Natural language processing-based compliance task generation method and system

Abstract

The invention relates to a compliance task generation method and system based on natural language processing. The method comprises the following steps: a pre-construction step: pre-constructing a label matrix system; an inputting step: inputting a to-be-processed text; a text preprocessing step: carrying out cleaning and word segmentation processing on the to-be-processed text; syntactic analysis: performing syntactic analysis on the text subjected to word segmentation processing to identify a syntactic structure of a sentence; an entity identification step: extracting an entity based on the grammatical structure of the identified sentence; a relation extraction step: extracting an association relation between the entities based on the grammatical structure of the recognized sentence and the extracted entities; and a compliance task generation step: generating a compliance task according to the pre-constructed label matrix system, the entities extracted in the entity identification step and the association relationship extracted in the relationship extraction step. According to the invention, the rule terms can be efficiently and accurately converted into compliance tasks.

Description

Technical Field

[0001] The present invention relates to computer information technology, and in particular to a compliance task generation method and generation system based on natural language processing. Background Art

[0002] Compliance and safety are the prerequisites for the development of securities and futures business. The "regulations" in compliance refer to the laws, regulations and related provisions that securities and futures operating institutions need to follow. When processing legal and regulatory clauses and generating compliance tasks, existing technologies mostly rely on manual operations, which are inefficient and prone to omissions.

[0003] For example, in the existing technology, manual analysis and management are mainly used, such as legal experts or compliance personnel manually reading and analyzing legal and regulatory provisions to generate corresponding compliance tasks. The disadvantages are that it is time-consuming and labor-intensive, prone to errors, and unable to be updated and adapted to changes in laws and regulations in a timely manner.

[0004] Some existing technologies also use natural language processing technology to assist in analysis. Generally, general NLP tools are used to perform basic processing on legal and regulatory texts, such as word segmentation, part-of-speech tagging, and entity recognition. However, there are still certain defects, such as the lack of optimization for legal and regulatory clauses, low accuracy and precision, and still require a lot of manual intervention.

[0005] On the other hand, the existing technology also adopts a rule-driven parsing system, for example, parsing legal and regulatory clauses through predefined rules and templates. The disadvantages are that the rule maintenance cost is high, it is difficult to deal with complex and changeable legal clauses, and the flexibility is insufficient.

[0006] Therefore, with the increasing complexity of laws and regulations and the improvement of corporate compliance requirements, there is an urgent need for an automated and intelligent system to assist in the parsing of legal clauses and the generation and management of compliance tasks. Summary of the invention

[0007] In order to solve the above problems in the prior art, the present invention provides a compliance task generation method and generation system based on natural language processing, which can efficiently and accurately convert rule clauses into compliance tasks.

[0008] A compliance task generation method based on natural language processing according to one aspect of the present invention comprises the following steps: Pre-construction step, pre-constructing the label matrix system; Input step: input the text to be processed; A text preprocessing step, cleaning and segmenting the text to be processed; Syntactic analysis step: syntactic analysis is performed on the text after word segmentation to identify the grammatical structure of the sentence; An entity recognition step of extracting entities based on the grammatical structure of the recognized sentence; A relationship extraction step, based on the grammatical structure of the identified sentence and the extracted entities, extracting the association relationship between entities; and The compliance task generation step generates a compliance task based on the pre-constructed label matrix system, the entities extracted by the entity recognition step, and the association relationships extracted by the relationship extraction step.

[0009] Optionally, the tag matrix system includes one or more of the following: subject, mandatory requirement, action, object, time limit and condition.

[0010] Optionally, in the syntactic analysis step, dependency parsing technology is used to perform syntactic analysis on the text after word segmentation processing.

[0011] Optionally, in the entity recognition step, a pre-trained named entity recognition model is used to extract entities.

[0012] A compliance task generation system based on natural language processing according to one aspect of the present invention includes: Pre-built modules for pre-building label matrix systems; Input module, used to input the text to be processed; A text preprocessing module, used for cleaning and segmenting the text to be processed; The syntactic analysis module is used to perform syntactic analysis on the text after word segmentation to identify the grammatical structure of the sentence; An entity recognition module, used for extracting entities based on the grammatical structure of the recognized sentence; A relation extraction module, configured to extract the association relationship between entities based on the grammatical structure of the identified sentence and the extracted entities; and The compliance task generation module is used to generate compliance tasks based on the pre-constructed label matrix system, the entities extracted by the entity recognition module, and the association relationships extracted by the relationship extraction module.

[0013] Optionally, the tag matrix system includes one or more of the following: subject, mandatory requirement, action, object, time limit and condition.

[0014] Optionally, in the syntactic analysis module, dependency parsing technology is used to perform syntactic analysis on the text after word segmentation processing.

[0015] Optionally, in the entity recognition module, a pre-trained named entity recognition model is used to extract entities.

[0016] A computer-readable medium according to one aspect of the present invention stores a computer program thereon, and when the computer program is executed by a processor, the compliance task generation method based on natural language processing is implemented.

[0017] A computer device according to one aspect of the present invention comprises a storage module, a processor, and a computer program stored in the storage module and executable on the processor, wherein the processor implements the compliance task generation method based on natural language processing when executing the computer program.

[0018] A computer program product according to one aspect of the present invention comprises a computer program, which, when executed by a processor, implements the compliance task generation method based on natural language processing. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is a flowchart of a compliance task generation method based on natural language processing according to an embodiment of the present invention.

[0020] Figure 2 It is a structural block diagram of a compliance task generation system based on natural language processing according to an embodiment of the present invention.

[0021] The above and other objects and advantages of the present application will become more fully apparent from the following detailed description in conjunction with the accompanying drawings, in which the same or similar elements are represented by the same reference numerals. DETAILED DESCRIPTION

[0022] The following describes some of the various embodiments of the present invention, which are intended to provide a basic understanding of the present invention, but are not intended to identify the key or decisive elements of the present invention or to limit the scope of protection.

[0023] For the purpose of brevity and illustration, the principles of the present invention are described herein mainly with reference to its exemplary embodiments. However, those skilled in the art will readily recognize that the same principles are equally applicable to all types of natural language processing-based compliance task generation methods and generation systems, and these same principles can be implemented therein, and any such changes do not depart from the true spirit and scope of the present patent application.

[0024] Moreover, in the following description, reference is made to the accompanying drawings, which illustrate specific exemplary embodiments. Electrical, mechanical, logical and structural changes may be made to these embodiments without departing from the spirit and scope of the present invention. In addition, although the features of the present invention are disclosed in conjunction with only one of several implementations / embodiments, this feature may be combined with one or more other features of other implementations / embodiments as may be desirable and / or advantageous for any given or identifiable function. Therefore, the following description should not be regarded as limiting, and the scope of the present invention is defined by the appended claims and their equivalents.

[0025] Terms such as “having” and “including” indicate that in addition to the units (modules) and steps directly and explicitly stated in the specification and claims, the technical solution of the present invention does not exclude the situation where it has other units (modules) and steps that are not directly or explicitly stated.

[0026] Figure 1 It is a flowchart of a compliance task generation method based on natural language processing according to an embodiment of the present invention.

[0027] like Figure 1 As shown, a compliance task generation method based on natural language processing according to an embodiment of the present invention includes the following steps: Pre-construction step S1: pre-constructing a label matrix system, for example, constructing a compliance task label matrix system based on expert experience; Input step S2: legal and regulatory documents, specifically including: inputting the legal and regulatory texts of the compliance tasks to be generated; Text preprocessing step S3: extracting clauses from the input legal and regulatory texts, and performing cleaning and word segmentation processing; Syntactic analysis step S4: performing syntactic analysis on the clause using, for example, dependency parsing techniques to identify the grammatical structure of the sentence; Entity recognition step S5: extracting key entities in the clauses, such as organization name, frequency, condition, and task, based on the grammatical structure of the identified sentences; Relationship extraction step S6: Based on the syntactic analysis results, identify and extract the relationship between entities; Compliance task generation step S7: Generate specific compliance tasks based on the pre-built label matrix.

[0028] These steps are described in detail below.

[0029] First, the label matrix system pre-constructed in the pre-construction step S1 is described. In the present invention application, the label matrix system mainly defines the following aspects, for example: (1) Subject Definition: In legal and regulatory terms, a subject refers to an institution or department that needs to perform compliance tasks.

[0030] Description: The subject is usually a specific organization that is required by laws and regulations to perform specific duties or tasks. The identification of the subject can help clarify the attribution of responsibilities. For example, the core organization, operating organization, etc.

[0031] (2) Mandatory requirements Definition: In legal and regulatory terms, mandatory requirements refer to the degree of mandatory nature of specific compliance requirements imposed on an entity.

[0032] Description: Mandatory requirements indicate the level of obligation of the subject when performing compliance tasks. Different mandatory requirements include should, must and encouraged. Should and must indicate mandatory requirements of laws and regulations on the subject, while encourage indicates a recommended or encouraged behavior.

[0033] (3) Action Definition: In legal and regulatory terms, an action refers to a specific compliance task or behavior that an entity needs to perform.

[0034] Description: Actions are usually specific tasks or behaviors specified in laws and regulations. For example, conduct, execute, evaluate, etc. The identification of action labels helps to clarify the specific content of compliance tasks.

[0035] (4) Target Definition: In legal and regulatory terms, an object refers to the specific thing or area that a compliance task targets.

[0036] Description: The object is the specific goal or scope of the compliance task that the subject needs to perform. For example, important information systems, market fluctuations, etc. The identification of the object helps to clarify the scope of application and specific implementation objects of the compliance task.

[0037] (5) Time limit Definition: In legal and regulatory terms, time limits refer to the frequency or periodicity requirements for performing compliance tasks.

[0038] Description: The time limit label indicates the frequency or periodicity of the compliance task. Common time limits include one-time, daily, quarterly, annual, etc. The identification of this label helps ensure that the compliance task is completed within the specified time.

[0039] (6) Conditions Definition: In legal and regulatory terms, a condition is a prerequisite for performing compliance tasks.

[0040] Description: The condition tag describes the specific premise or situation that triggers the execution of the compliance task. For example, "when a large market fluctuation is found", such a condition can help the system automatically trigger the corresponding compliance task under specific circumstances.

[0041] Next, in the input step S2, the original text of the compliance task to be generated (such as the legal and regulatory text) is input. As an example, a text file related to the compliance task can be selected from the legal and regulatory database, and the selected text file can be electronically input into the system to ensure the integrity and accuracy of the text. The format of the input text can also be further checked to ensure that it meets the requirements of subsequent processing.

[0042] Then, in the text preprocessing step S3, the input legal and regulatory text is cleaned and segmented to provide clean text data for subsequent steps.

[0043] As an example, the following process may be included: Text cleaning: remove irrelevant characters, formatting marks and redundant information from the text, and retain the core content of the clauses; Word segmentation: Use word segmentation algorithms to break text into separate vocabulary units for subsequent processing; and Part-of-speech tagging (optional): tag each lexical unit with its part of speech, such as noun, verb, adjective, etc., to provide a basis for syntactic analysis.

[0044] Next, in the lexical analysis step S4, the dependency parsing technology is used to perform syntactic analysis on the text after word segmentation to identify the grammatical structure of the sentence. Dependency parsing is an important task in natural language processing, which aims to analyze the dependency relationship between words in a sentence, that is, the syntactic relationship between words. Dependency parsing is based on dependency grammar. By analyzing the dependency relationship between words in a sentence, a dependency tree is constructed to represent these relationships. In the dependency tree, each word is a node, and the dependency relationship between words is a tree edge. The identification of dependency relationships is the core task of dependency parsing, which involves identifying the syntactic relationship between words, such as subject-predicate relationship, verb-object relationship, etc.

[0045] In the lexical analysis step S4, as an example, a dependency parsing model is used to perform syntactic analysis, identify the grammatical structure of the sentence, and annotate grammatical components such as the subject, predicate, and object relationships in the sentence, providing a basis for subsequent entity recognition steps and relationship extraction steps, that is, being able to accurately locate the information required for each label, such as subject, action, and object.

[0046] As an example, using dependency parsing technology to perform syntactic analysis on the segmented text to identify the grammatical structure of the sentence may include the following process: Dependency analysis: construct a dependency tree to represent the dependency relationship between words in a sentence; Dependency tagging: attaching a label to each dependency, describing the specific syntactic relationship between the dependent word and the core word; and Dependency tree representation: The dependency relationship of a sentence is represented in the form of a dependency tree, where each word is a node and the dependency relationship is the edge between the nodes.

[0047] Among them, the types of dependencies commonly include the following types: Subject-predicate relationship: indicates the dependence of the subject on the predicate.

[0048] Verb-object relationship: indicates the dependency of the subject on the object.

[0049] Attributive-predicate relationship: It indicates the dependency of a noun on the adjective or noun that modifies it.

[0050] Adverbial structure: It indicates the dependency of a verb on the adverb or prepositional phrase that modifies it.

[0051] Parallel relationship: Indicates the parallel relationship between two words, such as parallel subjects or parallel objects.

[0052] The following technical effects can be achieved by using dependency parsing for syntactic analysis: Improve the accuracy of syntactic analysis: By accurately identifying the dependency relationship between words, dependency parsing can more accurately analyze the grammatical structure of sentences; Enhanced semantic understanding: Dependency parsing not only focuses on the grammatical relationship between words, but also represents these relationships in the form of dependency trees, which helps to deeply understand the semantic information of sentences; Support the implementation of natural language processing tasks: Dependency parsing is a basic task in natural language processing, and its output provides important syntactic information support for other tasks such as information extraction, sentiment analysis, and machine translation. Next, in the entity recognition step S5, key entities in the clauses are identified and extracted, such as organization name, frequency, condition, task, etc.

[0053] As an example, the entity recognition step S5 may include the following process: Entity type definition: Based on the requirements of compliance tasks, define the entity types that need to be identified, such as institution name, time, conditions, tasks, etc.; Entity Recognition Algorithms: Use entity recognition algorithms to scan text and identify entities that meet the definition; and Entity information extraction: Extract relevant information of the identified entities, such as entity name, type, location, etc.

[0054] Among them, as a preferred method of entity recognition algorithm, a named entity recognition (NER) model is used in the present invention application to realize the recognition of entity information. The named entity recognition model belongs to the sequence annotation model in natural language processing (NLP). In the NER task, the goal of the model is to annotate each word or character in the text to determine whether they belong to a specific named entity category, such as names of people, places, names of institutions, time, date, etc. As an example, a pre-trained named entity recognition model is used in the present invention, and fine-tuned on a legal and regulatory data set to identify and extract key entities (such as institutions, frequencies, conditions, and tasks) in the clauses. Through the grammatical structure provided by the syntactic analysis step, the named entity recognition model can be used to more accurately extract key entities.

[0055] As entity recognition algorithms, the following other algorithms can also be listed: (1) Regular expressions: Identify entities through predefined pattern matching rules. For example, regular expressions can be used to match specific organization name formats or date formats.

[0056] (2) Hidden Markov Model (HMM): A generative model that describes the relationship between an observation sequence and a hidden state. In entity recognition, HMM can be used to model the relationship between entities and infer hidden states (such as entity categories) from observation sequences (such as words in text).

[0057] (3) Maximum Entropy Model (MEM): A probabilistic model based on entropy maximization, used to describe the relationship between observation sequences and hidden variables. In entity recognition, MEM can be used to model the relationship between entity instances and entity categories.

[0058] (4) Conditional Random Field (CRF): A discriminative model that takes into account the input state feature function and the label transfer feature function. It can utilize rich internal and contextual feature information in the labeling process. CRF is currently the mainstream model in named entity recognition (NER).

[0059] (5) Support Vector Machine (SVM): A classification algorithm based on statistical learning theory that separates entities of different categories by finding an optimal hyperplane.

[0060] (6) Convolutional Neural Network (CNN): Mainly used in image processing and natural language processing. In entity recognition, CNN can be used to extract the features of entities in text.

[0061] (7) Recurrent Neural Network (RNN): A recursive neural network that can process sequence data. In entity recognition, RNN can be used to model the relationship between entities. LSTM (Long Short-Term Memory) and GRU (Gated Recurrent Unit) are variants of RNN that can solve the problem of long-term dependencies.

[0062] (8) Attention mechanism: Entity recognition is performed by studying the connection between words through the attention model. In the deep learning model, an attention mechanism can be added to enhance the effect of entity recognition.

[0063] (9) Pre-trained language models: such as BERT and GPT, which learn rich language knowledge and contextual information through pre-training on large-scale corpora. These models can be adapted to entity recognition tasks through fine-tuning.

[0064] Then, in the relationship extraction step S6, based on the syntactic analysis results, the relationship between entities is identified and extracted, that is, the association relationship between entities is obtained, such as "subject-execution task", "condition-trigger task", etc.

[0065] Compliance task generation step S7: Generate specific compliance tasks based on the pre-constructed label matrix and the identified entities and the associations between entities.

[0066] In order to better understand the above steps S1 to S7, a hypothetical "data protection regulation" can be used as an example of a legal document to illustrate each step: Example: Legal and regulatory documents (data protection regulations) “All companies processing personal data must conduct a data security audit annually and report any data breach to the data protection authority within 24 hours of discovery. In addition, companies should ensure that appropriate technical and organizational measures are in place to protect personal data, which should include at least data encryption and access controls.” In the pre-construction step S1, based on the experts’ understanding of data protection regulations, a label matrix system is constructed, including labels such as subject (such as “company”), mandatory requirements (such as “must”), actions (such as “conduct data security audit”, “report”), objects (such as “personal data”, “data protection agency”), time limits (such as “every year”, “within 24 hours”), and conditions (such as “when data leakage is discovered”).

[0067] In the input step S2, the text content of the above data protection regulations is input as the original text of the compliance task to be generated.

[0068] In the text preprocessing step S3, the entered regulatory text is cleaned to remove irrelevant characters and formats, and then word segmentation is performed to split the text into independent vocabulary units, such as "all", "processing", "personal data", "company", etc.

[0069] In the syntactic analysis step S4, the text after word segmentation is analyzed using dependency parsing technology to identify the grammatical structure of the sentence, such as the subject-predicate-object relationship, the attributive-modification relationship, etc. For example, it is identified that "company" is the subject, "must" is the predicate, and "conduct data security audit" is the object.

[0070] In the entity identification step S5, based on the results of the syntactic analysis, key entities in the clause are identified and extracted. In this example, the entities may include "company" (subject), "personal data" (object), "data security audit" (action object), "data protection agency" (report object), "annually" (time limit), "within 24 hours" (time limit), and "upon discovery of data leakage" (condition).

[0071] In the relationship extraction step S6, based on the results of syntactic analysis and entity recognition, the relationship between entities is identified and extracted. For example, the subject-object relationship between "company" and "conducting data security audit" and the relationship between "company" and "data protection agency" in the reporting action are identified.

[0072] In the compliance task generation step S7, a specific compliance task is generated based on the pre-built label matrix, the extracted entities and the associations between them, for example: Task 1: All companies that process personal data must conduct a data security audit annually.

[0073] Task 2: Companies that process personal data must report a data breach to the data protection authority within 24 hours of becoming aware of it.

[0074] Task 3: Companies that process personal data should ensure that appropriate technical and organizational measures are in place to protect personal data, which should include at least data encryption and access controls.

[0075] As described above, according to the compliance task generation method based on natural language processing of the present invention, by defining a label matrix, it is possible to more accurately parse and understand the legal and regulatory clauses and generate standardized compliance tasks. Moreover, the language characteristics of the legal and regulatory clauses are optimized in particular to improve the accuracy and efficiency of parsing. Furthermore, by using advanced dependency parsing and NER technology, accurate parsing of complex legal clauses is ensured, and the relationship between entities is automatically identified and extracted based on dependency relationships to generate key information required for compliance tasks. In this way, integrated task generation can be achieved, that is, the clause parsing and compliance task generation functions can be integrated to achieve end-to-end automated compliance management.

[0076] Therefore, according to the compliance task generation method based on natural language processing of the present invention, the efficiency of compliance task generation can be improved, and the legal and regulatory provisions can be automatically parsed and compliance tasks can be generated, reducing the time and cost of manual parsing. Moreover, the risk of missing compliance tasks can be reduced, and legal and regulatory requirements can be accurately identified and implemented, reducing legal risks caused by compliance omissions. Furthermore, regulatory provisions can be updated in a timely manner, and the update of laws and regulations can be quickly adapted to, and new compliance tasks can be generated in a timely manner.

[0077] Figure 2 It is a structural block diagram of a compliance task generation system based on natural language processing according to an embodiment of the present invention.

[0078] like Figure 2 As shown, a compliance task generation system 100 based on natural language processing according to an embodiment of the present invention includes: A pre-construction module 110 is used to pre-construct a label matrix system; An input module 120, used for inputting text to be processed; A text preprocessing module 130, used for cleaning and segmenting the text to be processed; The syntactic analysis module 140 is used to perform syntactic analysis on the text after word segmentation processing to identify the grammatical structure of the sentence; An entity recognition module 150, configured to extract entities based on the grammatical structure of the identified sentence; A relation extraction module 160, configured to extract association relations between entities based on the grammatical structure of the identified sentence and the extracted entities; and The compliance task generation module 170 is used to generate compliance tasks according to the pre-constructed label matrix system, the entities extracted by the entity recognition module, and the association relationships extracted by the relationship extraction module.

[0079] The label matrix system includes one or more of the following: subject, mandatory requirement, action, object, time limit and condition.

[0080] In the syntactic analysis module 140, dependency parsing technology is used to perform syntactic analysis on the text after word segmentation processing.

[0081] In the entity recognition module 150, a pre-trained named entity recognition model is used to extract entities.

[0082] As described above, the compliance task generation system based on natural language processing according to the present invention can achieve the following technical effects: (1) Structured information extraction: By defining clear labels (such as subject, action, object, etc.), the label matrix system can extract key information from legal and regulatory texts in a structured manner. This structured information is easier for computers to process and analyze, providing a solid foundation for subsequent task generation and compliance management; (2) Improved accuracy: The label matrix system ensures the accuracy of information extraction through precise definitions and descriptions. For example, by clarifying that "subject" refers to the institution or department that needs to perform compliance tasks, the responsible entity in the text can be accurately identified to avoid confusion or omissions. (3) Enhanced readability: The structured label matrix allows compliance tasks to be presented in a clear and organized manner, which enhances readability. This is crucial for compliance personnel to understand and perform compliance tasks, and can greatly improve work efficiency and accuracy. (4) Flexibility and scalability: The label matrix system is flexible and can be adjusted and expanded according to specific laws, regulations and compliance requirements. For example, if new laws and regulations introduce new compliance requirements or concepts, these changes can be accommodated by adding new labels. (5) Automation and intelligence: The signature matrix system makes it possible to generate automated and intelligent compliance tasks. Through natural language processing technology and machine learning algorithms, it can automatically identify label information in the text and generate specific compliance tasks based on this information. This greatly reduces manual intervention and improves the efficiency and accuracy of compliance management; (6) Supporting compliance analysis: The label matrix system can also be used for compliance analysis. By statistically analyzing the extracted label information, we can understand the distribution, execution frequency, and completion status of compliance tasks, and provide decision support for compliance management.

[0083] The above are only specific implementations of the present application, but the protection scope of the present application is not limited thereto. Those skilled in the art can think of other feasible changes or substitutions based on the technical scope disclosed in the present application, and such changes or substitutions are all included in the protection scope of the present application. In the absence of conflict, the implementation modes of the present application and the features in the implementation modes can also be combined with each other. The protection scope of the present application shall be subject to the description of the claims.

Claims

1. A compliance task generation method based on natural language processing, characterized in that: The following steps are involved: Pre-construction step, pre-constructing the label matrix system; Input step: input the text to be processed; A text preprocessing step, cleaning and segmenting the text to be processed; A syntactic analysis step is to perform syntactic analysis on the text after word segmentation to identify the grammatical structure of the sentence, wherein a dependency parsing model is used to perform syntactic analysis to identify the grammatical structure of the sentence; An entity recognition step, extracting entities based on the grammatical structure of the sentence identified in the syntactic analysis step, wherein the entities are recognized and extracted using a pre-trained named entity recognition model; A relationship extraction step, extracting the association relationship between entities based on the grammatical structure of the sentence identified in the syntactic analysis step and the entities extracted in the entity recognition step; and The compliance task generation step generates a compliance task based on the label matrix system pre-constructed in the pre-construction step, the entities extracted in the entity recognition step, and the association relationships extracted in the relationship extraction step.

2. The method for generating compliance tasks based on natural language processing according to claim 1, characterized in that: In the pre-construction step, a label matrix system is pre-constructed to include one or more of the following: subject, mandatory requirement, action, object, time limit and condition.

3. A compliance task generation system based on natural language processing, characterized in that: include: Pre-built modules for pre-building label matrix systems; Input module, used to input the text to be processed; A text preprocessing module, used for cleaning and segmenting the text to be processed; A syntactic analysis module is used to perform syntactic analysis on the text after word segmentation processing to identify the grammatical structure of the sentence, wherein the syntactic analysis is performed on the text after word segmentation processing using dependency parsing technology; An entity recognition module, used for extracting entities based on the grammatical structure of the sentence recognized by the syntactic analysis module, wherein a pre-trained named entity recognition model is used to extract entities; A relationship extraction module, configured to extract the association relationship between entities based on the grammatical structure of the sentence identified by the syntactic analysis module and the entities extracted by the entity recognition module; and The compliance task generation module is used to generate compliance tasks based on the pre-constructed label matrix system, the entities extracted by the entity recognition module, and the association relationships extracted by the relationship extraction module.

4. The compliance task generation system based on natural language processing according to claim 3, characterized in that: The label matrix system includes one or more of the following: subject, mandatory requirement, action, object, time limit and condition.

5. A computer readable medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the compliance task generation method based on natural language processing described in claim 1 or 2 is implemented.

6. A computer device comprising a storage module, a processor, and a computer program stored in the storage module and executable on the processor, characterized in that: When the processor executes the computer program, the compliance task generation method based on natural language processing described in claim 1 or 2 is implemented.

7. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the compliance task generation method based on natural language processing described in claim 1 or 2 is implemented.

Images