CN-122021818-A - Knowledge structure determination method, device, equipment, medium and product of intelligent body

CN122021818ACN 122021818 ACN122021818 ACN 122021818ACN-122021818-A

Abstract

The embodiment of the disclosure discloses a knowledge structure determining method, device, equipment, medium and product of an intelligent agent, which comprise the steps of obtaining scene data and task data of a target intelligent agent in a current running task scene, inputting the scene data and the task data into an intelligent agent knowledge structure recommending model to obtain at least one recommended knowledge structure corresponding to the target intelligent agent, wherein the recommended knowledge structure is a knowledge structure defined in a preset knowledge structure base, determining a scoring result corresponding to the recommended knowledge structure by using a preset scoring algorithm, and determining a target knowledge structure adopted by the target intelligent agent in the current running task scene from the recommended knowledge structure according to the scoring result. According to the technical scheme, the accurate improvement of the skills of the intelligent agent is realized, and the capability of the intelligent agent for adapting to different industries and scenes is improved.

Inventors

CHEN JIEKUN
ZHONG WEIJUN
Zhao Kanghui
ZHOU JIA
ZHANG YI
SUN HAO
BAI GUOTAO

Assignees

中移信息技术有限公司
中国移动通信集团有限公司

Dates

Publication Date: 20260512
Application Date: 20260127

Claims (10)

1. A method for determining a knowledge structure of an agent, comprising: acquiring scene data and task data of a target intelligent agent in a current running task scene; Inputting the scene data and the task data into an agent knowledge structure recommendation model to obtain at least one recommendation knowledge structure corresponding to the target agent, wherein the recommendation knowledge structure is a knowledge structure defined in a preset knowledge structure library; determining a scoring result corresponding to the recommended knowledge structure by using a preset scoring algorithm; And determining a target knowledge structure adopted by the target intelligent agent in the current running task scene from the recommended knowledge structure according to the scoring result.
2. The method of claim 1, wherein the determining of the agent knowledge structure recommendation model comprises: determining a historical training data set based on a historical running log of the agent, the historical training data set comprising at least one historical training data, the historical training data comprising historical scene data and historical task data; For any one historical training data, determining a knowledge structure label corresponding to the historical training data according to an existing knowledge structure in a preset knowledge structure base; extracting scene characteristics of the historical scene data by using a first preset algorithm; extracting task characteristics of the historical task data by using a second preset algorithm; Splicing the scene features and the task features to determine joint features corresponding to the historical training data; Based on the corresponding joint characteristics and knowledge structure labels of each historical training data, a preset loss function is used as a training target, and model parameters of an initial knowledge structure recommendation model are adjusted by using a preset parameter adjustment algorithm so as to determine the knowledge structure recommendation model of the intelligent body.
3. The method according to claim 1, wherein the determining of the pre-set knowledge structure base comprises: determining a historical interaction data set based on the historical running log of the agent; performing cluster analysis on each historical interaction data in the historical interaction data set by using a preset clustering algorithm to determine a clustering result; and determining the preset knowledge structure base based on the clustering result, wherein the preset knowledge structure base comprises at least one knowledge structure.
4. The method of claim 3, wherein the determining a historical interaction dataset based on the historical travel log of the agent comprises: Determining each initial historical interaction data according to the historical operation log of the intelligent agent; Vectorizing each initial historical interaction data by using a text vector generation model to determine text vectors corresponding to each initial historical interaction data respectively; Calculating the similarity between the text vectors by using a similarity determination algorithm to determine a first similarity matrix; And merging all initial historical interaction data according to the first similarity matrix to determine the historical interaction data set.
5. The method of claim 4, wherein performing cluster analysis on each of the historical interaction data in the set of historical interaction data using a preset clustering algorithm to determine a cluster result comprises: Performing vectorization processing on the historical interaction data set by using the text vector generation model so as to determine text vectors corresponding to each historical interaction data respectively; and carrying out cluster analysis on each text vector by using a preset cluster algorithm to determine the cluster result.
6. The method of claim 5, wherein performing a clustering analysis on each text vector using a preset clustering algorithm to determine the clustering result comprises: calculating the similarity between the text vectors by using a similarity determination algorithm to determine a second similarity matrix; Clustering each text vector based on a second similarity matrix to determine an initial cluster set, wherein the initial cluster set comprises at least one initial cluster; calculating the inter-class distance between each initial cluster by using an inter-class distance determining algorithm to determine an inter-class distance matrix; Determining a target initial cluster according to the inter-class distance matrix, and merging the target initial clusters to update the initial cluster set; updating the inter-class distance matrix according to the updated initial cluster set; If the number of clusters in the updated initial cluster set does not reach the target cluster number, returning to the step of determining the target initial cluster according to the inter-class distance matrix until the number of clusters in the updated initial cluster set reaches the target cluster number, wherein the target cluster number is determined based on a preset evaluation rule and a preset evaluation index; And determining the updated initial cluster set as the clustering result.
7. A knowledge structure determination apparatus of an agent, comprising: the data acquisition module is used for acquiring scene data and task data of the target intelligent agent in a current running task scene; The recommendation knowledge structure determining module is used for inputting the scene data and the task data into an agent knowledge structure recommendation model to obtain at least one recommendation knowledge structure corresponding to the target agent, wherein the recommendation knowledge structure is a knowledge structure defined in a preset knowledge structure library; The scoring result determining module is used for determining scoring results corresponding to the recommended knowledge structure by using a preset scoring algorithm; And the knowledge structure determining module is used for determining a target knowledge structure adopted by the target intelligent agent in the current running task scene from the recommended knowledge structure according to the scoring result.
8. An electronic device, comprising: at least one processor, and A memory communicatively coupled to the at least one processor, wherein, The memory stores a computer program executable by the at least one processor to enable the at least one processor to perform the method of knowledge structure determination of an agent as claimed in any one of claims 1-6.
9. A computer-readable storage medium, on which a computer program is stored, characterized in that the program, when being executed by a processor, implements the method for determining the knowledge structure of an agent according to any one of claims 1-6.
10. A computer program product, characterized in that the computer program product comprises a computer program which, when executed by a processor, implements the method of knowledge structure determination of an agent according to any one of claims 1-6.

Description

Knowledge structure determination method, device, equipment, medium and product of intelligent body Technical Field The embodiment of the disclosure relates to the technical field of artificial intelligence and intelligent agents, in particular to a knowledge structure determining method, device, equipment, medium and product of an intelligent agent. Background Along with the application of large models related to the intelligent agent technology enabling industry, and particularly along with the maturation and popularization of intelligent agent innovation application platforms, more and more intelligent agents are created and integrated into the service, and rich use records and evaluations are accumulated. The current agent optimization technology mainly relies on manual experience adjustment or extensive optimization based on overall usage/evaluation, and lacks structural analysis means for the core thinking capability of the agent (multi-round session organization, task chain logic planning and task execution step design). The prior art mainly has the following limitations that the evaluation dimension is single, the traditional method optimizes the intelligent body through macroscopic indexes such as the overall satisfaction degree or click rate of a user, the specific thinking defects such as insufficient continuity of multiple rounds of conversations, redundant steps of a task chain and the like can not be positioned, the knowledge structure is fuzzy, the intelligent body skill text is mostly unstructured natural language, key semantic modules such as character definition, knowledge background, task instruction and illustration are difficult to automatically identify through a machine, the optimization efficiency is low, a great deal of manpower is consumed for manually analyzing massive interaction logs, and a reusable thinking structure template library is difficult to form. Disclosure of Invention The embodiment of the disclosure provides a knowledge structure determining method, device, equipment, medium and product of an intelligent agent, which realize the accurate improvement of the skill of the intelligent agent and promote the capability of the intelligent agent to adapt to different industries and scenes. In a first aspect, a method for determining a knowledge structure of an agent is provided, including: acquiring scene data and task data of a target intelligent agent in a current running task scene; Inputting the scene data and the task data into an agent knowledge structure recommendation model to obtain at least one recommendation knowledge structure corresponding to the target agent, wherein the recommendation knowledge structure is a knowledge structure defined in a preset knowledge structure library; determining a scoring result corresponding to the recommended knowledge structure by using a preset scoring algorithm; And determining a target knowledge structure adopted by the target intelligent agent in the current running task scene from the recommended knowledge structure according to the scoring result. In a second aspect, there is provided a knowledge structure determining apparatus of an agent, including: the data acquisition module is used for acquiring scene data and task data of the target intelligent agent in a current running task scene; The recommendation knowledge structure determining module is used for inputting the scene data and the task data into an agent knowledge structure recommendation model to obtain at least one recommendation knowledge structure corresponding to the target agent, wherein the recommendation knowledge structure is a knowledge structure defined in a preset knowledge structure library; The scoring result determining module is used for determining scoring results corresponding to the recommended knowledge structure by using a preset scoring algorithm; And the knowledge structure determining module is used for determining a target knowledge structure adopted by the target intelligent agent in the current running task scene from the recommended knowledge structure according to the scoring result. In a third aspect, an electronic device is provided, comprising: at least one processor, and A memory communicatively coupled to the at least one processor, wherein, The memory stores a computer program executable by the at least one processor to enable the at least one processor to perform the method of determining the knowledge structure of an agent as described in the first aspect above. In a fourth aspect, a computer-readable storage medium is provided, on which a computer program is stored, which program, when being executed by a processor, implements a method for determining a knowledge structure of an agent as described in the first aspect above. In a fifth aspect, a computer program product is provided, the computer program product comprising a computer program which, when executed by a processor, implements the method for determining the knowledge