Legal document and project management field association analysis method based on NLP

Abstract

The invention provides a legal instrument and project management field association analysis method based on NLP, and belongs to the technical field of project management. The method comprises the following steps: performing dispute focus extraction on a legal document through a large language model to obtain a dispute focus table, and mapping keywords in the dispute focus table to a project management knowledge field to obtain a case dispute reason mapping table containing a first project management field; then calculating the frequency between the first project management field and a dispute focus, the cosine distance and the weight of mutual information by using an entropy weight method, carrying out weight sorting, screening out a second project management field causing dispute, calculating the Pearson correlation between sub-fields in the first project management field, obtaining the correlation of project management sub-fields, and carrying out the calculation of the correlation of the sub-fields in the second project management field. And screening sub-fields strongly related to the second project management field according to the project management sub-field correlation. According to the invention, the reason for the project dispute can be analyzed from various related texts in the project management field, and the subjectivity of project dispute analysis is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of project management, and in particular to a method for analyzing the association between legal documents and project management based on NLP. Background Art

[0002] Construction projects encompass a wide range of sectors, including infrastructure development, real estate development, and municipal engineering. According to the Global Construction Project Dispute Report, the average value of global construction project disputes and the time it takes to resolve them have increased significantly in recent years. Identifying the root causes of disputes is the first step in resolving them, rather than attempting to resolve them after they arise.

[0003] At present, when analyzing the causes of project disputes, we usually locate the causes through internal project documents, expert meetings, or interviews. These methods have achieved relatively good results, but they also have the following problems: 1. Internal project data can only reveal information about that project. However, the unique nature of projects means that different projects have vastly different characteristics, leading to varying causes of disputes. Consequently, results from one project are difficult to apply to other projects, and analytical methods are less universal.

[0004] 2. The interview results of project disputes mainly rely on manual or expert analysis, and the causes of project disputes are relatively subjective. Summary of the Invention

[0005] In view of this, the present invention provides an NLP-based method for analyzing the association between legal documents and project management fields, so as to solve the problem that the current project dispute cause analysis technology is too subjective and has poor versatility.

[0006] The technical solution adopted in the present invention is: The present invention provides a method for analyzing the association between legal documents and project management based on NLP, including: Obtain legal documents and input them into a large language model to extract dispute points and obtain a dispute point table; Mapping the cause keywords in the dispute focus table to the project management knowledge domain to obtain a case dispute cause mapping table including the first project management domain; Count the number of times the sub-fields of the first project management field appear in the dispute focus table to obtain the frequency of occurrence of each sub-field in the first project management field; A joint vocabulary is constructed based on the dispute text in the dispute focus table and the case dispute cause mapping table, and the mutual information between the dispute text and the first project management field is calculated based on the joint vocabulary; Convert the dispute text in the dispute focus table into a dispute vector, and convert the first item management field in the case dispute cause mapping table into a management field vector, and calculate the cosine distance between each dispute vector and each management field vector; The entropy weight method is used to calculate the weight importance values ​​of frequency, cosine distance and mutual information, and the weight importance is sorted, and the second project management field is screened out from the first project management field according to the preset screening threshold; The cause keywords corresponding to the sub-fields in the first project management field are mapped into a group of project field vectors, and the Pearson correlation between each two groups of project field vectors is calculated to obtain the project management sub-field correlation. At the same time, based on the project management sub-field correlation, sub-fields that are strongly correlated with the second project management field are screened out from the first project management field.

[0007] Furthermore, the legal documents are obtained and input into the large language model to extract the focus of disputes, thereby obtaining a focus of disputes table, including: Obtain legal documents related to project management and input them into the big language model. The big language model is then used to extract the case number, plaintiff's claim text, and defendant's response text from each legal document to form the focus of the dispute. The TF-IDF algorithm is used to extract cause keywords from dispute focuses, and the cause keywords of multiple dispute focuses are combined to construct a dispute focus table for project management.

[0008] Furthermore, the cause keywords in the dispute focus table are mapped to the project management knowledge field to obtain a case dispute cause mapping table containing the first project management field, including: Use the Jieba word segmentation tool to segment the cause keywords in the dispute focus table, remove stop words, and obtain a backup keyword table; Create a stop word list WL based on the expert opinions of project management and the field of invention; Based on the stop word list, the TF-IDF algorithm is used to perform secondary word segmentation on the backup keyword table to obtain the cause keyword table; Extract the sub-domains of the project management process and the corresponding large project management knowledge domains, and use the sub-domains and large project management knowledge domains as the first project management domain; A semantic comparison is performed between the cause keywords in the cause keyword table and the text of the first project management field, and the cause keywords are classified into the first project management field with similar semantics to the cause keywords to form a case dispute cause mapping table; each first project management field has a project field identifier.

[0009] Furthermore, the counting of the number of times the sub-fields of the first project management field appear in the dispute focus table to obtain the frequency of occurrence of each sub-field in the first project management field includes: A dispute label is set for each dispute focus in the dispute focus table according to the project field identifier of the first project management field, and each dispute label represents the project management field that caused the dispute; The number of times each first project management field appears in the dispute focus table is counted according to the dispute label to obtain the frequency of occurrence of each first project management field.

[0010] Furthermore, the method of constructing a joint vocabulary based on the dispute text in the dispute focus table and the case dispute cause mapping table, and calculating the mutual information between the dispute text and the first project management field based on the joint vocabulary, includes: Using the Lawformer pre-trained model, we construct a joint vocabulary for each dispute text in the dispute focus table and each first project management field in the case-dispute cause mapping table. Counting the first and second frequencies of each cause keyword in the joint vocabulary in the dispute text and the first project management field, and counting the third frequency of each cause keyword appearing in both the dispute text and the first project management field; According to the first frequency, second frequency and third frequency of each reason keyword in the joint vocabulary, the marginal probability and joint probability of the dispute text and the first project management field are calculated to obtain the mutual information of multiple sub-fields in each first project management field, and the mutual information of multiple sub-fields is summed and averaged to obtain the average mutual information of each first project management field.

[0011] Furthermore, the process of converting the dispute text in the dispute focus table into a dispute vector, converting the first project management field in the case dispute cause mapping table into a management field vector, and calculating the cosine distance between each dispute vector and each management field vector includes: Use the Lawformer pre-trained model to map each dispute focus in the dispute focus table into a dispute vector. Also use the Lawformer pre-trained model to map each sub-field of the first project management field in the case dispute cause mapping table into a sub-field management field vector. Use the Lawformer pre-trained model to calculate the vector dot product between each dispute vector and each sub-domain management domain vector, as well as the vector length of each dispute vector and the vector length of each sub-domain management domain vector; Calculate the cosine similarity of each dispute vector and each sub-domain management domain vector based on the vector length of each dispute vector, the vector length of each sub-domain management domain vector, and the corresponding vector dot product, and subtract the cosine similarity from 1 to obtain the cosine distance of each sub-domain; The cosine distances of all sub-fields of the first project management field are summed and averaged to obtain the average cosine distance of the first project management field.

[0012] Furthermore, the entropy weight method is used to calculate the weight importance values ​​of frequency, cosine distance, and mutual information, and to sort the weight importance, and the second project management field is screened out from the first project management field according to a preset screening threshold, including: Obtaining frequency index values, cosine distance index values, and mutual information index values ​​for each sub-field in the first project management field to form an index value matrix; Based on the index value matrix, the frequency weight, cosine distance weight and mutual information weight are calculated using the entropy weight method; Multiply the frequency weight by the frequency index value, the cosine distance weight by the cosine distance index value, and the mutual information weight by the mutual information index value, and add the three product results to obtain the comprehensive importance value of each specific field; The comprehensive importance values ​​of all project management fields in the first project management field are sorted in descending order, and according to a preset screening threshold, the sub-fields whose comprehensive importance values ​​exceed the preset screening threshold are screened out from the first project management field as the second project management field.

[0013] Furthermore, the cause keywords corresponding to the sub-fields in the first project management field are mapped into a group of project field vectors, and the Pearson correlation between each two groups of project field vectors is calculated to obtain the project management sub-field correlation. At the same time, based on the project management sub-field correlation, sub-fields that are strongly correlated with the second project management field are screened from the first project management field, including: The Lawformer pre-trained model is used to map the cause keywords of the sub-fields in the first project management field in the case dispute cause mapping table into a set of project field vectors; The Pearson correlation between the project domain vectors of every two sub-domains is calculated, and the correlation of the project management sub-domains is determined based on the Pearson correlation calculation results. The sub-domains that are strongly correlated with the second project management domain are screened out from the first project management domain based on the correlation of the project management sub-domains.

[0014] In summary, the beneficial effects of the present invention are as follows: The present invention provides a method for analyzing the association between legal documents and project management fields based on NLP. The method obtains legal documents and inputs them into a large language model to extract dispute points, thereby obtaining a dispute point table. The method then maps the cause keywords in the dispute point table to project management knowledge fields, resulting in a case dispute cause mapping table containing the first project management field. The method then uses an entropy weight method to calculate the weights of the frequency, cosine distance, and mutual information between the first project management field and the dispute points, and sorts the weights. The second project management field that caused the dispute is screened from the first project management field according to a preset screening threshold. Finally, the cause keywords corresponding to the sub-fields in the first project management field are mapped into a set of project field vectors, and the Pearson correlation between each pair of project field vectors is calculated to obtain the project management sub-field correlation. Based on the project management sub-field correlation, the method screens out sub-fields from the first project management field that are strongly correlated with the second project management field, thereby obtaining the second project management field that requires special attention and its strongly correlated sub-fields. By using objective legal documents and project management fields for association analysis, the causes of project disputes can be analyzed from texts related to multiple project management fields, reducing the subjectivity of project dispute analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following is a brief introduction to the drawings required for use in the embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work, and these are all within the scope of protection of the present invention.

[0016] Figure 1 This is a flow chart of a method for analyzing the association between legal documents and project management based on NLP of the present invention; Figure 2 This is a schematic diagram of the legal dispute-project management knowledge domain correlation analysis framework of the present invention; Figure 3 It is the keyword table extraction flow chart of the present invention; Figure 4 This is a flowchart of the field mutual information calculation of the present invention; Figure 5 Flowchart of cosine distance calculation for the field of the present invention; Figure 6 This is a flow chart of the entropy weight method calculation of the present invention. DETAILED DESCRIPTION

[0017] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. It should be noted that, in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. If there is no conflict, the various features of the present invention and the embodiments can be combined with each other and are all within the scope of protection of the present invention.

[0018] The detailed implementation process of the present invention is shown in the following examples.

[0019] Example 1: Reference Figure 1 and Figure 2 As shown, Figure 1 This is a flowchart of the NLP-based method for analyzing the association between legal documents and project management. Figure 2 2 is a schematic diagram of a framework for analyzing the association between legal disputes and project management knowledge domains. The method of this embodiment includes: S1: Obtain legal documents and input them into the large language model to extract dispute points and obtain a dispute point table; S2: Map the cause keywords in the dispute focus table to the project management knowledge domain to obtain a case dispute cause mapping table containing the first project management domain; S3: Count the number of times the sub-fields of the first project management field appear in the dispute focus table to obtain the frequency of occurrence of each sub-field in the first project management field; S4: constructing a joint vocabulary based on the dispute text in the dispute focus table and the case dispute cause mapping table, and calculating the mutual information between the dispute text and the first project management field based on the joint vocabulary; S5: converting the dispute text in the dispute focus table into a dispute vector, and converting the first project management field in the case dispute cause mapping table into a management field vector, and calculating the cosine distance between each dispute vector and each management field vector; S6: Calculate the weighted importance values ​​of frequency, cosine distance, and mutual information using the entropy weight method, and perform weighted importance ranking, and select the second project management field from the first project management field according to a preset screening threshold; S7: Map the cause keywords corresponding to the sub-fields in the first project management field into a group of project field vectors, and calculate the Pearson correlation between each two groups of project field vectors to obtain the project management sub-field correlation. At the same time, based on the project management sub-field correlation, screen out the sub-fields that are strongly correlated with the second project management field from the first project management field.

[0020] This embodiment obtains legal documents and inputs them into a large language model to extract dispute points, thereby obtaining a dispute point table. The cause keywords in the dispute point table are then mapped to project management knowledge domains, resulting in a case dispute cause mapping table containing the first project management domain. The entropy weight method is then used to calculate the weights of the frequency, cosine distance, and mutual information between the first project management domain and the dispute point, and the weights are sorted. The second project management domain that caused the dispute is screened out from the first project management domain according to a preset screening threshold. Finally, the cause keywords corresponding to the sub-domains in the first project management domain are mapped into a set of project domain vectors, and the Pearson correlation between each two sets of project domain vectors is calculated to obtain the project management sub-domain correlation. Based on the project management sub-domain correlation, sub-domains that are strongly correlated with the second project management domain are screened out from the first project management domain. By using objective legal documents and project management domains for correlation analysis, the causes of project disputes can be analyzed from texts related to multiple project management domains, reducing the subjectivity of project dispute analysis.

[0021] Specifically, this embodiment describes the detailed implementation process of each step of the above method, as follows: In this embodiment, step S1 specifically includes the following sub-steps: S11: Obtain legal documents related to project management and input them into the big language model. The big language model is used to extract the case number and the plaintiff’s claim text (e.g. ) and the defendant's reply text (such as ) forms the focus of the dispute. Legal documents refer to legal documents such as legal rulings related to project management disputes.

[0022] S12: Use the TF-IDF algorithm to extract the cause keywords from the focus of the dispute and obtain the corresponding cause keyword table KL : ; in, Representative n Reason keywords, Indicates the n The weight of the reason keyword.

[0023] Finally, combine the keyword list of the causes of multiple controversial issues KL , get the controversial focus table of project management FL : ; Among them, each award corresponds to a row in the dispute focus table.

[0024] In this embodiment, refer to Figure 3In the keyword table extraction process shown in FIG, step S2 specifically includes the following sub-steps: S21: Use the Jieba word segmentation tool to segment the cause keywords in the dispute focus table, remove stop words, and obtain a backup keyword table; S22: Create a stop word list based on project management expertise and the invention field WL : ; in, Indicates the nth stop word.

[0025] S23: Based on the stop word list, the backup keyword table is segmented twice using the TF-IDF algorithm to obtain a cause keyword table. The specific calculation formula for the secondary segmentation using the TF-IDF algorithm is as follows: ; ; Among them, t represents the target word, d represents the current document, and D represents the document collection. represents the number of times the target word t appears in document d, represents the total number of words in the document set D, TF represents the term frequency, IDF(t,D) represents the inverse document frequency, which reflects the general importance of word t in the entire document set D, TF(t,d) represents the frequency of the target word t in document d, and TF-IDF(t,d,D) is the term frequency-inverse document frequency, which is used to quantify the importance of a word. In the documentation The importance of .

[0026] ; Where N represents the total number of documents in the document set D. Represents the number of documents containing the target word t.

[0027] S24: Extract sub-domains and corresponding large project management knowledge domains in the project management process, and use the sub-domains and large project management knowledge domain as the first project management domain. There are 49 sub-domains and 10 large project management knowledge domains.

[0028] S25: Perform semantic comparison between the cause keywords in the cause keyword table and the text of the first project management domain, and classify the cause keywords into the first project management domain with similar semantics to the cause keywords, thereby forming a case dispute cause mapping table. Each first project management domain has a project domain identifier. The case dispute cause mapping table is specifically represented as follows: ; in, Indicates the nth sequence number.

[0029] In this embodiment, step S3 specifically includes the following sub-steps: S31: Setting a dispute tag for each dispute focus in the dispute focus table according to the project field identifier of the first project management field, where each dispute tag represents the project management field that caused the dispute. FL_Label It can be expressed as: ; in, The label representing the nth dispute.

[0030] S32: Count the number of times each sub-field in each first project management field appears in the dispute focus table based on the dispute label to obtain the frequency of each sub-field in each first project management field. The corresponding frequency calculation formula is: ; Where N represents the number of documents, L represents the Lth label, Indicates that there is a d-th document with label K.

[0031] In this embodiment, refer to Figure 4 In the mutual information calculation process shown in FIG, step S4 specifically includes the following sub-steps: S41: Using the Lawformer pre-trained model, a joint vocabulary is constructed for each dispute text in the dispute focus table and each first project management field in the case-dispute cause mapping table; S42: Counting the first and second frequencies of each cause keyword in the joint vocabulary appearing in the dispute text and the first project management field, and counting the third frequency of each cause keyword appearing in both the dispute text and the first project management field; S43: Based on the first frequency, second frequency, and third frequency of each cause keyword in the joint vocabulary, calculate the marginal probability and joint probability of the dispute text and the first project management field, obtain the mutual information of multiple sub-fields in each first project management field, and sum and average the mutual information of multiple sub-fields to obtain the average mutual information of each first project management field. The mutual information calculation formula is as follows: ; ; in, x , y are two discrete random variables. yes , The joint probability of . , yes , The marginal probability of . Usually the natural logarithm or base 2 is used. is the jth mutual information value of the i-th sub-field, and n is the number of documents. represents the mutual information of sub-domains. represents the average mutual information of the first project management domain. r represents the number of subdomains, and c represents the number of mutual information values.

[0032] In this embodiment, refer to Figure 5 In the community cosine distance calculation process shown in FIG, step S5 specifically includes the following sub-steps: S51: Use the Lawformer pre-training model to map each dispute focus in the dispute focus table into a dispute vector. At the same time, use the Lawformer pre-training model to map each sub-field of the first project management field in the case dispute cause mapping table into a sub-field management field vector, so as to obtain the text vector of each dispute and each management field.

[0033] Among them, the Lawformer pre-training model is a legal pre-training language model for long legal texts.

[0034] S52: Use the Lawformer pre-trained model to calculate the vector dot product of each dispute vector and each sub-domain management domain vector, as well as the vector length of each dispute vector and the vector length of each sub-domain management domain vector.

[0035] S53: Calculate the cosine similarity between each dispute vector and each sub-domain management domain vector based on the vector length of each dispute vector, the vector length of each sub-domain management domain vector and the corresponding vector dot product, and subtract the cosine similarity from 1 to obtain the cosine distance of each sub-domain.

[0036] S54: Calculate the sum and average of the cosine distances of all sub-fields of the first project management field to obtain an average cosine distance of the first project management field.

[0037] The calculation formula of cosine distance is as follows: ; ; ; Among them, A and B are two n-dimensional vectors, representing the dispute vector and the sub-domain management domain vector respectively; is the angle between A and B; the symbol Represents the dot product operation of vectors; Represents the norm of a vector. , The vectors A and B are A portion. is the jth cosine value of the i-th management field, and n is the number of documents. represents the cosine similarity between the dispute vector and each sub-domain management domain vector; cosine distance represents the cosine distance of the sub-domain; Represents the mean cosine distance of the first project management domain.

[0038] In this embodiment, refer to Figure 6 In the domain cosine distance calculation process shown in FIG, step S6 specifically first obtains the frequency index value, cosine distance index value, and mutual information index value of each sub-domain in the first project management domain to form an index value matrix; Based on the indicator value matrix, the frequency weight, cosine distance weight and mutual information weight are calculated using the entropy weight method.

[0039] Multiply the frequency weight by the frequency index value, the cosine distance weight by the cosine distance index value, and the mutual information weight by the mutual information index value respectively, and add the three product results to obtain the comprehensive importance value of each specific field.

[0040] The comprehensive importance values ​​of all project management fields in the first project management field are sorted in descending order, and according to a preset screening threshold, the sub-fields whose comprehensive importance values ​​exceed the preset screening threshold are screened out from the first project management field as the second project management field, thereby obtaining the important management fields that cause disputes.

[0041] This embodiment calculates the frequency, cosine distance, and mutual information for each project management area within the first project management area. The weight of each indicator is multiplied by its own indicator value and then added together to obtain the final importance value for each sub-area. The importance values ​​are then sorted from largest to smallest. This allows the top n project management areas to be screened based on a preset screening threshold, i.e., the second project management areas. The selected second project management areas are the areas that require special attention in practice; the more important they are, the more likely they are to cause disputes.

[0042] In this embodiment, step S7 specifically includes the following sub-steps: S71: Mapping the cause keywords of the sub-fields in the first project management field in the case dispute cause mapping table into a set of project field vectors through the Lawformer pre-training model; S72: Calculate the Pearson correlation between the project domain vectors of every two sub-domains, determine the correlation of the project management sub-domains based on the Pearson correlation calculation results, and screen out the sub-domains that are strongly correlated with the second project management domain from the first project management domain based on the correlation of the project management sub-domains. When screening strongly correlated sub-domains, you can set a strong correlation threshold to perform the screening. Although the sub-domains that are strongly correlated with the second project management domain are not the project management domains that directly cause disputes, these sub-domains are strongly correlated with the project management domains that are prone to cause disputes. Therefore, they also need attention. Therefore, this embodiment performs a correlation analysis on legal documents and project management domains, and screens out the sub-domains that need to be reminded to pay attention to the second project management domain, as well as the sub-domains that are strongly correlated with the second project management domain.

[0043] The calculation formula of Pearson correlation is: ; in, is a variable and variables The covariance of . , The variables are ,variable When When , there is a complete positive linear correlation. There is a complete negative linear correlation when Time is not linearly related.

[0044] This embodiment has the following technical advantages: 1. By utilizing the Large Language Model (LLM) to automatically extract dispute points from legal documents and generate a keyword mapping table, the subjectivity and limitations of traditional manual analysis are effectively overcome, ensuring the comprehensiveness and objectivity of dispute point extraction and laying an accurate data foundation for subsequent correlation analysis.

[0045] 2. By mapping dispute focus keywords to project management knowledge areas to generate a case dispute cause mapping table, a direct connection between legal disputes and the field of standardized project management is achieved, and complex legal texts are structured into a project management framework that can be quantified and analyzed.

[0046] 3. By using the entropy weight method to comprehensively calculate and rank the project management sub-fields that lead to disputes, it is possible to scientifically and objectively identify the core project management fields that have a significant impact on the disputes, effectively avoiding the one-sidedness of single indicator analysis and improving the accuracy and reliability of screening key causal fields.

[0047] 4. By mapping the cause keywords in the project management sub-fields that lead to disputes into text vectors and calculating the Pearson correlation between the vectors to obtain the project management sub-field correlation, we can deeply reveal the intrinsic correlation strength and synergy pattern between different project sub-fields, thereby more comprehensively understanding the complex causes of disputes.

[0048] 5. Finally, combining the importance ranking of project management sub-fields and the correlation between sub-fields can help prevent sub-fields and their strongly related fields that may lead to disputes in advance in project management practice, and realize a closed loop of correlation analysis based on objective data (legal document text) and rigorous algorithms (LLM extraction, entropy weight method screening, vector mapping and correlation calculation), so that the final causes of project disputes are highly objective, explainable and targeted.

[0049] Therefore, the present invention comprehensively utilizes technical means such as natural language processing, domain mapping, multi-index weight calculation and correlation analysis, which can systematically and objectively analyze and dig out the root causes of project disputes and their correlation relationships from massive legal documents, significantly reducing the subjective arbitrariness of the analysis process, and providing strong data support and decision-making basis for project risk management.

[0050] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for analyzing the association between legal documents and project management based on NLP, characterized by: include: Obtain legal documents and input them into a large language model to extract dispute points and obtain a dispute point table; Mapping the cause keywords in the dispute focus table to the project management knowledge domain to obtain a case dispute cause mapping table including the first project management domain; Count the number of times the sub-fields of the first project management field appear in the dispute focus table to obtain the frequency of occurrence of each sub-field in the first project management field; A joint vocabulary is constructed based on the dispute text in the dispute focus table and the case dispute cause mapping table, and the mutual information between the dispute text and the first project management field is calculated based on the joint vocabulary; Convert the dispute text in the dispute focus table into a dispute vector, and convert the first item management field in the case dispute cause mapping table into a management field vector, and calculate the cosine distance between each dispute vector and each management field vector; The entropy weight method is used to calculate the weight importance values ​​of frequency, cosine distance and mutual information, and the weight importance is sorted, and the second project management field is screened out from the first project management field according to the preset screening threshold; The cause keywords corresponding to the sub-fields in the first project management field are mapped into a group of project field vectors, and the Pearson correlation between each two groups of project field vectors is calculated to obtain the project management sub-field correlation. At the same time, based on the project management sub-field correlation, sub-fields that are strongly correlated with the second project management field are screened out from the first project management field.

2. The NLP-based legal document and project management field correlation analysis method according to claim 1 is characterized in that: The legal documents are obtained and input into the large language model to extract the focus of disputes, thereby obtaining a focus of disputes table, including: Obtain legal documents related to project management and input them into the big language model. The big language model is then used to extract the case number, plaintiff's claim text, and defendant's response text from each legal document to form the focus of the dispute. The TF-IDF algorithm is used to extract cause keywords from dispute focuses, and the cause keywords of multiple dispute focuses are combined to construct a dispute focus table for project management.

3. The NLP-based legal document and project management domain correlation analysis method according to claim 1 is characterized in that: The cause keywords in the dispute focus table are mapped to the project management knowledge field to obtain a case dispute cause mapping table containing the first project management field, including: Use the Jieba word segmentation tool to segment the cause keywords in the dispute focus table, remove stop words, and obtain a backup keyword table; Create a stop word list WL based on the expert opinions of project management and the field of invention; Based on the stop word list, the TF-IDF algorithm is used to perform secondary word segmentation on the backup keyword table to obtain the cause keyword table; Extract the sub-domains of the project management process and the corresponding large project management knowledge domains, and use the sub-domains and large project management knowledge domains as the first project management domain; A semantic comparison is performed between the cause keywords in the cause keyword table and the text of the first project management field, and the cause keywords are classified into the first project management field with similar semantics to the cause keywords to form a case dispute cause mapping table; each first project management field has a project field identifier.

4. The NLP-based legal document and project management field correlation analysis method according to claim 1 is characterized in that: The counting of the number of times the sub-fields of the first project management field appear in the dispute focus table to obtain the frequency of occurrence of each sub-field in the first project management field includes: A dispute label is set for each dispute focus in the dispute focus table according to the project field identifier of the first project management field, and each dispute label represents the project management field that caused the dispute; The number of times each first project management field appears in the dispute focus table is counted according to the dispute label to obtain the frequency of occurrence of each first project management field.

5. The NLP-based legal document and project management field correlation analysis method according to claim 1 is characterized in that: The method of constructing a joint vocabulary based on the dispute text in the dispute focus table and the case dispute cause mapping table, and calculating the mutual information between the dispute text and the first project management field based on the joint vocabulary, includes: Using the Lawformer pre-trained model, we construct a joint vocabulary for each dispute text in the dispute focus table and each first project management field in the case-dispute cause mapping table. Counting the first and second frequencies of each cause keyword in the joint vocabulary in the dispute text and the first project management field, and counting the third frequency of each cause keyword appearing in both the dispute text and the first project management field; According to the first frequency, second frequency and third frequency of each reason keyword in the joint vocabulary, the marginal probability and joint probability of the dispute text and the first project management field are calculated to obtain the mutual information of multiple sub-fields in each first project management field, and the mutual information of multiple sub-fields is summed and averaged to obtain the average mutual information of each first project management field.

6. The NLP-based legal document and project management domain correlation analysis method according to claim 1 is characterized in that: The method converts the dispute text in the dispute focus table into a dispute vector, converts the first project management field in the case dispute cause mapping table into a management field vector, and calculates the cosine distance between each dispute vector and each management field vector, including: Use the Lawformer pre-trained model to map each dispute focus in the dispute focus table into a dispute vector. Also use the Lawformer pre-trained model to map each sub-field of the first project management field in the case dispute cause mapping table into a sub-field management field vector. Use the Lawformer pre-trained model to calculate the vector dot product between each dispute vector and each sub-domain management domain vector, as well as the vector length of each dispute vector and the vector length of each sub-domain management domain vector; Calculate the cosine similarity of each dispute vector and each sub-domain management domain vector based on the vector length of each dispute vector, the vector length of each sub-domain management domain vector, and the corresponding vector dot product, and subtract the cosine similarity from 1 to obtain the cosine distance of each sub-domain; The cosine distances of all sub-fields of the first project management field are summed and averaged to obtain the average cosine distance of each first project management field.

7. The NLP-based legal document and project management domain correlation analysis method according to claim 1 is characterized in that: The entropy weight method is used to calculate the weight importance values ​​of frequency, cosine distance and mutual information, and the weight importance is sorted, and the second project management field is screened out from the first project management field according to a preset screening threshold, including: Obtaining frequency index values, cosine distance index values, and mutual information index values ​​for each sub-field in the first project management field to form an index value matrix; Based on the index value matrix, the frequency weight, cosine distance weight and mutual information weight are calculated using the entropy weight method; Multiply the frequency weight by the frequency index value, the cosine distance weight by the cosine distance index value, and the mutual information weight by the mutual information index value, and add the three product results to obtain the comprehensive importance value of each specific field; The comprehensive importance values ​​of all project management fields in the first project management field are sorted in descending order, and according to a preset screening threshold, the sub-fields whose comprehensive importance values ​​exceed the preset screening threshold are screened out from the first project management field as the second project management field.

8. The NLP-based legal document and project management domain correlation analysis method according to claim 1 is characterized in that: The cause keywords corresponding to the sub-fields in the first project management field are mapped into a group of project field vectors, and the Pearson correlation between each two groups of project field vectors is calculated to obtain the project management sub-field correlation. At the same time, based on the project management sub-field correlation, sub-fields that are strongly correlated with the second project management field are screened from the first project management field, including: The Lawformer pre-trained model is used to map the cause keywords of the sub-fields in the first project management field in the case dispute cause mapping table into a set of project field vectors; The Pearson correlation between the project domain vectors of every two sub-domains is calculated, and the correlation of the project management sub-domains is determined based on the Pearson correlation calculation results. The sub-domains that are strongly correlated with the second project management domain are screened out from the first project management domain based on the correlation of the project management sub-domains.

Images