Intelligent identification system and method for applicability of laws and regulations for information technology consulting
By introducing intelligent audit of time difference characteristics and large language models in the legal and regulatory identification system, the problems of the alternation of new and old laws and regulations and the time difference between consultants are solved, the applicability and accuracy of legal and regulatory identification are improved, and dynamic and predictive audits are provided. result.
Patent Information
- Application Number
- CN202410344328.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-03-25
- Publication Date
- 2024-11-26
- Estimated Expiration
- 2044-03-25
AI Technical Summary
The existing legal and regulatory identification system ignores the two time differences between the old and new laws and regulations and the consultant consultation and problem planning to solve, resulting in the inappropriate identification results.
Adopt an intelligent identification system for the applicable legal and regulatory system including consulting module, feature extraction module, policy inspection module, labeling module, intelligent audit module and result module. Legal and regulatory data are collected from two time dimensions through time difference characteristics and designated search characteristics, and the legal and regulatory data is collected from two time dimensions. The large language model is intelligently audited and the cycle process optimization model is optimized through 'intelligent recognition → manual review correction → machine learning → intelligent recognition'.
Improve the applicability and accuracy of the identification of laws and regulations, can dynamically and predictably give reasonable audit results, resolve time conflicts, provide results suitable for consultants' subsequent operations, enhance the universality of the system and deal with sensitive issues. efficiency.
Smart Images

Figure CN118094193B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of information technology consulting, and in particular to an intelligent identification system and method for the applicability of laws and regulations for information technology consulting. Background Art
[0002] With the development of social economy and the promotion of legal knowledge, the number of laws and regulations in my country continues to increase. The massive amount of data makes people spend a lot of energy in searching for legal terms. Therefore, when people encounter problems and need to understand the laws and local regulations related to the problems, it is particularly important to have systems and methods that can help people intelligently, quickly and accurately identify laws and regulations that meet their needs from massive data.
[0003] At the same time, in addition to consultation on routine legal issues, information technology consultation also includes fair competition review. Fair competition review refers to the policy-making body's assessment of whether the proposed administrative regulations, rules, normative documents, other policy documents and specific policy measures in the form of "one case, one discussion" involving the economic activities of business entities exclude or restrict competition, and conducts internal supervision activities for prior review. The evaluation process also requires accurate identification of laws and regulations to achieve effective review.
[0004] Although various information technologies are currently being fully utilized to provide tool support to improve the efficiency and effectiveness of legal and regulatory identification, the problem with existing technologies is that existing identification focuses more on optimizing identification algorithms and models to improve the accuracy of legal and regulatory capture at the time of consultation, while ignoring the fact that my country's laws, in addition to being large in number, are constantly updated, and that old and new laws and regulations are continuously and dynamically alternating, with a time difference in alternating effectiveness and expiration, so that the identification result is only applicable at the time of consultation.
[0005] In order to solve the above-mentioned problem of the timeliness of laws and regulations, the conventional solution is to use the effective time of laws and regulations in the database to define the timeliness of laws and regulations, and then search and identify laws and regulations within the effective range. This method can help identify the update of laws and regulations to a certain extent, but it still ignores the applicability of the time difference between the consultant's consultation and specific use based on the continuously and dynamically updated laws and regulations. Summary of the invention
[0006] The present invention is intended to provide an intelligent identification system and method for the applicability of laws and regulations for information technology consulting, so as to solve the technical problem that the existing legal and regulatory identification ignores the simultaneous existence of the alternation of old and new laws and regulations and the time difference between the consultant's consultation and the problem plan solution, resulting in inapplicable identification.
[0007] The basic solution provided by the present invention is: an intelligent identification system for the applicability of laws and regulations for information technology consultation, including a consultation module, a feature extraction module, a policy inspection module, a labeling module, an intelligent review module and a result module;
[0008] The consulting module is used to input consulting data; the consulting data includes fair competition review documents;
[0009] The feature extraction module is used to extract consultation data features; the consultation data features include time difference features and specified search features; the time difference features include consultation time and planned implementation time;
[0010] The policy inspection module is used to collect legal and regulatory data from two time dimensions through big data collection technology according to time difference characteristics and specified search characteristics, and to build a policy document library;
[0011] The labeling module is used to label the collected legal and regulatory data using the time difference feature, wherein the labeled files include commonly valid policy files and / or individually valid policy files.
[0012] The intelligent audit module is used to call the large language model to conduct comparative audit on the consultation data and the common effective policy documents and / or the individual effective policy documents; the large language model is tuned through a number of fair competition review cases, conclusions and analysis and demonstration, and is continuously self-learned and optimized through intelligent audit;
[0013] The result module is used to output corresponding questionable review results and next operation suggestions based on the time difference characteristics.
[0014] The working principle and advantages of the present invention are as follows: This scheme first uses a large number of fair competition review cases, conclusions, analysis and demonstration, and optimizes the large language model through cases to make the large language model more suitable for fair competition review business, and through the "intelligent recognition → manual review and correction → machine learning → intelligent recognition" cycle process, continuously optimizes the large language model, making the large language model more and more accurate in intelligent review and recognition and giving conclusions. In the actual application of this system, after the consultation data is input, the features of the consultation data are extracted, especially the time difference features and the specified search features. Based on the time difference characteristic of the consultation data in the two stages of consultation and application of laws and regulations, this system collects laws and regulations from two time dimensions and uses the time difference characteristic of the consultation data to mark the collected laws and regulations as jointly valid and individually valid policy documents. The basis for marking is that there is a certain time difference between the draft opinions and the formal implementation of the laws and regulations collected in real time, and there are also laws and regulations that are valid during the consultation time but expire when the problem is planned to be solved. The marking can associate the time difference of the laws and regulations with the time difference of the consultation data, so that the system can perform intelligent review in different time periods, and dynamically and predictively give more reasonable review results in the two time dimensions. At the same time, the system also gives operational suggestions for the application of laws and regulations.
[0015] Unlike the existing technology, this system is not limited to a single suggestion given during the consultation time, and this suggestion is not applicable to the planned solution time of the problem; instead, it is based on the alternation of old and new laws and regulations and the time difference between the consultant's consultation and specific use. It not only solves time conflicts and improves the rationality of policy review in different time dimensions, but also gives results suitable for the consultant's subsequent operations, greatly improving the applicability of this system's legal and regulatory identification to consultants.
[0016] Furthermore, the consulting data also includes legal issues; and the specified search features include region, keyword and type.
[0017] Beneficial effects: Expand the scope of consulting data and improve the versatility of the system; adopt appropriate specified search features to increase the speed of collecting laws and regulations, and control the collection quantity and quality within a reasonable range.
[0018] Furthermore, it also includes a time difference adjustment module for expanding the plan implementation time according to the preset floating time.
[0019] Beneficial effects: The collection scope of policy documents is expanded to an appropriate range, the effective expiration marking is more accurate, and a more reasonable audit result is given by using a dynamic plan implementation time, avoiding the missed favorable policies due to unreasonable plan time set by customers.
[0020] Furthermore, the questionable review results include conflicting applicable clauses, typical cases and constraints; the next step operation suggestions include a form automatically filled out based on the consultation data.
[0021] Beneficial effects: It provides consultants with more comprehensive analysis and demonstration results, but is not limited to reviewing and comparing single results. Instead, it provides actionable suggestions based on the time difference characteristics, which is instructive for consultants and makes the functions of this system extensible.
[0022] Furthermore, it also includes a priority review module, which is used to directly trigger the priority review module for emergency processing when sensitive data is detected in the consulting data.
[0023] Beneficial effects: Improve the efficiency of the system in handling sensitive issues and strengthen legal protection for consultants.
[0024] Furthermore, it also includes an expert review module for conducting at least one expert review of questionable audit results and consulting suggestions; it also includes a consulting guidance module for guiding the consultant to enter the planned implementation time and planned implementation area when the input consulting data does not include the time difference feature and the specified search feature.
[0025] Beneficial effects: Utilizing the manual and system dual audit modes can improve the accuracy and professionalism of the audit results; consulting guidance can improve the dynamics of changes in collected laws and regulations over time, and further improve the applicability of the final results to the consultants in actually solving problems.
[0026] The present invention is based on an intelligent recognition system for the applicability of laws and regulations for information technology consulting, and also provides an intelligent recognition method for the applicability of laws and regulations for information technology consulting, which is used to solve the technical problem that the existing recognition of laws and regulations ignores the simultaneous existence of the alternation of old and new laws and regulations and the time difference between the consultant's consultation and the problem plan solution, resulting in inapplicable recognition.
[0027] The method comprises the following steps:
[0028] S01, input consulting data; the consulting data includes fair competition review documents;
[0029] S02, extracting consulting data features including time difference features and designated search features; the time difference features include consulting time and planned implementation time; the designated search features include region, keyword and type; when the extracted consulting data features do not include the time difference features and the designated search features, starting consulting guidance to guide the consultant to input the planned implementation time and planned implementation region;
[0030] S03, based on the time difference feature and the specified search feature, collect legal and regulatory data from two time dimensions through big data collection technology, and build a policy document library;
[0031] S04, using the time difference feature to mark the collected legal and regulatory data, wherein the marking includes commonly valid policy documents and / or individually valid policy documents;
[0032] S05, calling up a large language model to conduct a comparative review of the consultation data and the common effective policy documents and / or the individual effective policy documents; the large language model is optimized through a number of fair competition review cases, conclusions and analytical arguments, and is continuously self-learned and optimized through intelligent review;
[0033] S06: Output corresponding questionable audit results and next operation suggestions based on the time difference characteristics.
[0034] Beneficial effects: This solution first uses a large number of fair competition review cases, conclusions, analysis and argumentation to tune the large language model through cases, so that the large language model is more suitable for fair competition review business, so that the constructed review ideas are in line with the thinking logic of fair competition review staff, easy to understand, and more accurate; and through the "intelligent identification → manual review and correction → machine learning → intelligent identification" cycle process, the large language model is continuously optimized, so that the large language model is more and more accurate in intelligent review and identification and giving conclusions, thereby providing users with more effective suggestions; in addition, based on the simultaneous existence of the alternation of old and new laws and regulations and the time difference between the consultant's consultation and the application of laws and regulations, a new intelligent identification method for the applicability of laws and regulations is proposed. Taking the time difference as the main line, dynamic legal and regulatory data collection is carried out in different time periods, and a predictive It not only provides predictable audit results but also provides guiding operational suggestions, which solves the time conflict between laws and regulations and consultants. It can provide consultants with the best suggestions at different time stages through a more time-dynamic review of laws and regulations, which not only improves the accuracy of legal and regulatory identification, but also improves the applicability of laws and regulations to consultants, and is more in line with the actual situation with time difference. At the same time, under the premise of limited data volume query, the review results of fair competition review business can be accurately obtained through simple query calculations. The operation is simple, accurate and timely. Every input of consulting questions and output of review results can further enrich fair review cases, conclusions, analysis and argumentation, provide large sample data for big data analysis and system optimization, provide more complete auxiliary work for fair competition review staff, improve review efficiency, and make the entire system more applicable.
[0035] Furthermore, S01 also includes priority review. When sensitive data is detected in the consulting data of the input natural language description, file, picture or video, it is directly linked to the corresponding legal department for emergency processing. The uploaded consulting data and the extracted sensitive data are transmitted to the legal department as evidence to provide data support for subsequent legal channel processing.
[0036] Beneficial effects: Improve the efficiency of the system in handling sensitive issues and strengthen legal protection for consultants.
[0037] Furthermore, S04 also includes time difference adjustment. The preset floating time is a time within a range of 5-10 days before and after the planned implementation time as the midpoint. This time period is used as the planned implementation time for collecting and marking laws and regulations.
[0038] Beneficial effects: The collection scope of policy documents is expanded to an appropriate range, the effective expiration marking is more accurate, and a more reasonable audit result is given by using a dynamic plan implementation time, avoiding the missed favorable policies due to unreasonable plan time set by customers.
[0039] Furthermore, in S06, the questionable review results and the next operation suggestions are visualized, and the questionable content in the consulting data is marked; the questionable review results include conflicting applicable clauses, typical cases and constraints; the next operation suggestions include a form automatically filled in according to the consulting data, and the form is the form that the consultant needs to submit in the laws and regulations corresponding to the planned implementation time of the consulting data.
[0040] Beneficial effects: It provides consultants with more comprehensive analysis and demonstration results, but is not limited to reviewing and comparing single results. Instead, it provides actionable suggestions based on the time difference characteristics, which is instructive for consultants and makes the functions of this system extensible. BRIEF DESCRIPTION OF THE DRAWINGS
[0041] Figure 1 A schematic diagram of the structure of an intelligent identification system for the applicability of laws and regulations for information technology consultation provided by an embodiment of the present invention;
[0042] Figure 2 A flow chart of a method for intelligently identifying the applicability of laws and regulations for information technology consulting provided by an embodiment of the present invention;
[0043] Figure 3 A schematic diagram of consulting data input for an intelligent identification system for applicability of laws and regulations for information technology consulting provided by an embodiment of the present invention;
[0044] Figure 4 A schematic diagram of policy inspection results of the intelligent identification system for applicability of laws and regulations for information technology consultation provided by an embodiment of the present invention;
[0045] Figure 5 A schematic diagram of intelligent review results of an intelligent identification system for applicability of laws and regulations for information technology consultation provided by an embodiment of the present invention;
[0046] Figure 6A schematic diagram of an intelligent audit result form of the intelligent identification system for the applicability of laws and regulations for information technology consulting provided in an embodiment of the present invention. DETAILED DESCRIPTION
[0047] The following is a further detailed description through specific implementation methods:
[0048] The embodiment is basically as shown in the attached Figure 1 As shown: An intelligent identification system for the applicability of laws and regulations for information technology consulting, including a consulting module, a feature extraction module, a policy inspection module, a labeling module, an intelligent review module and a result module; all modules are implemented on a cloud server connected to the network.
[0049] The consulting module is used to input consulting data; the consulting data includes fair competition review documents.
[0050] The feature extraction module is used to extract consultation data features; the consultation data features include time difference features and specified search features; the time difference features include consultation time and plan implementation time.
[0051] It also includes a consultation guidance module, which is used to guide the consultant to enter the planned implementation time and planned implementation area when the input consultation data does not include the time difference feature and the specified search feature.
[0052] The policy inspection module is used to collect legal and regulatory data from two time dimensions based on time difference characteristics and specified search characteristics through big data collection technology, and to build a policy document library.
[0053] The labeling module is used to label the collected legal and regulatory data using the time difference feature, wherein the labeled files include commonly valid policy files and / or individually valid policy files.
[0054] It also includes a time difference adjustment module, which is used to expand the plan implementation time according to the preset floating time.
[0055] The intelligent audit module is used to call the large language model to compare and audit the consulting data and common valid policy documents and / or individual valid policy documents; the large language model is tuned through a number of fair competition review cases, conclusions and analytical arguments, and is continuously self-learned and optimized through intelligent audit.
[0056] The result module is used to output corresponding questionable review results and next operation suggestions based on the time difference characteristics.
[0057] It also includes a priority review module, which is used to directly trigger the priority review module for emergency processing when sensitive data is detected in the consulting data.
[0058] It also includes an expert review module, which is used to conduct at least one expert review of questionable audit results and consulting recommendations.
[0059] like Figure 2 As shown, based on the intelligent identification system for the applicability of laws and regulations for information technology consulting in this embodiment, an intelligent identification method for the applicability of laws and regulations for information technology consulting is also proposed to realize the use process of the above system or the corresponding functions of each module; the method includes the following steps:
[0060] S01, input consulting data; the consulting data includes fair competition review documents and legal issues;
[0061] Specifically, the input fair competition review data includes the proposed policy documents, including information such as the time, region, and type of implementation; legal issues include conventional natural language descriptions of the issues, which may involve legal issues related to family, education, and assets, such as Figure 3 As shown, you can upload relevant files directly in the system.
[0062] For example, the legal issue consulted by the consultant involves the division of property in a divorce. The time when the issue is entered into the system and the region extracted by the system IP are the consultation time and region. The consultation also includes entering the time and location requirements for the planned resolution of the issue, that is, when and where the actual processing is planned after the consultation. What is taken into consideration here is that when the consultant actually handles legal issues, time difference and geographical difference are very critical factors in the collection of the applicability of laws and regulations, which will directly determine the applicability of the final result to the consultant when actually solving the problem.
[0063] S01 also includes priority review. When sensitive data is detected in the consulting data, the priority review module is directly triggered for emergency processing. Specifically, the input consulting data can also upload files, pictures, videos, etc. Sensitive data may involve hot social legal issues, such as domestic violence, abuse and corruption. The legal evidence data collected from the uploaded descriptions, files, pictures or videos can be directly linked to the relevant legal department for emergency processing. At the same time, the sensitive data is saved as evidence to the legal department to provide data support for subsequent legal channel processing.
[0064] S02, extracting consulting data features including time difference features and designated search features; the time difference features include consulting time and planned implementation time; the designated search features include regions, keywords and types; when the extracted consulting data features do not include time difference features and designated search features, starting consulting guidance to guide the consultant to input the planned implementation time and planned implementation region.
[0065] Specifically, because for consultation on legal issues, there may be a situation where consultation is conducted before implementation, and fair competition review data may be published first and then implemented, the time difference here will lead to different application of laws and regulations. This solution is based on this time difference feature and can give multi-time dimension results that are different from the existing technology, providing consultants with more comprehensive, systematic and reasonable suggestions; the consultation guidance method can supplement the planned implementation time and region to improve the accuracy of the collection of laws and regulations, as well as the applicability of the review results to the consultant when actually solving the problem; the specified search features include region, keywords and type, such as Yubei District, Chongqing City.
[0066] S03, based on the time difference feature and the specified search feature, collect legal and regulatory data from two time dimensions through big data collection technology, and build a policy document library;
[0067] Specifically, crawler technology can be used for big data collection technology. The program uses network protocols to imitate human browsing behavior, access various resources on the Internet, obtain information and analyze it. While collecting data based on regions, keywords and types, corresponding crawling is carried out with the consultation time and region, and the plan implementation time and region. Only laws and regulations that are valid in the two time dimensions relative to the consultant are collected to improve the pertinence of data collection; Figure 4 As shown, the policy inspection results are displayed in the system.
[0068] S04, using the time difference feature to mark the collected legal and regulatory data, wherein the marking includes commonly valid policy documents and / or individually valid policy documents;
[0069] Specifically, a jointly valid policy document refers to a policy document that has been implemented and is valid at the consultation time and the problem plan resolution / implementation time; a separately valid policy document for the consultation time refers to a policy document that has been implemented but is valid at the consultation time and is about to expire at the problem plan resolution / implementation time; a separately valid policy document for the problem plan resolution / implementation time means that the draft opinion / public draft is invalid at the consultation time. Because policy documents also have a time difference, the time difference of the policy document and the time difference between the consultant's consultation and implementation are associated through marking, which can make the subsequent review results and suggestions more in line with the two time dimensions, making it easier for the consultant to comprehensively consider and solve the problem.
[0070] This also includes time difference adjustment, expanding the planned implementation time according to the preset floating time. The preset floating time can allow the planned implementation time to float 5-10 days before and after the midpoint. This increases the collection scope of policy documents to an appropriate range, makes effective expiration marking more accurate, and uses dynamic planned implementation time to give more reasonable audit results, avoiding unreasonable planned time set by customers and missing out on favorable policies.
[0071] S05, call up a large language model to conduct a comparative intelligent review of the consulting data and common valid policy documents and / or individual valid policy documents; the large language model is optimized through a number of fair competition review cases, conclusions and analytical arguments, and is continuously self-learned and optimized through intelligent review.
[0072] Specifically, this plan first uses a large number of fair competition review cases, conclusions, and analysis and demonstration to tune the large language model through cases, so that the large language model is more suitable for fair competition review business, and through the "intelligent identification → manual review correction → machine learning → intelligent identification" cycle process, continuously optimize the large language model, so that the large language model is more and more accurate in intelligent review recognition and conclusion. Model tuning is to optimize the performance of the model to the best by changing hyperparameters, adjusting optimizers, data preprocessing, model structure, etc., so as to achieve the best performance and minimum error. Good tuning can not only improve the accuracy of the model, but also make the model more stable, reliable, and interpretable, thereby improving the practical value of the model. Large language model (LLM (large language model)), that is, the number of parameters is usually in billions of weights or more language models, such as Llama2, chagpt, chatlaw, etc.; the open source large model architecture Llama2 can be selected; Llama2 is the next generation of large open source language models developed by Meta AI, including pre-trained and fine-tuned large language models, with model parameters ranging from 7 billion to 70 billion. The pre-trained model is trained on 2 trillion tokens, with twice the context length of Llama 1. The fine-tuned model is trained on more than 1 million human annotations. Currently, Llama2 has become the benchmark for text generation models. Most models use it as a benchmark for comparison. Many large open source models in China are fine-tuned and trained on it.
[0073] S05 also includes large language model training. The source data used for training includes accumulated business data, collected legal data, and legal and regulatory data. The model architecture used for training adopts the open source large model architecture Llama2, and uses open source model parameters to fine-tune the model using prompt learning and instruction learning in the legal field to obtain a vertical model for review.
[0074] Use its open source model parameters to fine-tune the model using prompt learning and instruction learning in the legal field to obtain a vertical model; the specific implementation method is as follows:
[0075] 1) Process the data to keep the input format consistent, for example:
[0076] enter:
[0077] Find out whether the following sentences are biased towards specific companies or groups of people. The superior law basis: Violation of Article 15 of the "Implementation Rules of the Fair Competition Review System" "Affecting the production and operation cost standards", Article (1) "No illegal preferential policies for specific operators" in the third situation (if there is no superior law basis, it is constructed as follows: Superior law basis: None) Content: Guide each high school to combine the actual situation of the school, use information technology, strive to explore classroom teaching methods with future education characteristics, and actively participate in classroom teaching reform. Complete the relevant work of high school comprehensive quality evaluation on time and with high quality; encourage schools to actively participate in the use of moors platform to do a good job in research-based learning special reports.
[0078] Output or label:
[0079] Biased crowd: "Encourage schools to actively participate in using the moors platform", directly points out the names of specific companies and points to specific operators.
[0080] Is there any bias: Yes
[0081] If there is no bias, the following example is output:
[0082] Preferred group: None; Is there a preference: No
[0083] 2) Loading model and model training
[0084] After loading the model, the model is trained using Lora fine-tuning (which freezes the pre-trained model weights and injects some trainable model blocks into each layer of the Transformer architecture to train these model blocks instead of training the entire model) to enable the model to understand legal knowledge.
[0085] 3) Prediction
[0086] First, use text similarity to search for the most similar superior law in the superior law policy database. If the similarity is greater than 95%, it is considered that there is a superior legal basis, otherwise it is considered that there is no superior legal basis. Then, the input is constructed based on the content to be judged and the superior legal basis, and the model determines whether there is bias.
[0087] S06: Output corresponding questionable audit results and next operation suggestions based on the time difference characteristics.
[0088] Specifically, the questionable audit results and the next step operation suggestions are displayed visually, such as Figure 5As shown, the questionable content in the consulting data is marked; the questionable review results include conflicting applicable clauses, typical cases and constraints, among which the conflicting applicable clauses include the legal clauses applicable to the consulting data in the collected laws and regulations and the fair competition clauses violated; typical cases include cases under review and cases issued by the State Administration for Market Regulation; constraints include higher-level policies, which include publicly released policies and regulations, in order to provide a more comprehensive analysis and argumentation on the consulting data, such as Figure 6 The result form shown presents the results in a clearer way. When a file is recognized, the recognition results are presented in a hierarchical order and then the time weight is added. The files with earlier time are associated first, so that the recognition results are logically hierarchical.
[0089] The next operation suggestion includes a form that is automatically filled in according to the consulting data, wherein the form is the form that the consultant needs to submit in the laws and regulations corresponding to the planned implementation time. There is a time period between the consultant's consultation and the specific implementation of the plan. In this process, the applicable laws and regulations may be different, and the corresponding registration form may also change. Therefore, when giving consulting suggestions, the forms related to the applicable laws and regulations at the time of planned implementation are automatically filled in using the consulting data, to help and guide the consultant's subsequent operations, improve the applicability of the system to the consultant, and enable more effective problem solving.
[0090] It also includes S07, expert review, which involves at least one expert review of questionable audit results and consulting suggestions, using a dual review mode of manual and system to improve the accuracy and professionalism of the audit results, which can be displayed in the system through task status.
[0091] The intelligent identification system and method for the applicability of laws and regulations for information technology consultation provided in this embodiment extracts time difference characteristics and specified search characteristics based on consultation data, and uses these characteristics to collect and mark corresponding laws and regulations in two time dimensions, and then reviews and gives suggestions for the next step of operation. Based on the time difference between the consultant's consultation and the planned solution of the problem and the time difference between the new and old laws and regulations, the results of the system and method are more predictive, and the time of the consultant's consultation and the actual planned solution of the problem are distinguished to give a more instructive next step of actionable results, which not only improves the accuracy of legal and regulatory identification, but also improves the applicability of laws and regulations to consultants, and is more in line with the actual situation with time difference; at the same time, through a large number of fair competition review cases, conclusions, analysis and demonstration, the large language model is tuned through cases to make the large language model more suitable for fair competition review business, and through the "intelligent identification → manual review and correction → machine learning → intelligent identification" cycle process, the large language model is continuously optimized, so that the large language model is more and more accurate in intelligent review and identification and giving conclusions, and more and more in line with the legal field, and has strong promotion and application.
[0092] The above is only an embodiment of the present invention. The common sense such as the known specific structure and characteristics in the scheme is not described in detail here. The ordinary technicians in the relevant field are aware of all the common technical knowledge in the technical field of the invention before the application date or priority date, can obtain all the existing technologies in the field, and have the ability to apply the conventional experimental means before that date. The ordinary technicians in the relevant field can improve and implement the scheme in combination with their own abilities under the enlightenment given by this application. Some typical known structures or known methods should not become obstacles for the ordinary technicians in the relevant field to implement this application. It should be pointed out that for the technicians in this field, without departing from the structure of the present invention, several deformations and improvements can be made, which should also be regarded as the protection scope of the present invention, and these will not affect the effect of the implementation of the present invention and the practicality of the patent.
Claims
1. An intelligent identification system for the applicability of laws and regulations for information technology consulting, characterized by: It includes consulting module, feature extraction module, policy inspection module, annotation module, intelligent review module and result module; The consulting module is used to input consulting data; the consulting data includes fair competition review documents; The feature extraction module is used to extract consulting data features; The consulting data features include specified search features; The specified search feature includes a time difference feature; the time difference feature includes consultation time and planned implementation time; The policy inspection module is used to collect legal and regulatory data from two time dimensions based on specified search features through big data collection technology, and to build a policy document library; The labeling module is used to label the collected legal and regulatory data using the time difference feature, wherein the labeled files include commonly valid policy files and / or individually valid policy files; The intelligent audit module is used to call the large language model to conduct comparative audit on the consultation data and the common effective policy documents and / or the individual effective policy documents; the large language model is tuned through a number of fair competition review cases, conclusions and analysis and demonstration, and is continuously self-learned and optimized through intelligent audit; The result module is used to output corresponding questionable audit results and next operation suggestions based on the time difference feature; Among them, the commonly effective policy documents refer to the policy documents that have been implemented and are valid at the time of consultation and the time when the problem is planned to be solved and implemented; Separately valid policy documents include separately valid policy documents for the consultation period, and separately valid policy documents for the plan to resolve and implement the issues. Separately valid policy documents for the consultation period indicate policy documents that have been implemented but are valid during the consultation period and are about to expire when the plan to resolve and implement the issues becomes invalid; separately valid policy documents for the plan to resolve and implement the issues indicate that draft opinions and public drafts are invalid during the consultation period.
2. The intelligent identification system for applicability of laws and regulations for information technology consultation according to claim 1 is characterized in that: The consulting data also includes legal issues; the specified search characteristics also include regions, keywords and types.
3. The intelligent identification system for applicability of laws and regulations for information technology consultation according to claim 1 is characterized in that: It also includes a time difference adjustment module, which is used to expand the plan implementation time according to the preset floating time.
4. The intelligent identification system for applicability of laws and regulations for information technology consultation according to claim 1 is characterized in that: The questionable review results include conflicting applicable clauses, typical cases and constraints; the next step operation suggestions include a form that is automatically filled out based on the consultation data.
5. The intelligent identification system for applicability of laws and regulations for information technology consultation according to claim 1 is characterized in that: It also includes a priority review module, which is used to directly trigger the priority review module for emergency processing when sensitive data is detected in the consulting data.
6. The intelligent identification system for applicability of laws and regulations for information technology consultation according to claim 1 is characterized in that: It also includes an expert review module for conducting at least one expert review of questionable review results and consulting suggestions; it also includes a consulting guidance module for guiding the consultant to enter the planned implementation time and planned implementation area when the input consulting data does not include the specified search features.
7. A method for intelligently identifying the applicability of laws and regulations for information technology consulting, characterized in that: The method comprises the following steps: S01, input consulting data; the consulting data includes fair competition review documents and legal issues; S02, extracting consulting data features including designated search features, wherein the designated search features include time difference features; the time difference features include consulting time and planned implementation time; the designated search features also include region, keyword and type; when the extracted consulting data features do not include the designated search features, starting consulting guidance to guide the consultant to input the planned implementation time and planned implementation region; S03, based on the specified search features, collect legal and regulatory data from two time dimensions through big data collection technology and build a policy document library; S04, using the time difference feature to mark the collected legal and regulatory data, wherein the marking includes commonly valid policy documents and / or individually valid policy documents; S05, calling up a large language model to conduct a comparative review of the consultation data and the common effective policy documents and / or the individual effective policy documents; the large language model is optimized through a number of fair competition review cases, conclusions and analytical arguments, and is continuously self-learned and optimized through intelligent review; S06: Output the corresponding questionable audit results and next operation suggestions based on the time difference characteristics; Among them, jointly valid policy documents refer to policy documents that have been implemented and are valid during the consultation period and the time when the problem plan is resolved and implemented; individually valid policy documents include individually valid policy documents for the consultation period, and individually valid policy documents for the time when the problem plan is resolved and implemented; among them, individually valid policy documents for the consultation period refer to policy documents that have been implemented but are valid during the consultation period and are about to expire when the problem plan is resolved and implemented; individually valid policy documents for the time when the problem plan is resolved and implemented indicate that draft opinions and public drafts are invalid during the consultation period.
8. The intelligent identification method for the applicability of laws and regulations for information technology consulting according to claim 7 is characterized in that: S01 also includes priority review. When sensitive data is detected in the consulting data of the input natural language description, file, picture or video, it will be directly linked to the corresponding legal department for emergency processing. The uploaded consulting data and extracted sensitive data will be transmitted to the legal department as evidence to provide data support for subsequent legal channel processing.
9. The intelligent identification method for the applicability of laws and regulations for information technology consulting according to claim 7 is characterized in that: S04 also includes time difference adjustment. The preset floating time is 5-10 days before and after the midpoint of the planned implementation time. The time within this range is used as the planned implementation time for collecting and marking laws and regulations.
10. The intelligent identification method for the applicability of laws and regulations for information technology consultation according to claim 7 is characterized in that: In S06, the questionable review results and the next operation suggestions are visualized, and the questionable content in the consulting data is marked; the questionable review results include conflicting applicable clauses, typical cases and constraints; the next operation suggestions include a form automatically filled in according to the consulting data, and the form is the form that the consultant needs to submit in the laws and regulations corresponding to the planned implementation time of the consulting data.
Citation Information
Cited By
Conflict-aware legal case judgment prediction method and system
CN121235859A
