CN-121599072-B - User knowledge acquisition method, device, equipment and medium based on artificial intelligence

CN121599072BCN 121599072 BCN121599072 BCN 121599072BCN-121599072-B

Abstract

The invention discloses a user knowledge acquisition method, device, equipment and medium based on artificial intelligence, wherein the method comprises the steps of acquiring current interaction task information corresponding to user interaction instructions, acquiring a knowledge engine temporary library from a knowledge engine according to a task scheduling strategy and the current interaction task information, acquiring current user input data corresponding to the current interaction task information, acquiring current target knowledge data from the knowledge engine temporary library, the knowledge engine or a server background according to the current user input data and the current interaction task type, and storing the current user input data and the current target knowledge data into a corresponding user data storage space if an intelligent interaction ending instruction corresponding to the current interaction task information is detected. According to the embodiment of the invention, in the process of intelligent interaction between the user terminal and the server, the target knowledge data can be more accurately acquired by the current user input data and the interaction scene corresponding to the current interaction task type, and the acquisition mode of the knowledge data is expanded.

Inventors

YANG XIUYUAN
Zeng Xiankang
CHEN FAN

Assignees

深圳知鸟教育科技有限公司

Dates

Publication Date: 20260508
Application Date: 20260130

Claims (8)

1. An artificial intelligence based user knowledge acquisition method, comprising: responding to a user interaction instruction sent by a user terminal, and acquiring current interaction task information corresponding to the user interaction instruction, wherein the current interaction task information at least comprises current user authorization information and current interaction task type; acquiring a corresponding knowledge engine temporary library from a local pre-constructed knowledge engine according to a preset task scheduling strategy and the current interaction task information; acquiring current user input data corresponding to the current interaction task information, and acquiring current target knowledge data from the knowledge engine temporary library, the knowledge engine or a server background according to the current user input data and the current interaction task type, wherein a preset interaction task type set corresponding to the current interaction task type at least comprises an intelligent learning scene type and an intelligent partner training scene type, and the current interaction task type is one of the preset interaction task type sets; If an intelligent interaction ending instruction corresponding to the current interaction task information is detected, storing the current user input data and the current target knowledge data into a user data storage space corresponding to the current interaction task information; Before the step of responding to the user interaction instruction sent by the user terminal and acquiring the current interaction task information corresponding to the user interaction instruction or before the step of acquiring the corresponding knowledge engine temporary library from the local pre-built knowledge engine according to the preset task scheduling strategy and the current interaction task information, the method further comprises the following steps: if the current uploading document file is detected and the current uploading document file is determined to be a non-repeated document file, acquiring document character layout, formula content, table content and chart content in the current uploading document file based on a preset file analysis strategy, and detecting to obtain a current document extraction result; performing field knowledge enhancement, structural fragmentation and document vectorization data acquisition on the current uploaded document file based on a preset file reconstruction strategy to obtain a current file reconstruction processing result; The current document extraction result, the current document reconstruction processing result and the current uploading document file are stored in a knowledge base after a mapping relation is established, so that the knowledge base is updated; The obtaining current target knowledge data from the knowledge engine temporary library, the knowledge engine or a server background according to the current user input data and the current interaction task type comprises the following steps: if the current interaction task type in the current interaction task information is determined to be the intelligent learning scene type, switching to a first current interaction interface corresponding to the intelligent learning scene type; acquiring current user input data input on the first current interactive interface; Performing intention recognition on the current user input data according to a pre-deployed intention recognition agent to obtain a current intention recognition result, wherein an intention recognition model is deployed in the intention recognition agent; Determining a current target data acquisition area according to the current interaction type corresponding to the current intention recognition result, and acquiring the current target knowledge data from the current target data acquisition area according to the current user input data and a preset knowledge data acquisition strategy; The obtaining current target knowledge data from the knowledge engine temporary library, the knowledge engine or a server background according to the current user input data and the current interaction task type comprises the following steps: if the current interaction task type in the current interaction task information is determined to be the intelligent partner training scene type, switching to a second current interaction interface corresponding to the intelligent partner training scene type; Starting a pre-built digital human model, and building an association relationship between the digital human model and the knowledge base; If the current user input data input by the user on the second current interactive interface is detected, acquiring current target knowledge data from the knowledge base through a knowledge data acquisition strategy in the digital human model, and sending the current target knowledge data to the user terminal; and if the ending interaction instruction input by the user on the second current interaction interface is detected, returning to the initial interface of the second current interaction interface.
2. The method according to claim 1, wherein the determining a current target data acquisition area according to the current interaction type corresponding to the current intention recognition result, and acquiring the current target knowledge data from the current target data acquisition area according to the current user input data and a preset knowledge data acquisition policy, includes: If the current intention recognition result is determined to be the intention recognition result of a common problem solving type, using the knowledge engine or the knowledge engine temporary library as the current target data acquisition area, acquiring first candidate knowledge data, of which the similarity with the current user input data exceeds a preset similarity threshold value, from the current target data acquisition area according to the knowledge data acquisition strategy, and using first candidate answer data corresponding to the first candidate knowledge data as the current target knowledge data; If the current intention recognition result is determined to be the intention recognition result of the knowledge question-answering type, using the knowledge engine or the knowledge engine temporary library as the current target data acquisition area, acquiring a current thinking chain disassembly result corresponding to the current user input data according to the knowledge data acquisition strategy, acquiring a thinking chain reply result corresponding to each piece of current thinking chain data in the current thinking chain disassembly result from the current target data acquisition area, and forming the current target knowledge data; And if the current intention recognition result is determined to be the intention recognition result of the appointed route type, acquiring a current route keyword corresponding to the current user input data, taking the server background as the current target data acquisition area, and acquiring the current target knowledge data corresponding to the previous route keyword from the current target data acquisition area according to the knowledge data acquisition strategy.
3. The method according to claim 2, wherein the obtaining a current mental chain disassembly result corresponding to the current user input data according to the knowledge data obtaining policy, and obtaining a mental chain reply result corresponding to each of the current mental chain disassembly results from the current target data obtaining area, comprises: Performing mental chain disassembly on the current user input data according to a mental chain disassembly sub-strategy in the knowledge data acquisition strategy to obtain a current mental chain disassembly result comprising a plurality of current mental chain data; And calling a local pre-deployed question answer intelligent agent, and inputting the current thinking chain disassembly result as a prompt word to the question answer intelligent agent to obtain a thinking chain reply result corresponding to each piece of current thinking chain data in the current thinking chain disassembly result.
4. The method according to claim 1, further comprising, after the step of returning to the initial interface of the second current interactive interface if the end interactive instruction input by the user on the second current interactive interface is detected: acquiring a plurality of current user input data and corresponding current target knowledge data which are interactively stored with the digital human model in a plurality of rounds, so as to form current intelligent coside comprehensive data according to a data acquisition time sequence; And acquiring standard intelligent coside comprehensive data corresponding to the current intelligent coside comprehensive data from the knowledge base, and determining semantic similarity between the current intelligent coside comprehensive data and the standard intelligent coside comprehensive data to serve as a current round coside evaluation result corresponding to the current intelligent coside comprehensive data.
5. The method according to any one of claims 1 to 4, further comprising, before the step of acquiring current interaction task information corresponding to a user interaction instruction in response to the user interaction instruction sent by the user terminal, or before the step of acquiring a corresponding knowledge engine temporary library from a locally pre-built knowledge engine according to a preset task scheduling policy and the current interaction task information: if the current uploading document file is detected and the current uploading document file is determined to be a non-repeated document file, carrying out knowledge extraction and knowledge graph construction on the current uploading document file based on a pre-trained knowledge extraction model to obtain current knowledge graph data; And acquiring stored knowledge-graph data, and fusing the current knowledge-graph data to the stored knowledge-graph data to update the stored knowledge-graph data.
6. An artificial intelligence based user knowledge acquisition device comprising means for performing the artificial intelligence based user knowledge acquisition method as claimed in any of claims 1-5.
7. A computer device comprising a memory and a processor, wherein the memory has stored thereon a computer program, and wherein the processor implements the artificial intelligence based user knowledge acquisition method of any of claims 1-5 when the computer program is executed.
8. A computer readable storage medium, characterized in that the computer readable storage medium stores a computer program comprising program instructions which, when executed by a processor, can implement the artificial intelligence based user knowledge acquisition method according to any one of claims 1-5.

Description

User knowledge acquisition method, device, equipment and medium based on artificial intelligence Technical Field The present invention relates to the field of artificial intelligence technologies, and in particular, to a method, an apparatus, a device, and a medium for obtaining user knowledge based on artificial intelligence. Background With the continuous development of enterprise technology and organization scale, knowledge documents are continuously accumulated and stored in servers or data center tables inside enterprises for staff users with knowledge acquisition requirements to acquire and review. When knowledge documents are stored in a current server or data center, limited information such as document names, document uploading information, document uploading time and the like of each knowledge document are generally extracted. And when the user logs in the server or the data center station and needs to search and acquire the needed knowledge data or knowledge document, a plurality of keywords can be input as search conditions to acquire the related target knowledge data or target knowledge document in the server or the data center station. However, in the method for acquiring the target knowledge data or the target knowledge document based on the keyword retrieval, all knowledge data in a server or a data center station need to be traversed, so that the retrieval result is less efficient to acquire. After that, knowledge acquisition modes based on artificial intelligence technology appear, namely, enterprises draw and tune experts from each professional field and management field, then extract, mark and extract knowledge from files generated in the production and management process, convert unstructured data into structured data, and store the structured data in a digital system of the enterprises. The knowledge is then sorted, aggregated and categorized and a knowledge graph is built by conventional knowledge extraction methods such as NLP (collectively Natural Language Processing and representing natural language processing) technology, OCR model (collectively Optical Character Recognition and representing optical character recognition), BERT model (collectively Bidirectional Encoder Representations from Transformers, representing a transducer-based bi-directional encoder), keyword normalization, rules engine, and knowledge graph. Finally, enterprise staff obtain the required knowledge data through keyword search and conditional query. However, the deposition of enterprise knowledge data in the above manner involves subsequent applications, and has the following drawbacks: 1) Enterprises need to systematically deposit files in a uniform format in a purposeful and systematic manner in the production management process, and then an expert can conveniently and rapidly mark the files and then extract knowledge so as to acquire knowledge data, and the novel document processing technology or the novel knowledge extraction method is lack of innovative exploration in the mode, such as document processing or knowledge extraction by combining technologies such as multi-mode learning, self-supervision learning or generation type artificial intelligence; 2) The knowledge acquisition mode mainly depends on strategies such as similarity matching, keyword routing and the like, and the related knowledge data cannot be acquired from a server according to the current knowledge acquisition scene of the user, but can only be acquired in a simple dialogue question-answering mode. Disclosure of Invention The embodiment of the invention provides a user knowledge acquisition method, device, equipment and medium based on artificial intelligence, which aim to solve the problem that in the prior art, the acquisition of related knowledge data cannot be carried out from a server according to the corresponding of the current knowledge acquisition scene of a user, but the acquisition of the related knowledge data can only be carried out in a simple dialogue question-answering mode. In a first aspect, an embodiment of the present invention provides a method for obtaining user knowledge based on artificial intelligence, including: responding to a user interaction instruction sent by a user terminal, and acquiring current interaction task information corresponding to the user interaction instruction, wherein the current interaction task information at least comprises current user authorization information and current interaction task type; acquiring a corresponding knowledge engine temporary library from a local pre-constructed knowledge engine according to a preset task scheduling strategy and the current interaction task information; acquiring current user input data corresponding to the current interaction task information, and acquiring current target knowledge data from the knowledge engine temporary library, the knowledge engine or a server background according to the current user input data and the current interaction tas