic index of the characteristic parameter, or/and the operation index of the searching thinking module and the reasoning thinking module comprises: The operation indexes comprise accuracy and delay rate; acquiring the accuracy and the delay rate of each module in a set time period, and obtaining an accuracy index and a delay rate index; Obtaining the difference value between the accuracy indexes of each module and the difference value between the delay indexes of each module to obtain an accuracy difference value and a delay rate difference value; Obtaining the maximum value in each delay rate index as a normalization parameter to normalize the delay rate difference value to obtain a normalization value; and summing the accuracy difference value and the normalization value according to the prior accuracy coefficient and the time coefficient to obtain an updated difficulty threshold.
3. The method according to claim 2, wherein the step of dynamically updating the difficulty threshold according to the characteristic index of the characteristic parameter, or/and the operation index of the searching thinking module and the reasoning thinking module, further comprises: acquiring a current difficulty threshold value, setting a buffer coefficient and a buffer reference value; obtaining the product of the set buffer coefficient and the set buffer reference value as a buffer threshold; subtracting the buffer threshold from the current difficulty threshold to obtain a first difficulty threshold, and adding the buffer threshold to the current difficulty threshold to obtain a second difficulty threshold.
4. A method according to claim 3, wherein the step of determining whether the difficulty factor exceeds a difficulty threshold, selectively invoking a search thinking module or an inference thinking module based on the determination result, and outputting a query result comprises: When the difficulty coefficient is not smaller than the first difficulty threshold value and not larger than the second difficulty threshold value, manually judging whether the corresponding query problem is an easy problem or a difficult problem; According to the classification result of each inquiry problem, the easy problem is input into a searching thinking module, the difficult problem is input into an reasoning thinking module, and the inquiry result is output.
5. The method according to claim 1, wherein the step of dynamically updating the difficulty threshold according to the characteristic index of the characteristic parameter, or/and the operation index of the searching thinking module and the reasoning thinking module, further comprises: acquiring the historical accuracy of each module to obtain an updated difficulty threshold according to the historical accuracy and the difficulty threshold; Judging whether the current load of the retrieval thinking module is larger than a corresponding safe load threshold value, if not, keeping unchanged, if so, determining a temporary floating threshold value, and subtracting the temporary floating threshold value from the updated difficulty threshold value to obtain a floating difficulty threshold value; Judging whether the current load of the reasoning thinking module is larger than a corresponding safe load threshold, if not, keeping unchanged, if so, determining a temporary floating threshold, and adding the temporary floating threshold to the updated difficulty threshold to obtain a floating difficulty threshold.
6. The method of claim 5, wherein the step of obtaining the historical accuracy of each module to obtain an updated difficulty threshold based on the historical accuracy and the difficulty threshold comprises: acquiring the historical accuracy of each module, and obtaining an accuracy difference value between each module; obtaining the product of the accuracy difference and the learning coefficient to obtain a threshold correction; and summing the current difficulty threshold and the threshold correction amount to obtain an updated difficulty threshold.
7. The method of claim 5, wherein the step of determining a temporary float threshold comprises: obtaining the maximum load and the safety coefficient of each module, and calculating the product of the maximum load and the safety coefficient of each module to obtain the safety load of each module; And obtaining the difference value between the current load and the safety load of each module, and calculating the product of the priori sensitivity coefficient and each difference value to obtain the temporary floating threshold value.
8. The method according to any one of claims 1-7, further comprising: acquiring and setting monitoring index data in real time in the running process of each module, and obtaining the current abnormal mode type according to the priori judging condition, wherein the abnormal mode type comprises any one or more of characteristic drift, performance attenuation, engine failure and system fusing; Obtaining an abnormal grade corresponding to the current abnormal mode type, and adopting corresponding priori repair operation according to the abnormal grade, wherein the priori repair operation comprises any one or more of automatic repair, degradation operation and fusing treatment.
9. A computer storage medium, characterized in that executable program code is stored for performing the method for dual path generation based on elastic difficulty threshold decisions according to any of claims 1-8.
10. A terminal system comprising a memory and a processor, wherein the memory stores program code executable by the processor, and wherein the program code is configured to perform the method for generating a dual path based on elastic difficulty threshold decisions as set forth in any one of claims 1-8.

Description

Dual-path generation method, medium and system based on elastic difficulty threshold decision Technical Field The present invention relates to the field of information retrieval technologies, and in particular, to a dual-path generating method, medium, and system based on elastic difficulty threshold decision. Background Retrieval thinking modules, such as retrieval enhanced large language models (RETRIEVAL-Augmented Generation, RAG), have become the dominant mode of question-answer interactions by users through natural language. However, this technique presents a number of core challenges in the application process. In one aspect, user queries are ubiquitous with spoken language expressions, ambiguous descriptions, and contextual ambiguity issues. Especially in complex task scenes, such as scenes requiring multiple rounds of information integration, deep logic analysis or relying on domain expertise, the traditional RAG system is difficult to accurately capture user intention, so that the retrieval result deviates from the actual requirement, and the expertise and comprehensiveness of answer generation are obviously limited. On the other hand, the widely adopted query rewrite technology can optimize input expression through semantic refining, but the existing method depends on a single rewrite strategy and has limitations. The method can not effectively analyze the multi-level characteristics of the user semantics and is difficult to realize balance among multi-dimensional targets. This limitation has led to resolution accuracy of complex problems of less than 70% over long periods of time, and often results in logic breaks or knowledge conflicts. Under the situation, an inference thinking module is generated, based on the human inference process, a plurality of intelligent agents are created, and various thinking templates, such as transverse thinking, sequential thinking, criticizing thinking, integration thinking and the like, are combined to excite potential thinking capability of a language model or other intelligent systems, so that the inference problem is solved, and the resolution accuracy of the complex problem is greatly improved. But in the face of simple problems, the problem handling efficiency is reduced. How to balance the time delay rate and the problem accuracy rate and selectively select the two is a hot spot and a difficult point of research in recent years. The Chinese patent application with the application number 202510092762.7 discloses an answer generation method, a knowledge question-answering system and electronic equipment, and discloses a technical scheme for generating answers to query questions by selecting corresponding search and answer generation strategies according to complexity levels corresponding to the query questions, but the current running condition of each generation module is not considered, so that the problem that the load of a certain generation module is overlarge due to the fact that the number of query questions of a certain complexity is too large cannot be avoided. How to balance the resource allocation of different methods to improve the query efficiency and reduce the resource waste, and to improve the problems of unreasonable resource allocation, low query efficiency, insufficient accuracy and the like of the dynamic adjustment in the complex task scene in the prior art, is a technical problem to be solved in the field. Disclosure of Invention Based on the above, the present application aims to provide a dual-path generating method, medium and system based on elastic difficulty threshold decision, so as to solve at least one technical problem mentioned in the above background art. In a first aspect, the present application provides a dual-path generation method based on elastic difficulty threshold decision, including: acquiring a current query problem of a user, and extracting characteristic parameters of the query problem; determining a difficulty coefficient of the query problem according to the characteristic parameters; Dynamically updating the difficulty threshold according to the characteristic indexes of the characteristic parameters or/and the operation indexes of the searching thinking module and the reasoning thinking module; Judging whether the difficulty coefficient exceeds a difficulty threshold, selectively calling a retrieval thinking module or an inference thinking module according to the judging result, and outputting a query result. Further, the step of dynamically updating the difficulty threshold according to the characteristic index of the characteristic parameter or/and the operation index of the searching thinking module and the reasoning thinking module comprises the following steps: The operation indexes comprise accuracy and delay rate; acquiring the accuracy and the delay rate of each module in a set time period, and obtaining an accuracy index and a delay rate index; Obtaining the difference value between the accuracy inde