Multi-dimensional attribute fusion contract number recommendation method, device, equipment and medium

Abstract

The present application relates to the technical field of contract management, and specifically to a method, apparatus, device, and medium for recommending contract numbers based on multi-dimensional attribute fusion. The method comprises the following steps: S1. Obtaining unstructured information of a contract, extracting and standardizing it into multi-dimensional contract attribute information; S2. Based on the multi-dimensional contract attribute information, generating contract numbering rules by weight allocation and interactive component calculation; S3. Based on the multi-dimensional contract attribute information, determining the generation node of the contract number by S-type decision function calculation, including generation at creation, generation after approval, or generation upon completion of signing; S4. Manually reviewing the generated contract numbering rules and generation nodes, and after the review is passed, outputting the contract number according to the contract numbering rules at the determined generation node. Generating numbers at relatively safe nodes such as archiving reduces the risk of confidentiality leaks caused by predictable numbering patterns, and enhances the protection of highly confidential contracts.

Description

Technical Field

[0001] The present application relates to the technical field of contract management, and in particular to a method, apparatus, device and medium for recommending contract numbers based on multi-dimensional attribute fusion. Background Art

[0002] Current contract management systems mostly use semi-automated coding rules, generating contract numbers based on preset templates (such as type + year + serial number). While some companies have introduced basic classification fields (such as department and region) to improve management efficiency, in terms of security, most systems rely on permission control and simple encryption. Numbering rules for sensitive contracts still use reversible logic (such as SECRET-001), which poses the risk of information leakage. Furthermore, while mainstream technologies enable basic data interoperability with external systems (such as ERP and electronic signature platforms) through APIs, complex business scenarios (such as multi-national, multi-version contracts) still require manual intervention, resulting in limited intelligence. The industry as a whole is in the early stages of transitioning from static to dynamic rules.

[0003] The fixed coding templates of traditional systems cannot adapt to dynamic contract attributes (such as amount, urgency, and multilingualism), resulting in the mixing of high-value contracts with ordinary contracts and low retrieval efficiency. This is due to the lack of algorithmic support for multi-attribute joint decision-making, which makes it impossible to flexibly generate numbering rules based on the multi-dimensional attributes of the contract. Confidential contracts are identified only by prefixes, making the numbering rules easy to reverse-parse; at the same time, premature number generation (such as at submission) leads to the exposure of sensitive information in the process. This is due to the lack of in-depth coding rule design and the lack of a dynamic generation node decision-making mechanism, making it impossible to reasonably determine the timing of number generation while ensuring security. Version management relies on manual labeling of attachments and supplementary agreements, which is prone to errors; multi-language and multi-regional rules are difficult to unify in cross-border scenarios. This is because the technical solution does not introduce intelligent engines such as weighting functions and automated version associations, and is overly dependent on manual experience, making it difficult to adapt to business expansion needs.

[0004] Traditional systems are focused on functional implementation and lack systematic modeling of the complex attributes of the entire contract lifecycle. Furthermore, they have long neglected the deep integration of dynamic rule engines and security algorithms, resulting in technological iteration lagging behind the globalization and high-compliance needs of enterprises. Furthermore, most solutions compromise between security and efficiency (e.g., reducing numbering complexity to increase generation speed), failing to achieve synergy between the two, hindering the industry's progress towards intelligent and secure development. Summary of the Invention

[0005] The present invention provides a contract number recommendation method, device, equipment and medium with multi-dimensional attribute fusion. The method solves the problems of traditional contract management systems such as rigid number generation rules, easy leakage of sensitive information and inefficient multi-version management through multi-dimensional data collection, multi-dimensional attribute fusion calculation and manual final confirmation.

[0006] In a first aspect, the technical solution of the present invention provides a method for recommending contract numbers based on multi-dimensional attribute fusion, comprising the following steps:

[0007] S1. Obtain unstructured contract information, extract and standardize it into multi-dimensional contract attribute information, including contract type, confidentiality level, contract amount, importance, approval method, attachment tag, contract version, language version, and contract source;

[0008] S2. Based on the multi-dimensional contract attribute information, generate contract numbering rules by weight allocation and interactive component calculation; including simple code rules, standard rules and enhanced rules;

[0009] S3. Based on the multi-dimensional contract attribute information, determine the generation node of the contract number through S-type decision function calculation, including generation at creation, generation after approval, or generation when signing is completed;

[0010] S4. Manually review the generated contract numbering rules and generation nodes. After the review is passed, the contract number is output according to the contract numbering rules at the determined generation node.

[0011] By acquiring and standardizing the multi-dimensional attribute information of the contract, and using weight distribution and interactive component calculation to generate flexible contract numbering rules, the rules can be dynamically adapted according to the different attributes of the contract, solving the problems of rigid rules and disconnection from business. At the same time, the generation node of the contract number is determined through the S-type decision function, and the timing of number generation is reasonably determined under the premise of ensuring security, avoiding the imbalance between security and efficiency. In addition, the method introduces intelligent algorithms and weight functions to realize the automatic generation and version association of contract numbers, improve intelligence and scalability, and adapt to the needs of complex business scenarios.

[0012] As a further limitation of the technical solution of the present invention, step S1 specifically includes:

[0013] S11. Extract unstructured information from the contract text, including contract type, amount, contract source, language version, attachments, and contract version, through OCR parsing and NLP keyword extraction.

[0014] S12. Extract unstructured information including approval level, importance, and confidentiality level from the approval process system through API connection to the workflow engine;

[0015] S13. Convert the extracted unstructured information into multi-dimensional parameters including contract type, confidentiality level, contract amount, importance, approval method, attachment mark, contract version, language version and contract source and generate a parameter table.

[0016] By extracting information from the contract text through OCR parsing and NLP keyword extraction, as well as extracting information from the approval process system through API docking, we can comprehensively obtain the multi-dimensional attribute information of the contract, providing an accurate data basis for subsequent rule generation and node decision-making; converting the extracted information into multi-dimensional parameters and generating a parameter table to facilitate subsequent calculations and processing, thereby improving the operability and accuracy of the system.

[0017] As a further limitation of the technical solution of the present invention, the steps in S11 specifically include:

[0018] S111. Use OCR technology to parse contract texts in PDF or scanned format;

[0019] S112. Use NLP technology to extract unstructured information from the parsed contract text, including identification of contract type, amount, contract source, language version, attachment information, and contract version.

[0020] Using OCR technology to parse contract texts in different formats can process contract documents from multiple sources, improving the versatility of the system; using NLP technology to extract information from the parsed text can accurately identify the key attributes of the contract, providing accurate data for subsequent standardized processing.

[0021] As a further limitation of the technical solution of the present invention, step S13 specifically includes:

[0022] S131. Convert the extracted contract type, contract amount, contract source, language version, attachment information, and contract version into multi-dimensional parameters including contract type, confidentiality level, contract amount, importance, approval method, attachment tag, contract version, language version, and contract source;

[0023] S132. Quantify the contract type, confidentiality level, contract amount, importance, approval method, attachment mark, contract version, language version, and contract source to form standardized parameter values;

[0024] S133. Generate a parameter table based on the multi-dimensional parameters and corresponding parameter values.

[0025] The extracted information is converted into multidimensional parameters, and the parameters are quantified to form standardized parameter values, which facilitates subsequent weight distribution and calculation; a parameter table is generated to make the correspondence between parameters and parameter values ​​clear, thereby improving the maintainability and scalability of the system.

[0026] As a further limitation of the technical solution of the present invention, in S132, the step of performing quantization processing includes:

[0027] Quantify contract types into type codes;

[0028] The contract amount is quantified into segmented values; if the amount is less than A, the amount level is small; if the amount is between A and B, the amount level is medium; if the amount is greater than B, the amount level is large. The segmented values ​​are: small = 1, medium = 2, large = 3;

[0029] Quantify the contract source as a source coefficient; including INT=0, EXT=1;

[0030] Quantify language versions into language weights;

[0031] Quantize attachment information into Boolean values;

[0032] Quantify the contract version into version order; where the main contract = 0 and the supplementary agreement is equal to the order;

[0033] Quantify the approval level and importance into corresponding grade values;

[0034] The confidentiality level is quantified as a confidentiality coefficient, which takes values ​​of 0, 1, and 2.

[0035] Quantify the different attributes of the contract and convert unstructured information into structured numerical values ​​to facilitate subsequent mathematical calculations and rule generation; different quantification methods can accurately reflect the characteristics of the contract attributes and provide a scientific basis for the generation of contract numbering rules.

[0036] As a further limitation of the technical solution of the present invention, step S2 specifically includes:

[0037] S21. Construct a parameter matrix based on the parameter table, where each parameter corresponds to a dimension;

[0038] S22. Assign a weight to each parameter, where the weight is determined based on the business impact and security compliance requirements;

[0039] S23, multiplying each parameter and its weight by linear combination and summing them to obtain a basic component;

[0040] S24. For contracts that meet a specific set of conditions, the interaction component is calculated by multiplying the interaction factor and the multiplication coefficient;

[0041] S25. Determine a contract numbering rule based on the sum of the basic component and the interactive component, wherein the contract numbering rule includes a simple code rule, a standard rule, and an enhanced rule;

[0042] The specific condition set is: {confidentiality coefficient ≥ 1, contract amount level is greater than medium, and the approval method is not exempt from review}.

[0043] By constructing a parameter matrix and assigning weights to each parameter, the importance of each parameter in the generation of contract numbering rules can be reasonably determined based on the degree of business impact and security compliance requirements; by calculating the basic components through linear combination, the influence of each parameter can be comprehensively considered; for contracts that meet a specific set of conditions, the interactive components are calculated, which can further refine the contract numbering rules to make them more in line with the actual needs of the contract; the contract numbering rules are determined based on the sum of the basic components and the interactive components, which realizes the dynamic generation of rules and improves the flexibility and adaptability of the rules.

[0044] As a further limitation of the technical solution of the present invention, in S22, the step of assigning a weight to each parameter includes:

[0045] S221. Determine a set of contract attributes to be weighted, where the elements of the set include contract type, confidentiality level, contract amount, importance, approval method, attachment tag, contract version, language version, and contract source;

[0046] S222. Construct an attribute importance comparison matrix and score the relative importance of each two attributes;

[0047] S223. Calculate the eigenvector of the comparison matrix to obtain the initial weight of each attribute;

[0048] S224, perform consistency test, when the consistency ratio If < 0.1, the weight assignment is accepted, otherwise the comparison matrix is ​​re-adjusted;

[0049] S225. Fine-tune the initial weights according to business scenario requirements to determine the final weight vector.

[0050] By constructing an attribute importance comparison matrix and calculating the eigenvector, the initial weights of each attribute can be scientifically determined; consistency checks can be performed to ensure the rationality and accuracy of the weight distribution; and the initial weights can be fine-tuned according to business scenario requirements to make the weight distribution more in line with actual business needs, thereby improving the practicality and effectiveness of the contract numbering rules.

[0051] As a further limitation of the technical solution of the present invention, in S224, the step of performing consistency checking includes:

[0052] Compute the largest eigenvalue of the comparison matrix ;

[0053] Calculating consistency index ,in is the dimension of the comparison matrix, i.e. the number of attributes;

[0054] Calculating the consistency ratio ,in is the random consistency index;

[0055] when When <0.1, the weight distribution is considered consistent, otherwise the comparison matrix needs to be readjusted.

[0056] By calculating the maximum eigenvalue, consistency index and consistency ratio, the consistency of weight distribution can be scientifically evaluated; when the consistency ratio is less than 0.1, the weight distribution is accepted, otherwise the comparison matrix is ​​readjusted to ensure the rationality and accuracy of the weight distribution and improve the reliability of the generation of contract numbering rules.

[0057] As a further limitation of the technical solution of the present invention, in S23, the basic component calculation formula is:

[0058]

[0059] Where, is the weight corresponding to the parameter, is the element in the parameter matrix, the parameter value.

[0060] The basic component calculation formula multiplies each parameter and its weight and sums them up through linear combination. It can comprehensively consider the influence of each parameter and accurately calculate the basic component, providing an important basis for the subsequent determination of contract numbering rules.

[0061] As a further limitation of the technical solution of the present invention, in S24, the calculation formula of the interaction component is as follows:

[0062]

[0063] Where, is the interaction factor, is the multiplication coefficient when the condition is met, and C is a specific set of conditions.

[0064] For contracts that meet a specific set of conditions, the interaction component calculation formula calculates the interaction component by multiplying the interaction factor and the multiplication coefficient. This can further refine the contract numbering rules, making them more in line with the actual needs of the contract and improving the flexibility and adaptability of the contract numbering rules.

[0065] As a further limitation of the technical solution of the present invention, in S25, the step of determining the contract numbering rule based on the sum of the basic component and the interactive component includes:

[0066] S251. Add the basic component and the interactive component to obtain the contract number rule. value;

[0067] S252, according to The value range determines the numbering rule type, including simple code rule, standard rule, and enhanced rule.

[0068] The value of the contract numbering rule is obtained by adding the basic component and the interactive component, and the numbering rule type is determined according to the value range. This realizes the dynamic generation of contract numbering rules and can automatically select the appropriate numbering rule according to the different attributes of the contract, thereby improving the flexibility and adaptability of the rules.

[0069] As a further limitation of the technical solution of the present invention, in S252, according to The steps for determining the number sequence type using a value range include:

[0070] Value less than When using the short code rule, it includes: type + year and month + sequence number;

[0071] Value arrive Between and including When using the standard rules, including: type + unit + year and month + sequence number;

[0072] Value greater than or equal to When using the enhanced rules, including: type + confidentiality level + approval code + hash tail number.

[0073] According to different The value range determines different numbering rule types, so that the contract numbering rule can be reasonably selected according to the importance and complexity of the contract; the short code rule is applicable to simple contracts, the standard rule is applicable to general contracts, and the enhanced rule is applicable to important contracts, which improves the standardization and readability of contract numbers.

[0074] As a further limitation of the technical solution of the present invention, the calculation of the hash tail number includes:

[0075] Obtain the key attributes of the contract and perform a hash operation on the parameter combination values ​​of the key attributes of the contract; the key attributes include contract type, contract amount, importance, version number, and timestamp;

[0076] The last set number of digits of the hash value as the unique identification segment is the hash tail number.

[0077] By obtaining the key attributes of the contract and performing a hash operation to generate a hash tail number, a unique identifier can be provided for the contract number, thereby improving the security and uniqueness of the contract number; truncating the last set digits of the hash value as the hash tail number ensures uniqueness while avoiding the problem of the hash value being too long.

[0078] As a further limitation of the technical solution of the present invention, step S3 specifically includes:

[0079] S31. Determine the decision dimensions that influence the contract number generation node, including confidentiality level, contract amount and importance, approval complexity, and contract version relevance; set thresholds and weights for each dimension.

[0080] S32. Normalize the parameters of each dimension and calculate the detection item value of each dimension;

[0081] S33, using the S-type decision function to comprehensively evaluate the detection item values ​​of each dimension and calculate the comprehensive decision value;

[0082] S34. Determine the generation node of the contract number based on the comprehensive decision value and the preset node rule mapping table.

[0083] By identifying multiple decision dimensions that affect the contract number generation node and setting thresholds and weights for each dimension, the impact of different contract attributes on the generation node can be fully considered; the parameters of each dimension are normalized and the detection item value is calculated to facilitate subsequent comprehensive evaluation; the detection item values ​​of each dimension are comprehensively evaluated using an S-type decision function, and the comprehensive decision value can be scientifically calculated; the generation node is determined based on the comprehensive decision value combined with the preset node rule mapping table, realizing dynamic decision-making of the contract number generation node and improving the security and efficiency of the system.

[0084] As a further limitation of the technical solution of the present invention, in S31, the decision dimensions and detection items include:

[0085] In the confidentiality level dimension, the detection item is the ratio of the confidentiality level to the highest level;

[0086] Contract amount and importance dimension: the test item is (amount coefficient × importance) / (highest amount coefficient × highest importance);

[0087] Approval complexity dimension: the detection item is the ratio of approval complexity to the highest approval complexity;

[0088] Contract version correlation dimension, the detection item is contract version correlation.

[0089] Determining different decision dimensions and corresponding detection items can accurately reflect the impact of different contract attributes on the generation node; the detection items for the confidentiality level dimension, contract amount and importance dimension, approval complexity dimension and contract version correlation dimension are reasonably designed, which can comprehensively consider the key factors of the contract and provide a basis for the decision-making of the contract number generation node.

[0090] As a further limitation of the technical solution of the present invention, in S33, the step of using an S-type decision function to comprehensively evaluate the detection item values ​​of each dimension and calculating the comprehensive decision value includes:

[0091] Calculate the sign function value of each dimension based on the detection items and corresponding thresholds of each dimension;

[0092] The decision parameter value is obtained by weighted summing the sign function value of each dimension and the corresponding weight w;

[0093] The decision parameter values ​​are input into the S-type decision function to obtain the comprehensive decision value.

[0094] By calculating the sign function value of each dimension and summing the sign function value with the corresponding weight to obtain the decision parameter value, the influence of each dimension can be comprehensively considered; inputting the decision parameter value into the S-type decision function to obtain the comprehensive decision value can scientifically evaluate the detection item value of each dimension, providing an accurate basis for determining the contract number generation node.

[0095] As a further limitation of the technical solution of the present invention, in S33, the calculation formula is as follows:

[0096]

[0097] in,

[0098] Where, is the comprehensive decision value, is the weight of the k-th dimension, is the detection item value of the kth dimension, is the threshold of the kth dimension, m is the number of decision dimensions, for The function's variables, is the variable of the S-type decision function.

[0099] The comprehensive decision value calculation formula can scientifically calculate the comprehensive decision value by comprehensively considering the weights, detection item values ​​and thresholds of each dimension, providing an accurate basis for determining the contract number generation node; the application of the S-type decision function makes the calculation of the comprehensive decision value more reasonable and accurate, and improves the decision-making ability of the system.

[0100] In a second aspect, the technical solution of the present invention further provides a contract number recommendation system integrating multi-dimensional attributes, including:

[0101] A data collection module is used to obtain unstructured contract information, extract and standardize it into multi-dimensional contract attribute information, including contract type, confidentiality level, contract amount, importance, approval method, attachment tag, contract version, language version, and contract source;

[0102] A rule generation module, configured to generate contract numbering rules based on the multi-dimensional contract attribute information by weight allocation and interactive component calculation, wherein the contract numbering rules include a shortcode rule, a standard rule, and an enhanced rule;

[0103] A node decision module is used to determine, based on the multi-dimensional contract attribute information, a generation node of the contract number through an S-type decision function calculation, wherein the generation node includes generation at creation, generation after approval, or generation upon completion of signing;

[0104] The audit execution module is used to manually audit the generated contract numbering rules and generation nodes, and after the audit is passed, output the contract number according to the contract numbering rules at the determined generation node.

[0105] As a further limitation of the technical solution of the present invention, the data acquisition module includes:

[0106] The text parsing unit is used to extract unstructured information from the contract text, including contract type, amount, contract source, language version, attachment information, and contract version, through OCR parsing and NLP keyword extraction;

[0107] The approval information extraction unit is used to extract unstructured information including approval level, importance, and confidentiality level from the approval process system through API docking with the workflow engine;

[0108] The parameter standardization unit is used to convert the extracted unstructured information into multi-dimensional parameters including contract type, confidentiality level, contract amount, importance, approval method, attachment mark, contract version, language version and contract source and generate a parameter table.

[0109] The formula for calculating the basic component is:

[0110]

[0111] Where, is the weight corresponding to the parameter, is the element in the parameter matrix, the parameter value.

[0112] The calculation formula for calculating the interaction component is as follows:

[0113]

[0114] Where, is the interaction factor, is the multiplication coefficient when the condition is met, and C is a specific set of conditions.

[0115] The calculation formula of the S-type decision function is as follows:

[0116]

[0117] in,

[0118] Where, is the comprehensive decision value, is the weight of the k-th dimension, is the detection item value of the kth dimension, is the threshold of the kth dimension, m is the number of decision dimensions, for The function's variables, is the variable of the S-type decision function.

[0119] It can be seen from the above technical solutions that this application has the following advantages: it comprehensively considers multiple key attribute factors of the contract, and can generate precisely matched numbering rules for contracts of different types and situations, so that the numbering becomes a concentrated reflection of the core information of the contract, which facilitates the rapid identification of the key characteristics of the contract during the contract management process and improves the efficiency of contract retrieval, classification and management.

[0120] For top-secret contracts, an irregular random numbering method is adopted, and it is recommended to generate numbers at relatively safe nodes such as archiving. This effectively reduces the risk of leaks caused by predictable numbering patterns and enhances the protection of highly confidential contracts. The optimal numbering generation node is recommended based on the attributes of different contracts, taking into account both the process simplicity requirements of ordinary contracts and the rigorous process control needs of important and complex contracts, ensuring that the timing of number generation is highly consistent with the actual business process of the contract, and avoiding unnecessary interference with the contract business flow or incorrect associations caused by the numbering process. In particular, for contracts such as supplementary agreements that have an associated relationship with the main contract, their subordinate and related information can be clearly reflected in the numbering, which helps maintain the integrity and consistency of the contract system, and facilitates users to view and trace the evolution of the contract and the internal connections between related contracts, providing strong support for the full life cycle management of the contract.

[0121] It realizes the automatic generation of contract numbers and intelligent recommendation of generation nodes, reduces manual intervention, avoids errors, duplications or irregularities that may be caused by manual numbering, improves the automation level and standardization of contract number management, helps enterprises or organizations establish a unified and orderly contract management system, and improves the overall contract management level and operational efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0122] In order to more clearly illustrate the technical solution of the present application, the following is a brief introduction to the drawings required for the description. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0123] Figure 1 A flowchart of a method provided in an embodiment of the present invention.

[0124] Figure 2 This is a flowchart of determining contract numbering rules in an embodiment of the present invention.

[0125] Figure 3 This is a flowchart of determining a generation node in an embodiment of the present invention.

[0126] Figure 4 This is a flow chart of assigning weights to parameters in an embodiment of the present invention.

[0127] Figure 5 A system block diagram provided for an embodiment of the present invention. DETAILED DESCRIPTION

[0128] In order to make the application objectives, features, and advantages of this application more obvious and easy to understand, the technical solutions protected by this application will be clearly and completely described below using specific embodiments and drawings. Obviously, the embodiments described below are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0129] like Figure 1 As shown, an embodiment of the present invention provides a method for recommending contract numbers based on multi-dimensional attribute fusion, comprising the following steps:

[0130] S1. Obtain unstructured contract information, extract and standardize it into multi-dimensional contract attribute information, including contract type, confidentiality level, contract amount, importance, approval method, attachment tag, contract version, language version, and contract source;

[0131] This step specifically includes: using OCR technology to parse the contract text in PDF or scanned format; using NLP technology to extract unstructured information from the parsed contract text, including identifying the contract type, amount, contract source, language version, attachments, and contract version;

[0132] S11. Use OCR technology to parse contract text in PDF or scanned format. Specific methods include: Preprocessing the PDF file or scanned image. For PDF files, use a PDF parsing library (such as PyMuPDF or PDFMiner in Python) to convert the PDF pages to images or directly extract the text. For scanned images, perform image preprocessing, including grayscale conversion, binarization, and noise removal, to improve OCR recognition accuracy and obtain a clear image or initially extract the text content. Use an OCR tool (such as Tesseract OCR, ABBYY FineReader, or other commercial OCR software) to perform text recognition on the preprocessed image or PDF page. For PDF files, use an OCR tool that supports PDF directly for recognition. Configure the OCR tool's language model to support the language of the contract (such as Chinese, English, etc.). Obtain the recognized text content, which may contain some recognition errors. Correct the recognized text content to correct common OCR errors (such as misidentifying a "0" as an "O" or a "1" as an "I"). If the text content is long, it can be processed in segments to improve the efficiency of subsequent processing. The corrected text content is used for subsequent NLP processing.

[0133] S12. Use NLP technology to extract unstructured information from the parsed contract text. Segment the OCR-corrected text into paragraphs or logical structures. Clean the text, removing extra spaces, line breaks, and special characters, to produce cleaned segmented text. Based on the cleaned segmented text, use keyword matching or a pre-trained text classification model (such as a BERT-based classification model) to identify the contract type.

[0134] Contract category identification process:

[0135] Define a keyword dictionary, for example:

[0136] Sales contract: Keywords include "sales", "purchase", "payment", etc.

[0137] Lease contract: Keywords include "lease period", "house", "rent payment", etc.

[0138] Scan the text for keywords, determine the contract type after matching the corresponding keywords, and obtain the identified contract type (such as "sales contract" or "lease contract").

[0139] Contract amount extraction process:

[0140] Based on the cleaned segmented text, regular expressions are used to match dollar amount formats (e.g., "￥1000," "$2000," "1000 yuan," etc.). For complex dollar amount expressions (e.g., "One thousand yuan"), a pre-trained NER (Named Entity Recognition) model is used to extract the dollar amount. The amount is then normalized (e.g., converting "One thousand yuan" to "1000 yuan"). This yields the contract amount and its normalized value.

[0141] Contract Source Identification Process:

[0142] Check whether the cleaned segmented text contains specific identifying information, such as "internal contract" or "external contract." If there is no clear identification, you can use the company name, address, and other information in the contract to determine the source of the contract (e.g., "internal contract" or "external contract").

[0143] Language version identification process:

[0144] Use a language detection tool (such as the langdetect library) to detect the language of the cleaned segmented text. If the contract contains multiple languages, you can detect the language of each segment separately to obtain the language version of the contract (e.g., "Chinese" or "English").

[0145] Attachment information identification process:

[0146] Check whether the cleaned segmented text contains keywords such as "attachment" and "appendix". If there is attachment information, extract the attachment description content. Get the existence and description of the attachment information.

[0147] Contract version identification process:

[0148] Check whether the cleaned segmented text contains version identifiers, such as "Main Contract," "Supplementary Agreement 1," "Version 2," etc. If there are no clear identifiers, you can infer the contract version information based on the structure and logic of the text content (such as "Main Contract," "Supplementary Agreement 1").

[0149] Here, the training steps of the pre-trained model used above are well-known and will not be described in detail here.

[0150] Extract unstructured information including approval level, importance, and confidentiality level from the approval process system through API connection to the workflow engine; specifically:

[0151] Determine the API interface for the approval process system, including the interface address, request method, authentication method, parameter list, etc. Confirm the API functions supported by the approval process system, such as obtaining approval process details, approval levels, importance, confidentiality level, and other information. Confirm the API call methods supported by the workflow engine, such as HTTP requests and webhooks. Design the trigger conditions for API calls, such as automatically calling the approval process system API when a contract is submitted for approval.

[0152] Define the data fields to be extracted:

[0153] Approval level: indicates the hierarchical structure of the approval process, such as first-level approval, second-level approval, etc.

[0154] Importance: Indicates the importance of the contract, such as general, important, and urgent.

[0155] Confidentiality level: Indicates the confidentiality level of the contract, such as public, secret, and top secret.

[0156] Design data extraction logic:

[0157] Design the data extraction logic according to the API documentation of the approval process system, including how to extract the required fields from the returned JSON or XML data.

[0158] For example, by parsing the JSON data returned by the API, extract fields such as approvalLevel, importance, and confidentiality.

[0159] Use programming languages ​​(such as Python, Java, etc.) to write API call code to connect with the approval process system and obtain information such as approval level, importance, and confidentiality level.

[0160] S13. Convert the extracted unstructured information into multi-dimensional parameters including contract type, confidentiality level, contract amount, importance, approval method, attachment mark, contract version, language version and contract source and generate a parameter table. Specifically, it includes:

[0161] S131. Convert the extracted contract type, contract amount, contract source, language version, attachment information, and contract version into multi-dimensional parameters including contract type, confidentiality level, contract amount, importance, approval method, attachment tag, contract version, language version, and contract source;

[0162] S132: Quantify the contract type, confidentiality level, contract amount, importance, approval method, attachment mark, contract version, language version, and contract source to form standardized parameter values. In this step, the quantification steps in S132 include:

[0163] Quantify contract types into type codes;

[0164] The contract amount is quantified into segmented values; if the amount is less than A, the amount level is small; if the amount is between A and B, the amount level is medium; if the amount is greater than B, the amount level is large. The segmented values ​​are: small = 1, medium = 2, large = 3;

[0165] Quantify the contract source as a source coefficient; including INT=0, EXT=1;

[0166] Quantify language versions into language weights;

[0167] Quantize attachment information into Boolean values;

[0168] Quantify the contract version into version order; where the main contract = 0 and the supplementary agreement is equal to the order;

[0169] Quantify the approval level and importance into corresponding grade values;

[0170] The confidentiality level is quantified as a confidentiality coefficient, which takes values ​​of 0, 1, and 2.

[0171] S133. Generate a parameter table based on the multi-dimensional parameters and corresponding parameter values, as shown in Table 1.

[0172] Table 1: Parameters

[0173]

[0174] S2. Based on the multi-dimensional contract attribute information, generate contract numbering rules through weight allocation and interactive component calculation; including simple code rules, standard rules and enhanced rules; such as Figure 2 As shown, specifically including:

[0175] S21. Construct a parameter matrix based on the parameter table, where each parameter corresponds to a dimension; P = [contract type, confidentiality level, contract amount, importance, approval method, attachment mark, contract version, language version, contract source]; the characteristics of a contract can be represented by a parameter matrix.

[0176] S22. Assign a weight to each parameter, where the weight is determined based on the business impact and security compliance requirements;

[0177] In this embodiment of the present invention, parameter prioritization is based on business impact and security and compliance requirements. Business impact: Attributes with a significant impact on the contract management process should be given a higher weight. Security and compliance requirements: Attributes related to legal validity and information security must be prioritized. ① Contract type: Impacts classification management and can be assigned a high priority. ② Confidentiality level: Involves trade secrets, and if leaked, the consequences are severe; therefore, the highest priority can be assigned. ③ Contract amount: Directly impacts financial risk and can be assigned a high priority. ④ Importance: Involves operational impact and can be assigned a medium-high priority. ④ Approval method: Determines whether the contract requires multi-level review, affecting process length and resource utilization; can be assigned a medium-high priority. ⑤ Attachment tagging and language version: Involves management convenience and can be assigned a medium-low priority. ⑥ Contract version: Involves the traceability of the contract text (e.g., the relationship between the main contract and the supplementary agreement); can be assigned a medium-low priority. ⑦ Contract source: Internal / external drafting affects the traceability mechanism; can be assigned a medium-low priority. Based on priority, weights are assigned using the Analytic Hierarchy Process (AHP) to quantify the importance of business rules. Based on the business scores, the nine attributes are compared pairwise to dynamically generate a parameter mapping table, as shown in Table 2.

[0178] Table 2: Parameter mapping table

[0179]

[0180] S23, multiplying each parameter and its weight by linear combination and summing them to obtain a basic component;

[0181] Basic component calculation formula:

[0182]

[0183] Where, is the weight corresponding to the parameter, is the element in the parameter matrix, the parameter value.

[0184] S24. For contracts that meet a specific set of conditions, the interaction component is calculated by multiplying the interaction factor and the multiplication coefficient;

[0185] Contracts with a confidentiality level greater than or equal to 1, a contract amount greater than medium, and approval not equal to exemption from review are relatively important contracts and require special treatment. Therefore, interaction variables are designed to enhance the weight of such contracts.

[0186] The calculation formula of the interaction component is as follows:

[0187]

[0188] Where, is the interaction factor, which is 0.2 in this embodiment, is the multiplication factor when the condition is met ( , , ), C is a specific condition set. The specific condition set is: {confidentiality coefficient ≥ 1, contract amount level greater than medium, approval method is not exempt from review}.

[0189] S25. Determine the contract numbering rule based on the sum of the basic component and the interactive component. Specifically, add the basic component and the interactive component to obtain the contract numbering rule. value; according to The value range determines the numbering rule type, including simple code rule, standard rule, and enhanced rule.

[0190] It should be noted here that ① the contract number rule generation function is obtained by adding the calculated values ​​of the basic component and the interactive component; ② The calculation of the value corresponds to the security, efficiency, and cost of the above-mentioned weight system; ③ The source of the multiplication coefficient in the interactive variable needs to be analyzed from the perspectives of business logic and mathematics. Business logic: The three r coefficients correspond to the three dimensions of confidentiality, amount, and approval, which are related to the risk control level. The highest confidentiality coefficient of 1.5 is in line with common sense (confidentiality has the greatest weight), and the lowest approval coefficient of 1.2 is also reasonable (process control is relatively minor). Mathematical level: The design of r>1 ensures that a multiplier effect is generated when conditions are superimposed. For example, when confidentiality and amount are triggered at the same time, they are not simply added (1.5+1.3=2.8), but multiplied (1.5*1.3=1.95), which can better reflect the nonlinear growth of risk superposition and needs to be explained. =0.2 is used to suppress the damping coefficient of the overfitting design. , the superposition of multiple conditions will cause the risk value to soar too quickly (such as three r's multiplied by 2.34 times), The actual increase was only 0.468 times, which is more in line with business reality.

[0191] For the maintainability of parameters, these three r values ​​can be adjusted in the background according to the actual situation of the enterprise. Since r>2 will lead to excessive expansion of risk value, r is best controlled in the range of 1.0-2.0. ④ Multiplication coefficient Once determined, the interaction component is not a constant value because its calculation is triple dynamic, as shown in Table 3.

[0192] Table 3: Dynamic Description Table

[0193]

[0194] in accordance with The value is combined with the preset numbering rule mapping table, as shown in Table 4, to infer the type of numbering rule.

[0195] Table 4: Numbering sequence mapping table

[0196]

[0197] The calculation of the hash tail number includes: obtaining the key attributes of the contract and performing hash operations on the parameter combination values ​​of the key attributes of the contract; the key attributes include contract type, contract amount, importance, version number, and timestamp; intercepting the hash value as the last set number of digits of the unique identification segment, which is the hash tail number.

[0198] The hash algorithm uses SHA-256, and the last 6 hexadecimal characters of the hash value are used as the tail. For example:

[0199] Key contract attributes include:

[0200] Contract Type: XS (Sales) (Quantitative Value P1 = 2)

[0201] Contract amount: 6 million (quantified value P3=3)

[0202] Importance: Urgent (quantitative value P4=2)

[0203] Contract version number: Main contract (P7=0)

[0204] Timestamp: 2025-06-05 08:30:22

[0205] First, build the hash input string: input="XS_600_2_0_20250605083022"

[0206] Then, perform SHA-256 hash calculation: Original hash: e3b0c44298fc1c14......a3b2c8 Truncated tail number: a3b2c8

[0207] Embedded numbering rules: Enhanced rules: type + confidentiality level + approval code + hash tail number

[0208] Final number: XS-SEC-L3-a3b2c8.

[0209] S3. Based on the multi-dimensional contract attribute information, determine the generation node of the contract number through the S-type decision function calculation, including generation at creation, generation after approval, or generation when signing is completed; Figure 3 As shown, specifically including:

[0210] S31. Determine the decision dimensions that affect the contract number generation node, including the confidentiality level dimension, the contract amount and importance dimension, the approval complexity dimension, and the contract version correlation dimension; set thresholds and weights for each dimension; and generate a decision dimension table, as shown in Table 5.

[0211] Table 5: Decision Dimension Table

[0212]

[0213] Here, dimension 1 is the confidentiality level, the threshold θ is set to 0.7, the weight w value is set to 0.3, and the detection item T is the confidentiality level / highest level.

[0214] Dimension 2 is the contract amount and importance. The threshold θ is set to 0.6, the weight w is set to 0.25, and the detection item T is (amount coefficient × importance) / (highest amount coefficient × highest importance). The highest amount coefficient × highest importance in the parameter mapping table is 3 × 2.

[0215] Dimension 3 is the approval method, the threshold θ is set to 0.5, the weight w is set to 0.25, and the detection item T is the approval complexity / maximum approval complexity.

[0216] Dimension 4 is the contract version, the threshold θ is set to 0.4, the weight w value is set to 0.2, and the detection item T is the contract version correlation.

[0217] S32. Normalize the parameters of each dimension and calculate the detection item value of each dimension;

[0218] S33. Use the S-type decision function to comprehensively evaluate the detection item values ​​of each dimension and calculate the comprehensive decision value; specifically, the following steps are included:

[0219] Calculate the sign function value of each dimension based on the detection items and corresponding thresholds of each dimension;

[0220] The decision parameter value is obtained by weighted summing the sign function value of each dimension and the corresponding weight w;

[0221] The decision parameter values ​​are input into the S-type decision function to obtain the comprehensive decision value.

[0222] The calculation formula is as follows:

[0223]

[0224] in,

[0225] Where, is the comprehensive decision value, is the weight of the k-th dimension, is the detection item value of the kth dimension, is the threshold of the kth dimension, m is the number of decision dimensions, for The function's variables, is the variable of the S-type decision function.

[0226] Here, an S-shaped decision function is used to set an inflection point at a threshold of 0.5, reflecting the business characteristic of "stable at low risk and steep at high risk." The exponential coefficient of 5 amplifies the gradient change between 0.5 and 1, and the offset of 0.6 (60% is the medium risk threshold) matches the company's risk appetite.

[0227] Analysis of the S decision function shows that the final value shows a nonlinear response according to the change of the input parameter. When the parameter is less than or equal to 0.5, it grows slowly, which meets the requirement of low-risk stability. When the parameter is greater than 0.5, it grows rapidly, which meets the requirement of high-risk steep rise.

[0228] S34, based on the comprehensive decision value , combined with the preset node rule mapping table, as shown in Table 6, determine the generation node of the contract number.

[0229] Table 6: Node rule mapping table

[0230]

[0231] In some embodiments, such as Figure 4 As shown, the steps to assign weights to each parameter include:

[0232] S221. Determine a set of contract attributes to be weighted, where the elements of the set include contract type, confidentiality level, contract amount, importance, approval method, attachment tag, contract version, language version, and contract source;

[0233] S222. Construct an attribute importance comparison matrix and score the relative importance of each two attributes; specifically, the following steps are performed: construct an attribute importance comparison matrix; domain experts score the relative importance of each two attributes based on business experience and security compliance requirements; the scoring adopts a 1-9 scale, where 1 indicates that the two attributes are equally important and 9 indicates that one attribute is significantly more important than the other.

[0234] A scale of 1 means that both attributes are equally important;

[0235] Scale 3 means the former is slightly more important than the latter;

[0236] A scale of 5 indicates that the former is significantly more important than the latter;

[0237] A scale of 7 indicates that the former is strongly more important than the latter;

[0238] A scale of 9 indicates that the former is extremely more important than the latter.

[0239] S223. Calculate the eigenvector of the comparison matrix to obtain the initial weight of each attribute;

[0240] S224, perform consistency test, when the consistency ratio If < 0.1, the weight assignment is accepted, otherwise the comparison matrix is ​​re-adjusted;

[0241] S225. Fine-tune the initial weights according to business scenario requirements to determine the final weight vector.

[0242] In S224, the steps of performing consistency check include:

[0243] Compute the largest eigenvalue of the comparison matrix ;

[0244] Calculating consistency index ,in is the dimension of the comparison matrix, i.e. the number of attributes;

[0245] Calculating the consistency ratio ,in is the random consistency index;

[0246] when When <0.1, the weight distribution is considered consistent, otherwise the comparison matrix needs to be readjusted.

[0247] In S225, the weight fine-tuning step includes:

[0248] Adjust the initial weights based on actual business scenarios and risk appetite to ensure that the weight distribution reflects the impact of different parameters on the generation of contract numbering rules;

[0249] The adjusted weights need to be re-checked for consistency to ensure that the adjusted weight distribution remains reasonable.

[0250] The above calculation logic can be used to infer the generation rules and generation nodes of the contract number.

[0251] Specific examples are as follows:

[0252] Assume the following input parameters: Contract Type = XS, Confidentiality Level = Top Secret, Contract Amount = Large, Importance = Urgent, Approval Method = Multi-Level Approval, Attachment Existence = Yes, Contract Version = Main Contract, Language Version = EN, Contract Source = EXT;

[0253] Parameter matrix P = [2,2,3,2,2,1,0,1,1]

[0254] Contract number rule generation calculation:

[0255] (1) Calculate the basic components

[0256] 0.15×2+0.25×2+0.18×3+0.12×2+0.10×2+0.05×1+0.08×0+0.04×1+0.03×1=0.3+0.5+0.54+0.24+0.2+0.05+0+0.04+0.03=1.9

[0257] (2) Calculation of interaction components

[0258] Because the input parameters meet the confidentiality of the contract ( ), Larger amount contracts ( ), multi-level approval ( ) Therefore, the sub-factors in the interaction component are enhanced, and the calculation is as follows:

[0259] 0.2×(1.5×1.3×1.2)=0.2×2.34=0.468

[0260] (3) =1.9+0.468=2.368

[0261] Refer to the number sequence mapping table to see that the enhanced rule should be used.

[0262] Node decision calculation:

[0263]

[0264]

[0265]

[0266]

[0267]

[0268] Referring to the node rule mapping table, we can see that 0.731>0.7 high-risk contracts require the node to generate a contract number after signing.

[0269] S4. Manually review the generated contract numbering rules and generation nodes. Once the review is passed, the contract number is output at the determined generation node according to the contract numbering rules. The addition of a manual review node is primarily intended to review the coding rules and generation nodes generated by the recommendation algorithm. Manual review is essential to ensure the accuracy of the final result.

[0270] like Figure 5 As shown, an embodiment of the present invention further provides a contract number recommendation system integrating multi-dimensional attributes, including:

[0271] A data collection module is used to obtain unstructured contract information, extract and standardize it into multi-dimensional contract attribute information, including contract type, confidentiality level, contract amount, importance, approval method, attachment tag, contract version, language version, and contract source;

[0272] A rule generation module, configured to generate contract numbering rules based on the multi-dimensional contract attribute information by weight allocation and interactive component calculation, wherein the contract numbering rules include a shortcode rule, a standard rule, and an enhanced rule;

[0273] A node decision module is used to determine, based on the multi-dimensional contract attribute information, a generation node of the contract number through an S-type decision function calculation, wherein the generation node includes generation at creation, generation after approval, or generation upon completion of signing;

[0274] The audit execution module is used to manually audit the generated contract numbering rules and generation nodes, and after the audit is passed, output the contract number according to the contract numbering rules at the determined generation node.

[0275] In some embodiments, the data acquisition module includes:

[0276] The text parsing unit is used to extract unstructured information from the contract text, including contract type, amount, contract source, language version, attachment information, and contract version, through OCR parsing and NLP keyword extraction;

[0277] The approval information extraction unit is used to extract unstructured information including approval level, importance, and confidentiality level from the approval process system through API docking with the workflow engine;

[0278] The parameter standardization unit is used to convert the extracted unstructured information into multi-dimensional parameters including contract type, confidentiality level, contract amount, importance, approval method, attachment mark, contract version, language version and contract source and generate a parameter table.

[0279] In some embodiments, the text parsing unit includes:

[0280] The OCR parsing subunit is used to parse the contract text in PDF or scanned format using OCR technology;

[0281] The NLP extraction subunit is used to use NLP technology to extract unstructured information from the parsed contract text, including identification of contract type, amount, contract source, language version, attachment information and contract version.

[0282] In some embodiments, the parameter normalization unit includes:

[0283] The parameter conversion sub-unit is used to convert the extracted contract type, contract amount, contract source, language version, attachment information and contract version into multi-dimensional parameters of contract type, confidentiality level, contract amount, importance, approval method, attachment tag, contract version, language version and contract source;

[0284] The parameter quantification sub-unit is used to quantify the contract type, confidentiality level, contract amount, importance, approval method, attachment mark, contract version, language version and contract source to form standardized parameter values;

[0285] The parameter table generating subunit is used to generate a parameter table from multi-dimensional parameters and corresponding parameter values.

[0286] In some embodiments, the parameter quantization subunit includes:

[0287] The contract type quantification submodule is used to quantify the contract type into a type code;

[0288] The contract amount quantification submodule is used to quantify the contract amount into segmented values. When the amount is less than A, the amount level is small; when the amount is between A and B, the amount level is medium; when the amount is greater than B, the amount level is large. The segmented values ​​are: small = 1, medium = 2, and large = 3.

[0289] The contract source quantification submodule is used to quantify the contract source into a source coefficient; including INT=0, EXT=1;

[0290] The language version quantification submodule is used to quantify the language version into language weights;

[0291] An attachment information quantization submodule, used to quantize the attachment information into a Boolean value;

[0292] The contract version quantification submodule is used to quantify the contract version into version orders; where the main contract = 0 and the supplementary agreement is equal to the order;

[0293] The approval level and importance quantification submodule is used to quantify the approval level and importance into corresponding grade values;

[0294] The confidentiality level quantification submodule is used to quantify the confidentiality level into a confidentiality coefficient with values ​​of 0, 1, and 2.

[0295] In some embodiments, the rule generation module includes:

[0296] The parameter matrix construction unit is used to construct a parameter matrix according to the parameter table, where each parameter corresponds to a dimension;

[0297] A weight allocation unit, configured to allocate a weight to each parameter, wherein the weight is determined based on the business impact and security compliance requirements;

[0298] A basic component calculation unit is used to multiply and sum each parameter with its weight in a linear combination manner to obtain a basic component;

[0299] An interaction component calculation unit, for calculating the interaction component by multiplying the interaction factor and the multiplication coefficient for contracts that meet a specific set of conditions;

[0300] a numbering rule determining unit, configured to determine a contract numbering rule based on the sum of the basic component and the interactive component, wherein the contract numbering rule includes a shortcode rule, a standard rule, and an enhanced rule;

[0301] The specific condition set is: {confidentiality coefficient ≥ 1, contract amount level is greater than medium, and approval method is not exempt from review}.

[0302] In some embodiments, the weight allocation unit includes:

[0303] The attribute set determination subunit is used to determine the contract attribute set to be weighted. The set elements include contract type, confidentiality level, contract amount, importance, approval method, attachment mark, contract version, language version and contract source;

[0304] The comparison matrix construction subunit is used to construct an attribute importance comparison matrix and score the relative importance of every two attributes;

[0305] The initial weight calculation subunit is used to calculate the eigenvector of the comparison matrix and obtain the initial weight of each attribute;

[0306] The consistency check subunit is used to perform consistency check and accept weight assignment when the consistency ratio is less than 0.1, otherwise the comparison matrix is ​​readjusted;

[0307] The weight fine-tuning subunit is used to fine-tune the initial weight according to the business scenario requirements and determine the final weight vector.

[0308] In some embodiments, the consistency check subunit includes:

[0309] Maximum eigenvalue calculation submodule, used to calculate the maximum eigenvalue of the comparison matrix ;

[0310] Consistency index calculation submodule, used to calculate consistency index ,in is the dimension of the comparison matrix, i.e. the number of attributes;

[0311] Consistency ratio calculation submodule, used to calculate consistency ratio ,in is the random consistency index;

[0312] The consistency judgment submodule is used to When <0.1, the weight distribution is considered consistent, otherwise the comparison matrix needs to be readjusted.

[0313] In some embodiments, the basic component calculation unit calculates the basic component using the formula:

[0314]

[0315] Where, is the weight corresponding to the parameter, is the element in the parameter matrix, the parameter value.

[0316] The calculation formula of the interaction component calculation unit for calculating the interaction component is as follows:

[0317]

[0318] Where, is the interaction factor, is the multiplication coefficient when the condition is met, and C is a specific set of conditions.

[0319] In some embodiments, the numbering rule determining unit includes:

[0320] The calculation subunit is used to add the basic component and the interactive component to obtain the contract number rule. value;

[0321] Rule type determination subunit, used to determine the The value range determines the numbering rule type, including simple code rule, standard rule, and enhanced rule.

[0322] Value less than When using the short code rule, it includes: type + year and month + sequence number;

[0323] Value arrive Between and including When using the standard rules, including: type + unit + year and month + sequence number;

[0324] Value greater than or equal to When using the enhanced rules, including: type + confidentiality level + approval code + hash tail number.

[0325] In some embodiments, the hash tail number calculation unit includes:

[0326] The key attribute acquisition sub-unit is used to obtain the key attributes of the contract and perform hash operations on the parameter combination values ​​of the key attributes of the contract; the key attributes include contract type, contract amount, importance, version number, and timestamp;

[0327] The hash tail number truncation subunit is used to truncate the hash value as the last set number of digits of the unique identification segment, that is, the hash tail number.

[0328] In some embodiments, the node decision module includes:

[0329] The decision dimension determination unit is used to determine the decision dimensions that affect the contract number generation node, including the confidentiality level dimension, the contract amount and importance dimension, the approval complexity dimension, and the contract version correlation dimension; and set thresholds and weights for each dimension;

[0330] A detection item calculation unit, used to normalize the parameters of each dimension and calculate the detection item value of each dimension;

[0331] A comprehensive decision value calculation unit is used to use an S-type decision function to comprehensively evaluate the detection item values ​​of each dimension and calculate a comprehensive decision value;

[0332] The generation node determination unit is used to determine the generation node of the contract number based on the comprehensive decision value and the preset node rule mapping table.

[0333] In some embodiments, the decision dimension determination unit includes:

[0334] The decision dimension and detection item determination subunit is used to determine the decision dimension and detection items, including:

[0335] In the confidentiality level dimension, the detection item is the ratio of the confidentiality level to the highest level;

[0336] Contract amount and importance dimension: the test item is (amount coefficient × importance) / (highest amount coefficient × highest importance);

[0337] Approval complexity dimension: the detection item is the ratio of approval complexity to the highest approval complexity;

[0338] Contract version correlation dimension, the detection item is contract version correlation.

[0339] In some embodiments, the comprehensive decision value calculation unit includes:

[0340] The sign function value calculation subunit is used to calculate the sign function value of each dimension based on the detection items and corresponding thresholds of each dimension;

[0341] The decision parameter value calculation subunit is used to obtain the decision parameter value by weighted summing the sign function value of each dimension and the corresponding weight w;

[0342] The comprehensive decision value calculation subunit is used to input the decision parameter value into the S-type decision function to obtain the comprehensive decision value.

[0343] In some embodiments, the calculation formula of the comprehensive decision value calculation unit is as follows:

[0344]

[0345] in,

[0346] Where, is the comprehensive decision value, is the weight of the k-th dimension, is the detection item value of the kth dimension, is the threshold of the kth dimension, m is the number of decision dimensions, for The function's variables, is the variable of the S-type decision function.

[0347] An embodiment of the present invention further provides an electronic device comprising: a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other via the communication bus. The communication bus can be used to transmit information between the electronic device and a sensor. The processor can invoke logic instructions in the memory to execute the following method: S1. Obtaining unstructured contract information, extracting and standardizing it into multidimensional contract attribute information, including contract type, confidentiality level, contract amount, importance, approval method, attachment tag, contract version, language version, and contract source; S2. Based on the multidimensional contract attribute information, generating a contract numbering rule through weight allocation and interactive component calculation, including a simplified code rule, a standard rule, and an enhanced rule; S3. Based on the multidimensional contract attribute information, determining a generation node for the contract number through an S-type decision function calculation, including generation upon creation, generation after approval, or generation upon signing; S4. Manually reviewing the generated contract numbering rule and generation node. Upon successful review, outputting the contract number at the determined generation node according to the contract numbering rule.

[0348] Furthermore, the logical instructions in the aforementioned memory can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage media include various media capable of storing program code, such as USB flash drives, mobile hard drives, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical disks.

[0349] An embodiment of the present invention provides a non-transitory computer-readable storage medium, which stores computer instructions, and the computer instructions enable a computer to execute the method provided by the above-mentioned method embodiment, for example, including: S1, obtaining unstructured information of the contract, extracting and standardizing it into multi-dimensional contract attribute information, including contract type, confidentiality level, contract amount, importance, approval method, attachment mark, contract version, language version and contract source; S2, based on the multi-dimensional contract attribute information, generating contract numbering rules through weight distribution and interactive component calculation; including simple code rules, standard rules and enhanced rules; S3, based on the multi-dimensional contract attribute information, determining the generation node of the contract number through S-type decision function calculation, including generation at creation, generation after approval or generation when signing is completed; S4, manually reviewing the generated contract numbering rules and generation nodes, and after the review is passed, outputting the contract number according to the contract numbering rules at the determined generation node.

[0350] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not limited to the embodiments shown herein but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A multi-dimensional attribute fusion contract number recommendation method, characterized in that: The following steps are involved: S1. Obtain unstructured contract information, extract and standardize it into multi-dimensional contract attribute information, including contract type, confidentiality level, contract amount, importance, approval method, attachment tag, contract version, language version, and contract source. Specifically, the extracted unstructured information is converted into multi-dimensional parameters including contract type, confidentiality level, contract amount, importance, approval method, attachment tag, contract version, language version, and contract source, and a parameter table is generated. S2. Based on the multi-dimensional contract attribute information, generate contract numbering rules by weight allocation and interactive component calculation; including simple code rules, standard rules and enhanced rules; S3. Based on the multi-dimensional contract attribute information, determine the generation node of the contract number through S-type decision function calculation, including generation at creation, generation after approval, or generation when signing is completed; S4. Manually review the generated contract numbering rules and generation nodes. After passing the review, output the contract number according to the contract numbering rules at the determined generation node; The steps of S2 specifically include: S21. Construct a parameter matrix based on the parameter table, where each parameter corresponds to a dimension; S22. Assign a weight to each parameter, where the weight is determined based on the business impact and security compliance requirements; S23, multiplying each parameter and its weight by linear combination and summing them to obtain a basic component; S24. For contracts that meet a specific set of conditions, the interaction component is calculated by multiplying the interaction factor and the multiplication coefficient; S25. Determine a contract numbering rule based on the sum of the basic component and the interactive component. The contract numbering rule includes a simple code rule, a standard rule, and an enhanced rule. Specifically, the contract numbering rule includes: S251. Add the basic component and the interactive component to obtain the contract number rule. value; S252, according to The value range determines the numbering scheme type, including: Value less than When using the short code rule, it includes: type + year and month + sequence number; Value arrive Between and including When using the standard rules, including: type + unit + year and month + sequence number; Value greater than or equal to When using the enhanced rules, including: type + confidentiality level + approval code + hash tail number; The specific condition set is: {confidentiality coefficient ≥ 1, contract amount level is greater than medium, and approval method is not exempt from review}; the confidentiality level is quantified as the confidentiality coefficient, with values ​​of 0, 1, and 2, where 0 represents non-confidential, 1 represents secret, and 2 represents top secret; when the contract amount is less than A, the contract amount level is small; when the contract amount is between A and B, the contract amount level is medium; when the contract amount is greater than B, the contract amount level is large; the approval methods are divided into exempt from review, general approval, and multi-level approval.

2. The multi-dimensional attribute fusion contract number recommendation method according to claim 1 is characterized in that: The steps of S1 specifically include: S11. Extract unstructured information from the contract text, including contract type, amount, contract source, language version, attachments, and contract version, through OCR parsing and NLP keyword extraction. S12. Extract unstructured information including approval level, importance, and confidentiality level from the approval process system through API connection to the workflow engine; S13. Convert the extracted unstructured information into multi-dimensional parameters including contract type, confidentiality level, contract amount, importance, approval method, attachment mark, contract version, language version and contract source and generate a parameter table.

3. The multi-dimensional attribute fusion contract number recommendation method according to claim 2 is characterized in that: The steps of S13 specifically include: S131. Convert the extracted contract type, contract amount, contract source, language version, attachment information, and contract version into multi-dimensional parameters including contract type, confidentiality level, contract amount, importance, approval method, attachment tag, contract version, language version, and contract source; S132. Quantify the contract type, confidentiality level, contract amount, importance, approval method, attachment mark, contract version, language version, and contract source to form standardized parameter values; S133. Generate a parameter table based on the multi-dimensional parameters and corresponding parameter values.

4. The multi-dimensional attribute fusion contract number recommendation method according to claim 3 is characterized in that: The steps of S3 include: S31. Determine the decision dimensions that influence the contract number generation node, including confidentiality level, contract amount and importance, approval complexity, and contract version relevance; set thresholds and weights for each dimension. S32. Normalize the parameters of each dimension and calculate the detection item value of each dimension; S33, using the S-type decision function to comprehensively evaluate the detection item values ​​of each dimension and calculate the comprehensive decision value; S34. Determine the generation node of the contract number based on the comprehensive decision value and the preset node rule mapping table.

5. The multi-dimensional attribute fusion contract number recommendation method according to claim 4 is characterized in that: In S31, the decision dimensions and detection items include: In the confidentiality level dimension, the detection item is the ratio of the confidentiality level to the highest level; Contract amount and importance dimension: the test item is (amount coefficient × importance) / (highest amount coefficient × highest importance); Approval complexity dimension: the detection item is the ratio of approval complexity to the highest approval complexity; Contract version correlation dimension, the detection item is contract version correlation.

6. A multi-dimensional attribute fusion contract number recommendation system, characterized by: include: The data acquisition module is used to obtain unstructured contract information, extract and standardize it into multi-dimensional contract attribute information, including contract type, confidentiality level, contract amount, importance, approval method, attachment mark, contract version, language version, and contract source; specifically, it is used to convert the extracted unstructured information into multi-dimensional parameters including contract type, confidentiality level, contract amount, importance, approval method, attachment mark, contract version, language version, and contract source, and generate a parameter table; A rule generation module, configured to generate contract numbering rules based on the multi-dimensional contract attribute information by weight allocation and interactive component calculation, wherein the contract numbering rules include a shortcode rule, a standard rule, and an enhanced rule; A node decision module is used to determine, based on the multi-dimensional contract attribute information, a generation node of the contract number through an S-type decision function calculation, wherein the generation node includes generation at creation, generation after approval, or generation upon completion of signing; An audit execution module is used to manually audit the generated contract numbering rules and generation nodes, and after passing the audit, output the contract number according to the contract numbering rules at the determined generation node; The rule generation module includes: The parameter matrix construction unit is used to construct a parameter matrix according to the parameter table, where each parameter corresponds to a dimension; A weight allocation unit, configured to allocate a weight to each parameter, wherein the weight is determined based on the business impact and security compliance requirements; A basic component calculation unit is used to multiply and sum each parameter with its weight in a linear combination manner to obtain a basic component; An interaction component calculation unit, for calculating the interaction component by multiplying the interaction factor and the multiplication coefficient for contracts that meet a specific set of conditions; The numbering rule determination unit is used to determine the contract numbering rule according to the sum of the basic component and the interactive component, wherein the contract numbering rule includes a simple code rule, a standard rule and an enhanced rule; specifically, the unit is used to add the basic component and the interactive component to obtain the contract numbering rule. value; according to The value range determines the numbering scheme type, including: Value less than When using the short code rule, it includes: type + year and month + sequence number; Value arrive Between and including When using the standard rules, including: type + unit + year and month + sequence number; Value greater than or equal to When using the enhanced rules, including: type + confidentiality level + approval code + hash tail number; The specific condition set is: {confidentiality coefficient ≥ 1, contract amount level is greater than medium, and the approval method is not exempt from review}; the confidentiality level is quantified as the confidentiality coefficient, with values ​​of 0, 1, and 2, where 0 represents non-confidential, 1 represents secret, and 2 represents top secret; when the contract amount is less than A, the contract amount level is small; when the contract amount is between A and B, the contract amount level is medium; when the contract amount is greater than B, the contract amount level is large; the approval methods are divided into exempt from review, general approval, and multi-level approval.

7. An electronic device, characterized in that: The electronic device includes: at least one processor; and a memory communicatively connected to the at least one processor; the memory stores computer program instructions that can be executed by the at least one processor, and the computer program instructions are executed by the at least one processor to enable the at least one processor to execute the contract number recommendation method based on multi-dimensional attribute fusion as described in any one of claims 1 to 5.

8. A non-transitory computer-readable storage medium, characterized in that The non-transitory computer-readable storage medium stores computer instructions, and the computer instructions enable the computer to execute the contract number recommendation method based on multi-dimensional attribute fusion according to any one of claims 1 to 5.

Images