A document specification review method and system

Abstract

The application discloses a document specification review method and system, comprising: obtaining a to-be-reviewed document and preprocessing, calling a pre-constructed professional vocabulary error correction model of a corresponding field to perform error word identification, and obtaining an error word identification result; judging whether the pre-processed to-be-reviewed document contains a reference standard, if the reference standard is not identified, generating a reference standard missing prompt; if the reference standard is identified, extracting information of the reference standard, comparing the identified reference standard information with a pre-constructed standard specification database, and judging whether the reference standard in the to-be-reviewed document is a valid version; if the reference standard in the to-be-reviewed document is a valid version, performing content consistency comparison between the to-be-reviewed document and the pre-constructed standard specification knowledge base, and judging whether there is a description error. Through the method, error words and standard specifications can be identified, and the identification accuracy and efficiency are improved.

Description

Technical Field

[0001] This application relates to the field of document review technology, and in particular to a document standard review method and system. Background Technology

[0002] In scenarios such as digital image review, drug application, and financial risk control, companies need to perform text recognition and compliance checks on massive amounts of scanned documents every day. Although OCR technology can quickly convert images into text, the conversion results often contain similar-looking misspellings, miswritten technical terms, and expressions inconsistent with the latest standards. Once these hidden textual errors enter the subsequent business process, they will directly lead to rework of the review, regulatory fines, and even security risks.

[0003] In existing technologies, manual verification of typos is required, which is not only inefficient but also unable to accurately determine whether the text is correct or conforms to standards. Therefore, there is an urgent need for an integrated automatic verification method that immediately performs "text correctness" and "standard compliance" after text recognition. Summary of the Invention

[0004] Based on this, and to address the aforementioned technical issues, a document specification review method and system are provided, which can improve the accuracy and efficiency of identifying typos and standard specification errors in documents.

[0005] Firstly, a document specification review method, the method comprising:

[0006] Obtain the document to be reviewed; preprocess the document to be reviewed;

[0007] The preprocessed documents to be reviewed are classified by domain, and a pre-built professional vocabulary error correction model for the corresponding domain is called according to the classification structure to identify misspelled words and obtain the misspelled word identification results.

[0008] Based on predefined reference standard feature rules, it is determined whether the preprocessed document to be reviewed contains reference standards. If no reference standards are identified, a reference standard missing prompt is generated.

[0009] If a referenced standard is identified, its information is extracted, and the identified referenced standard information is compared with a pre-built standard and specification database to determine whether the referenced standard in the document to be reviewed is a valid version. If the referenced standard is not a valid version, a referenced standard adjustment prompt is generated. The standard and specification database stores the name, standard number, promulgation time, implementation time, and status information of the standard and specification.

[0010] If the referenced standard in the document to be reviewed is a valid version, then the document to be reviewed is compared with the pre-built standard specification knowledge base for content consistency to determine whether there is a description error and generate a standard specification description error result. The standard specification knowledge base stores a vectorized representation of the full text of the standard specification.

[0011] The results of word error identification, missing standard citation prompts, standard adjustment prompts, and standard specification description errors are compiled and the review comments are output.

[0012] Optionally, in the above scheme, preprocessing the document data to be reviewed includes:

[0013] If the document data to be reviewed is non-editable text data, the document data to be reviewed is converted into editable text data using OCR technology to obtain pre-processed document data to be reviewed.

[0014] If the document data to be reviewed contains editable text data and non-editable text data, the non-editable text data in the document data to be reviewed is converted into editable text data using OCR technology, and then merged with the original editable text data to obtain the preprocessed document data to be reviewed.

[0015] A standards and specifications knowledge base is constructed based on the specific content of the standards and specifications regulations in the aforementioned standards and specifications database.

[0016] Optionally, in the above scheme, the professional terminology error correction model performs error recognition by including:

[0017] Obtain a terminology set for the target field;

[0018] The open-source large language model is trained using a set of specialized terms from the target domain to generate a basic model with domain knowledge.

[0019] Construct error correction training samples containing common misspellings and easily confused terms in the field, and train the error correction task based on the base model to obtain the professional terminology error correction model.

[0020] In the above scheme, optionally, determining whether the preprocessed document to be reviewed contains standard references based on predefined standard reference feature rules includes:

[0021] Text analysis is performed on the document to be reviewed, and based on predefined standard citation feature rules, it is detected whether predefined standard citation key features appear in it; the standard citation feature rules include at least one of standard book title format characters, standard number format characters, and standard citation chapter title keywords;

[0022] If no of the aforementioned key features are detected, the document to be reviewed is determined to not contain standard references; if at least one of the aforementioned key features is detected, the document to be reviewed is determined to contain standard references.

[0023] Optionally, in the above scheme, comparing the identified referenced standard information with a pre-built standard and specification database to determine whether the referenced standard in the document to be reviewed is a valid version specifically includes:

[0024] From the identified referenced standard information, the referenced standard name and version number are parsed out;

[0025] The parsed standard name or version number is used as the query condition to match the standard entries in the standard specification database to verify whether the referenced standard information is correct.

[0026] Based on the premise that the citation is authentic, the document time attribute of the document to be reviewed is obtained, and it is logically compared with the promulgation time attribute of the standard entry in the matching standard specification database to verify whether the cited standard is in a valid state at the time indicated by the time attribute of the document to be reviewed.

[0027] Optionally, in the above scheme, the document to be reviewed is compared with a pre-built standard specification knowledge base to determine whether there are any descriptive errors. This specifically includes:

[0028] In the document to be reviewed, technical description statements that have a semantic connection with the valid reference standard are identified as target judgment statements;

[0029] Based on the valid reference standards associated with the target judgment sentence, the corresponding full text of the standard specification is obtained from the standard specification knowledge base;

[0030] Based on the valid reference standards associated with the target judgment sentence, the corresponding full text of the standard specification is obtained from the standard specification knowledge base;

[0031] Based on the results of the semantic analysis, it is determined whether the target judgment sentence contains a description error that contradicts the standard specification clause; if so, a description error result is generated.

[0032] In the above scheme, optionally, the technical description statements in the document to be reviewed that have a semantic relationship with the valid reference standard are identified as target determination statements, specifically including:

[0033] Identify the standard association identifiers used to associate technical descriptions and referenced standards in the document to be reviewed; wherein, the standard association identifiers include text units or reference symbols that directly contain standard information;

[0034] Extract the technical description text that is within a preset association range of the standard association identifier, and use it as the target judgment sentence; and determine the reference standard pointed to by the standard association identifier.

[0035] Secondly, a document specification review system, the system comprising:

[0036] Preprocessing module: used to acquire the document to be reviewed; and to preprocess the document to be reviewed;

[0037] The misspelling detection module is used to classify the preprocessed documents to be reviewed into different domains, and to call a pre-built professional vocabulary error correction model for the corresponding domain to identify misspellings based on the classification structure, so as to obtain the misspelling detection results.

[0038] The module for determining the existence of reference standards is used to determine whether the preprocessed document to be reviewed contains reference standards based on predefined reference standard feature rules. If no reference standards are identified, a reference standard missing prompt is generated.

[0039] The Validity Judgment Module for Referenced Standards: If a referenced standard is identified, the module extracts the information of the referenced standard, compares the identified referenced standard information with a pre-built standard and specification database, and determines whether the referenced standard in the document to be reviewed is a valid version. If the referenced standard is not a valid version, the module generates a referenced standard adjustment prompt. The standard and specification database stores the name, standard number, promulgation time, implementation time, and status information of the standard and specification.

[0040] Standard specification description correctness judgment module: If the referenced standard in the document to be reviewed is a valid version, the module compares the content of the document to be reviewed with the pre-built standard specification knowledge base to determine whether there is a description error and generates a standard specification description error result. The standard specification knowledge base stores a vectorized representation of the full text of the standard specification.

[0041] The review opinion output module is used to organize the results of word error identification, missing standard citation prompts, standard adjustment prompts, and standard specification description errors, and output the review opinion results.

[0042] This application has at least the following beneficial effects:

[0043] This application categorizes the documents under review into different domains, then uses a domain-specific vocabulary correction model to identify errors. Because the model incorporates specialized vocabulary during training, it can check both common and specialized terms, automatically and quickly identifying typos and errors in technical terms. Next, based on predefined citation standard feature rules, it determines whether the document contains citation standards and provides a missing citation standard warning. It also compares the identified citation standard information with a pre-built standard and specification database to determine if the citation standards in the document are valid and provides a non-valid version warning. Finally, it compares the content consistency of the document under review with a pre-built standard and specification knowledge base to determine if there are any descriptive errors and generates a standard and specification description error result. Therefore, this application can automatically identify typos, technical terminology errors, the existence and validity of citation standards, and standard and specification errors, greatly improving accuracy and efficiency. Attached Figure Description

[0044] Figure 1 A flowchart of a document specification review system provided in one embodiment of this application. Detailed Implementation

[0045] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0046] In one embodiment, such as Figure 1 As shown, a document specification review method is provided, the method comprising:

[0047] Step S1: Obtain the document to be reviewed; preprocess the document to be reviewed.

[0048] In step S1, if the document data to be reviewed is non-editable text data, the document data to be reviewed is converted into editable text data using OCR technology to obtain preprocessed document data to be reviewed;

[0049] If the document data to be reviewed contains editable text data and non-editable text data, the non-editable text data in the document data to be reviewed is converted into editable text data using OCR technology, and then merged with the original editable text data to obtain the preprocessed document data to be reviewed.

[0050] Step S2: Classify the preprocessed document to be reviewed into different domains, and use the pre-built professional vocabulary error correction model for the corresponding domain to identify misspelled words based on the classification structure, so as to obtain the misspelled word identification result.

[0051] In step S2, keywords are extracted from the preprocessed document data to be reviewed, and domain classification is performed based on the keywords using a domain classification model. Methods for constructing domain classification models for text domain analysis have been described in existing technologies and will not be repeated here.

[0052] Meanwhile, the construction of the professional terminology error correction model includes: obtaining a set of professional terms in the target domain; using the set of professional terms in the target domain to train an open-source large language model to generate a basic model with domain knowledge; constructing error correction training samples containing common misspellings and easily confused terms in the domain; and training the error correction task based on the basic model to obtain the professional terminology error correction model.

[0053] In other words, by using an open-source large language model and writing a Prompt, we can identify common typos. By building a collection of professional terms and using the open-source large language model as a base, we can fine-tune it to form a small model for professional terminology recognition. By applying the model and writing a Prompt, we can achieve professional terminology recognition.

[0054] Step S3: Based on predefined reference standard feature rules, determine whether the preprocessed document to be reviewed contains reference standards. If no reference standards are identified, generate a reference standard missing prompt.

[0055] In step S3, text analysis is performed on the document to be reviewed. Based on predefined standard citation feature rules, it is detected whether predefined standard citation key features appear. These standard citation feature rules include at least one of the following: standard book title format characters, standard number format characters, and standard citation chapter title keywords. If the document contains standard citation chapters, the existence of standard citation feature rules is determined by identifying the specific content under "standard citation chapters," or by describing phrases such as "《Safety Technical Specification for High-Altitude Operations in Building Construction》(JGJ 80-2016)" or "must comply with the requirements of GB 50016-2014." Specifically, this includes the following:

[0056] (1) A text pattern containing Chinese book titles with standard numbers in parentheses;

[0057] (2) The standard numbering text format conforms to the national and industry standard numbering and naming specifications;

[0058] (3) The text content located under the headings of the sections “Referenced Standards”, “Basis for Compilation”, and “Normative References”;

[0059] If no of the aforementioned key features are detected, the document to be reviewed is determined to not contain standard references; if at least one of the aforementioned key features is detected, the document to be reviewed is determined to contain standard references.

[0060] Step S4: If a referenced standard is identified, extract the information of the referenced standard, compare the identified referenced standard information with the pre-built standard specification database, and determine whether the referenced standard in the document to be reviewed is a valid version. If the referenced standard is not a valid version, generate a referenced standard adjustment prompt. The standard specification database stores the name, standard number, promulgation time, implementation time and status information of the standard specification.

[0061] In step S4, the referenced standard name and version number are parsed from the identified referenced standard information;

[0062] The parsed standard name or version number is used as the query condition to match the standard entries in the standard specification database to verify whether the referenced standard information is correct.

[0063] Based on the premise that the citation is authentic, the document time attribute of the document to be reviewed is obtained, and it is logically compared with the promulgation time attribute of the standard entry in the matching standard specification database to verify whether the cited standard is in a valid state at the time indicated by the time attribute of the document to be reviewed.

[0064] Specifically, the standards and specifications database contains information on the names, standard numbers, promulgation dates, implementation dates, and status of many standards and specifications. Users can search by name or number to determine whether a standard is valid.

[0065] Step S5: If the referenced standard in the document to be reviewed is a valid version, then the document to be reviewed is compared with the pre-built standard specification knowledge base for content consistency to determine whether there is a description error and generate a standard specification description error result. The standard specification knowledge base stores the vectorized representation of the full text of the standard specification.

[0066] In step S5, in the document to be reviewed, technical description statements that have a semantic relationship with the valid reference standard are identified as target judgment statements;

[0067] Based on the valid reference standards associated with the target judgment sentence, the corresponding full text of the standard specification is obtained from the standard specification knowledge base;

[0068] Based on the valid reference standards associated with the target judgment sentence, the corresponding full text of the standard specification is obtained from the standard specification knowledge base;

[0069] Based on the results of the semantic analysis, it is determined whether the target judgment sentence contains a description error that contradicts the standard specification clause; if so, a description error result is generated.

[0070] Specifically, the technical description statements that are determined to have a semantic association with the valid reference standard, as target determination statements, include:

[0071] Identify the standard association identifiers used to associate technical descriptions and referenced standards in the document to be reviewed; wherein, the standard association identifiers include text units or reference symbols that directly contain standard information;

[0072] Extract the technical description text that is within a preset association range of the standard association identifier, and use it as the target judgment sentence; and determine the reference standard pointed to by the standard association identifier.

[0073] Step S6: Organize the results of incorrect word identification, missing standard citation prompts, standard citation adjustment prompts, and standard specification description errors, and output the review comments.

[0074] In this embodiment, the present application specifically includes:

[0075] I. Identification of misspelled words and incorrect terminology:

[0076] 1. Use an open-source large language model for training to achieve the recognition of common typos.

[0077] 2. By constructing a collection of professional terms, and using an open-source large language model as a foundation, a small model for professional term recognition is formed through fine-tuning. The application model is then used to realize the recognition of professional terms.

[0078] II. Standard and Specification Error Identification:

[0079] 1. Construct a standards and specifications database to maintain specific information such as the name, version, and promulgation date of standards and specifications;

[0080] 2. Apply RAG technology to build a standard specification knowledge base, and vectorize the standard specifications for comparison of document content standard compliance.

[0081] 3. Using full-text search technology, locate the names and versions of the standards cited in the document under review. If the document does not cite any standards, it is directly deemed non-compliant, and feedback is provided that the cited standards need to be added. If the document specifies the cited standards, it needs to be compared with the standard and specification database, including whether the standard name, version, promulgation date, and document writing date are consistent. If they are not consistent, feedback is provided that the cited standards need to be adjusted.

[0082] 4. Based on the previous step's successful review, the document content is segmented and vectorized. Using the established standard and specification knowledge base, the document's content is reviewed for compliance with standards and specifications. If any discrepancies are found between the descriptions in the document and the referenced standards and specifications, feedback is provided on which specific standard and which provision of the document violates the specified content.

[0083] III. Review Report Generation:

[0084] The feedback information generated in the previous two steps is used as input and organized to finally form a standard format review opinion document.

[0085] In the aforementioned document specification review method, the document data to be reviewed is classified by domain. Then, a domain-specific terminology correction model is used to identify errors. Because the professional terminology correction model incorporates professional vocabulary during training, it can check both common and professional terms, automatically and quickly identifying typos and errors in professional terminology. Next, based on predefined citation standard feature rules, it determines whether the document to be reviewed contains citation standards and provides a missing citation standard warning. Furthermore, it compares the identified citation standard information with a pre-built standard and specification database to determine whether the citation standards in the document are valid versions and provides a non-valid version warning. Finally, it compares the content consistency of the document to be reviewed with a pre-built standard and specification knowledge base to determine if there are any descriptive errors and generates a standard and specification description error result. Therefore, this application can automatically identify typos, errors in professional terminology, the existence and validity of citation standards, and errors in standard and specification, greatly improving accuracy and efficiency.

[0086] In one embodiment, a document specification review system is provided, the system comprising:

[0087] Preprocessing module: used to acquire the document to be reviewed; and to preprocess the document to be reviewed;

[0088] The misspelling detection module is used to classify the preprocessed documents to be reviewed into different domains, and to call a pre-built professional vocabulary error correction model for the corresponding domain to identify misspellings based on the classification structure, so as to obtain the misspelling detection results.

[0089] The module for determining the existence of reference standards is used to determine whether the preprocessed document to be reviewed contains reference standards based on predefined reference standard feature rules. If no reference standards are identified, a reference standard missing prompt is generated.

[0090] The Validity Judgment Module for Referenced Standards: If a referenced standard is identified, the module extracts the information of the referenced standard, compares the identified referenced standard information with a pre-built standard and specification database, and determines whether the referenced standard in the document to be reviewed is a valid version. If the referenced standard is not a valid version, the module generates a referenced standard adjustment prompt. The standard and specification database stores the name, standard number, promulgation time, implementation time, and status information of the standard and specification.

[0091] Standard specification description correctness judgment module: If the referenced standard in the document to be reviewed is a valid version, the module compares the content of the document to be reviewed with the pre-built standard specification knowledge base to determine whether there is a description error and generates a standard specification description error result. The standard specification knowledge base stores a vectorized representation of the full text of the standard specification.

[0092] The review opinion output module is used to organize the results of word error identification, missing standard citation prompts, standard adjustment prompts, and standard specification description errors, and output the review opinion results.

[0093] For specific limitations regarding a document standard review system, please refer to the limitations of a document standard review method described above, which will not be repeated here. Each module in the aforementioned document standard review system can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0094] In one embodiment, a computer device is provided, which may be a server. The computer device includes a processor, memory, and a network interface connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage medium. The network interface is used to communicate with external terminals via a network connection. When the computer program is executed by the processor, it implements the aforementioned document specification review method.

[0095] In one embodiment, a computer-readable storage medium is also provided, on which a computer program is stored relating to all or part of the processes in the methods of the above embodiments.

[0096] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the methods described above. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, or optical storage, etc. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc.

[0097] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0098] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A document specification review method, characterized in that, The method includes: Obtain the document to be reviewed; preprocess the document to be reviewed; The preprocessed documents to be reviewed are classified by domain, and a pre-built professional vocabulary error correction model for the corresponding domain is called according to the classification structure to identify misspelled words and obtain the misspelled word identification results. Based on predefined reference standard feature rules, it is determined whether the preprocessed document to be reviewed contains reference standards. If no reference standards are identified, a reference standard missing prompt is generated. If a referenced standard is identified, its information is extracted, and the identified referenced standard information is compared with a pre-built standard and specification database to determine whether the referenced standard in the document to be reviewed is a valid version. If the referenced standard is not a valid version, a referenced standard adjustment prompt is generated. The standard and specification database stores the name, standard number, promulgation time, implementation time, and status information of the standard and specification. If the referenced standard in the document to be reviewed is a valid version, then the document to be reviewed is compared with the pre-built standard specification knowledge base for content consistency to determine whether there is a description error and generate a standard specification description error result. The standard specification knowledge base stores a vectorized representation of the full text of the standard specification. The results of word error identification, missing standard citation prompts, standard adjustment prompts, and standard specification description errors are compiled and the review comments are output.

2. The document standardization review method according to claim 1, characterized in that, Preprocessing the document data to be reviewed includes: If the document data to be reviewed is non-editable text data, the document data to be reviewed is converted into editable text data using OCR technology to obtain pre-processed document data to be reviewed. If the document data to be reviewed contains editable text data and non-editable text data, the non-editable text data in the document data to be reviewed is converted into editable text data using OCR technology, and then merged with the original editable text data to obtain the preprocessed document data to be reviewed.

3. The document standardization review method according to claim 1, characterized in that, The construction of the specialized terminology error correction model includes: Obtain a terminology set for the target field; The open-source large language model is trained using a set of specialized terms from the target domain to generate a basic model with domain knowledge. Construct error correction training samples containing common misspellings and easily confused terms in the field, and train the error correction task based on the base model to obtain the professional terminology error correction model.

4. The document standardization review method according to claim 1, characterized in that, Determining whether the preprocessed document to be reviewed contains standard references based on predefined standard reference feature rules includes: Text analysis is performed on the document to be reviewed, and based on predefined standard citation feature rules, it is detected whether predefined standard citation key features appear in it; the standard citation feature rules include at least one of standard book title format characters, standard number format characters, and standard citation chapter title keywords; If no of the aforementioned key features are detected, the document to be reviewed is determined to not contain standard references; if at least one of the aforementioned key features is detected, the document to be reviewed is determined to contain standard references.

5. The document standardization review method according to claim 1, characterized in that, The process of comparing the identified referenced standards information with a pre-built database of standards and specifications to determine whether the referenced standards in the document to be reviewed are valid versions specifically includes: From the identified referenced standard information, the referenced standard name and version number are parsed out; The parsed standard name or version number is used as the query condition to match the standard entries in the standard specification database to verify whether the referenced standard information is correct. Based on the premise that the citation is authentic, the document time attribute of the document to be reviewed is obtained, and it is logically compared with the promulgation time attribute of the standard entry in the matching standard specification database to verify whether the cited standard is in a valid state at the time indicated by the time attribute of the document to be reviewed.

6. The document standardization review method according to claim 1, characterized in that, The document to be reviewed is compared with a pre-built standard and specification knowledge base to determine whether there are any descriptive errors. This includes: In the document to be reviewed, technical description statements that have a semantic connection with the valid reference standard are identified as target judgment statements; Based on the valid reference standards associated with the target judgment sentence, the corresponding full text of the standard specification is obtained from the standard specification knowledge base; Based on the valid reference standards associated with the target judgment sentence, the corresponding full text of the standard specification is obtained from the standard specification knowledge base; Based on the results of the semantic analysis, it is determined whether the target judgment sentence contains a description error that contradicts the standard specification clause; if so, a description error result is generated.

7. The document standardization review method according to claim 6, characterized in that, In the document to be reviewed, technical description statements that are semantically related to the valid reference standard are identified as target judgment statements, specifically including: Identify the standard association identifiers used to associate technical descriptions and referenced standards in the document to be reviewed; wherein, the standard association identifiers include text units or reference symbols that directly contain standard information; Extract the technical description text that is within a preset association range of the standard association identifier, and use it as the target judgment sentence; and determine the reference standard pointed to by the standard association identifier.

8. A document specification review system, characterized in that, The system includes: Preprocessing module: used to acquire the document to be reviewed; and to preprocess the document to be reviewed; The misspelling detection module is used to classify the preprocessed documents to be reviewed into different domains, and to call a pre-built professional vocabulary error correction model for the corresponding domain to identify misspellings based on the classification structure, so as to obtain the misspelling detection results. The module for determining the existence of reference standards is used to determine whether the preprocessed document to be reviewed contains reference standards based on predefined reference standard feature rules. If no reference standards are identified, a reference standard missing prompt is generated. The Validity Judgment Module for Referenced Standards: If a referenced standard is identified, the module extracts the information of the referenced standard, compares the identified referenced standard information with a pre-built standard and specification database, and determines whether the referenced standard in the document to be reviewed is a valid version. If the referenced standard is not a valid version, the module generates a referenced standard adjustment prompt. The standard and specification database stores the name, standard number, promulgation time, implementation time, and status information of the standard and specification. Standard specification description correctness judgment module: If the referenced standard in the document to be reviewed is a valid version, the module compares the content of the document to be reviewed with the pre-built standard specification knowledge base to determine whether there is a description error and generates a standard specification description error result. The standard specification knowledge base stores a vectorized representation of the full text of the standard specification. The review opinion output module is used to organize the results of word error identification, missing standard citation prompts, standard adjustment prompts, and standard specification description errors, and output the review opinion results.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

Images