Search

CN-122022177-A - Enterprise management method based on artificial intelligence

CN122022177ACN 122022177 ACN122022177 ACN 122022177ACN-122022177-A

Abstract

The invention discloses an enterprise management method based on artificial intelligence, which comprises a rule data acquisition and structuring step for acquiring latest rule, standard and policy information, a business data access step for accessing an enterprise internal business system and acquiring business data, a compliance knowledge graph construction and reasoning step for carrying out entity extraction and relation mapping on structured rule data and preprocessed business data to construct a field compliance knowledge graph, a risk identification and early warning step for checking the business data based on the compliance knowledge graph and a machine learning model to realize risk identification, a task allocation and modification tracking step for automatically generating modification tasks based on inference results of the knowledge graph, and a report generation step for generating compliance reports according to supervision requirements and internal management requirements.

Inventors

  • YAN HAISHUI

Assignees

  • 中合数联(苏州)科技有限公司

Dates

Publication Date
20260512
Application Date
20260203

Claims (10)

  1. 1. An enterprise management method based on artificial intelligence is characterized by comprising the following steps: A regulation data acquisition and structuring step for acquiring the latest regulation, standard and policy information; a service data access step for accessing an enterprise internal service system and obtaining service data; a compliance knowledge graph construction and reasoning step, which is used for carrying out entity extraction and relation mapping on structured regulation data and preprocessed business data to construct a field compliance knowledge graph; a risk identification and early warning step, namely verifying service data based on a compliance knowledge graph and a machine learning model to realize risk identification; Task assignment and rectification tracking, namely automatically generating rectification tasks comprising problem description, rectification advice and responsibility main body according to regulation clauses based on the reasoning result of the knowledge graph; and a report generation step of generating a compliance report according to the supervision requirement and the internal management requirement.
  2. 2. The method for enterprise management based on artificial intelligence of claim 1, wherein in the step of collecting and structuring the rule data, the unstructured rule text is subjected to text extraction and chapter division by using a natural language processing technology and a sequence labeling model, and term numbers, applicable subjects, penalty regulations and effective date elements are extracted with high precision to convert the unstructured text into structured data.
  3. 3. The method for enterprise management based on artificial intelligence of claim 2, wherein in the step of collecting and structuring the regulatory data, specifically: writing a data acquisition program script to periodically access government websites and analyzing legal contents in an HTML or PDF file; Text extraction and chapter division are carried out by using a sequence labeling model based on a Transformer, and named entity identification is carried out on the rule and regulation, so that clause numbers, application ranges and punishment regulations are extracted with high precision; Storing the parsed rule entries into a rule base in a structured data format, and recording sources, release dates and update dates.
  4. 4. The enterprise management method based on artificial intelligence of claim 1, wherein in the service data accessing step, the related service data of the enterprise internal service system is accessed in real time or at regular time through a security API or a database interface, and field mapping, data cleaning and anonymization processing are performed.
  5. 5. The artificial intelligence based enterprise management method of claim 1, The entities include regulatory clauses, business activities, liability bodies, risk points, and penalty types; the relationship includes various keywords associated with the relationship statement.
  6. 6. The method for enterprise management based on artificial intelligence of claim 5, wherein in the step of building and reasoning the compliance knowledge graph, specifically: Extracting rule terms, business process steps, responsibility bodies and punishment types from rule treaty and business data by using named entity recognition technology; Identifying the relationship among the entities by using a relationship extraction model, and constructing a triplet; And storing the entity and the relation by using a graph database Neo4j to form an inferable compliance knowledge graph.
  7. 7. The method for enterprise management based on artificial intelligence of claim 1, wherein in the risk identification and pre-warning step, specifically: According to mandatory and deterministic compliance requirements defined in the knowledge graph, carrying out logic comparison on service data through graph path reasoning, and identifying dominant violations; Training a risk scoring model based on the historical violation records and the business data, and outputting risk probability according to business behaviors; and carrying out weighted fusion on the identified explicit violation result and the risk probability obtained by the risk scoring model to obtain a final unified risk level, and automatically generating and pushing an early warning when the risk level exceeds a preset threshold value.
  8. 8. The artificial intelligence based enterprise management method of claim 7, wherein the training a risk scoring model based on historical violation records and business data and outputting risk probabilities based on business behavior, in particular: Extracting business data characteristics and characteristic vectors based on a knowledge graph; Training a risk scoring model based on the history violation records and the business data; And a risk calculating step, wherein the risk scoring model outputs the risk probability of the business event.
  9. 9. The method for enterprise management based on artificial intelligence of claim 1, wherein in the task assignment and modification tracking step, specifically, when risk pre-warning is generated, the system automatically generates modification tasks based on the reasoning result of the knowledge graph, including problem description, modification suggestion and completion period, and the workflow engine drives task circulation to realize closed-loop recording and tracking of task states.
  10. 10. The artificial intelligence based enterprise management method of claim 1, further comprising a question-answering module comprising the following elements: the vector retrieval unit is used for carrying out semantic coding on the user question by using a vectorization technology, and retrieving similar questions and standard answers from a compliance question-answer corpus; The knowledge graph reasoning unit is used for calling the knowledge graph reasoning unit when the problem relates to complex logic or requires legal basis, searching related legal terms and business processes through a graph path and generating an interpretable answer; And the large language model auxiliary unit is used for calling the large language model to answer and marking the source of the regulation clause of the reference if the vector retrieval unit and the knowledge graph reasoning unit cannot solve the problem.

Description

Enterprise management method based on artificial intelligence Technical Field The invention relates to the technical field of enterprise operation management and information, in particular to an enterprise management method based on artificial intelligence. Background With the increasing complexity of the global regulatory environment, businesses need to follow a large number of laws and regulations, industry standards, and internal regulations during business processes. Traditional compliance management relies mainly on manual search regulations, recording compliance matters and follow-up corrective measures, and is low in efficiency and prone to error. In recent years, some compliance management software based on rule engines or keyword matching has appeared on the market (i.e., the closest prior art). The system compares the business data of the enterprises through a preset hard rule or keyword library so as to identify potential illegal behaviors. However, such prior art solutions still have the following significant technical drawbacks: 1. The semantic relevance is poor, the risk identification accuracy is low, the prior art mainly relies on manually maintained rules and keywords, and the complex semantics of the rule and the regulation and the deep association between the rule and the specific business process of an enterprise are difficult to understand. This results in a high false positive rate (misjudging compliance as a risk) and a high false negative rate (missing potential violations) when risk comparison is performed. 2. The dynamic adaptability is weak, the knowledge updating efficiency is low, the supervision policies and industry standards are updated frequently, and the existing rule engine needs to modify and maintain the rule base manually one by one. The non-automatic and hysteresis updating mechanism cannot guarantee the real-time adaptability of the system to the latest supervision requirements. 3. The system lacks of deep reasoning capability and cannot perform pre-warning, and most of the existing systems audit afterwards and can only simply match the generated business data. Due to the lack of technical means for structuring and reasoning-associating legal knowledge with business data, real-time and prospective risk early warning is difficult to carry out in the business handling process. Therefore, an intelligent compliance management system and method capable of realizing automation and structured extraction of legal knowledge and constructing a inferable knowledge graph by utilizing an artificial intelligence technology are urgently needed, so that the compliance information collection efficiency, the risk identification accuracy and the rectifying and tracking closed-loop capability are greatly improved. Disclosure of Invention The technical problem solved by the invention is to provide an enterprise management method based on artificial intelligence, which can greatly improve the collection efficiency of compliance information, the accuracy of risk identification and the capability of rectifying and modifying tracking closed loops. The technical scheme adopted by the invention for solving the technical problems is that the enterprise management method based on artificial intelligence comprises the following steps: A regulation data acquisition and structuring step for acquiring the latest regulation, standard and policy information; a service data access step for accessing an enterprise internal service system and obtaining service data; a compliance knowledge graph construction and reasoning step, which is used for carrying out entity extraction and relation mapping on structured regulation data and preprocessed business data to construct a field compliance knowledge graph; a risk identification and early warning step, namely verifying service data based on a compliance knowledge graph and a machine learning model to realize risk identification; Task assignment and rectification tracking, namely automatically generating rectification tasks comprising problem description, rectification advice and responsibility main body according to regulation clauses based on the reasoning result of the knowledge graph; and a report generation step of generating a compliance report according to the supervision requirement and the internal management requirement. The method comprises the steps of collecting and structuring rule data, namely, extracting text and chapter division of unstructured rule texts by using a natural language processing technology and a sequence labeling model, extracting clause numbers, applicable bodies, punishment regulations and effective date elements with high precision, and converting the unstructured texts into structured data. Further, in the step of the regulation data acquisition and structuring, the method specifically comprises the following steps: writing a data acquisition program script to periodically access government websites and analyzing legal conte