Intelligent legal text analysis method and system based on NLP

Abstract

The invention relates to the technical field of text data processing, in particular to a legal text intelligent analysis method and system based on NLP, and the method comprises the steps: obtaining the correlation degree between legal vocabularies in each sentence of legal text data to be analyzed and other vocabularies; determining an association degree fluctuation index of each legal word and each word in the sentence to which the legal word belongs; determining an associated vocabulary of each legal vocabulary; analyzing the overall similar characteristics of the splicing vector when each legal vocabulary in each sentence and the associated vocabulary thereof appear together in each legal case, and determining the influence association degree between the legal vocabulary and the legal case; and analyzing the similarity between the splicing vectors of the legal vocabularies and the associated vocabularies thereof in each sentence, determining a matching similarity index of the legal vocabularies and the legal cases, and obtaining an optimal case corresponding to the legal vocabularies and the associated vocabularies thereof. The invention aims to improve the matching precision of the legal text and the case and provide more accurate case support.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of text data processing, in particular to a legal text intelligent analysis method and system based on NLP. BACKGROUND

[0002] Based on natural language processing (NLP), legal text intelligent analysis technology emerges as the times require, aiming to deal with the complex and professional terminology of legal texts and the high requirement for logical rigor. The core process of this technology usually includes basic processing such as word segmentation, part-of-speech tagging and named entity recognition (NER) of legal texts, aiming to extract key information such as names, places, organization names and legal provisions. Through these processing steps, NLP can efficiently identify and analyze important elements in legal texts. Further, through relationship extraction technology, the rights and obligations between parties are found out, and text classification technology is used to automatically classify legal documents. In recent years, deep learning models such as BERT, GPT and other pre-trained language models have performed well in legal text processing, and they can understand the context semantics of legal texts, significantly improving the accuracy of understanding. At the same time, combined with knowledge graph technology, entities and relationship structures in legal texts can be structured, and a relationship network between legal concept systems can be constructed to support intelligent retrieval and legal reasoning.

[0003] After obtaining the legal text data, considering that legal terms often have different meanings in different sentences, the legal terms in different contexts are affected by context words, resulting in changes in their interpretation. When matching and analyzing legal terms and corresponding cases, errors often occur, which may interfere with the user's reference to the wrong legal cases and have a serious impact. Therefore, when performing intelligent analysis on legal texts based on NLP, more accurate analysis and matching of legal terms are needed. SUMMARY

[0004] In view of the above, it is necessary to provide a legal text intelligent analysis method and system based on NLP to solve the above problems.

[0005] The first aspect of the present application provides a legal text intelligent analysis method based on NLP, which comprises: performing word segmentation processing on each sentence in the legal text data to be analyzed, and performing part-of-speech tagging; obtaining legal terms in each sentence, determining the correlation degree between each legal term and other terms in the sentence by comparing the word vectors between each legal term and other terms in the sentence; sequentially removing the remaining terms in each sentence except each legal term, analyzing the change characteristics of all correlation degrees obtained by each legal term before and after removing the terms, and determining the correlation degree fluctuation index of each legal term and each term in the sentence to which the legal term belongs; all the correlation degree fluctuation indexes obtained by analyzing each legal vocabulary, determine the associated vocabulary of each legal vocabulary; splice the part of speech and the word vector of each vocabulary to obtain the spliced vector of each vocabulary, analyze the overall similarity characteristics of the spliced vector when each legal vocabulary and its associated vocabulary appear together in each legal case, and determine the influence correlation degree between each legal vocabulary and each legal case; analyze the similarity between the spliced vectors of each legal vocabulary and its associated vocabulary in each sentence, combine the influence correlation degree, determine the matching similarity index of each legal vocabulary and each legal case, and obtain the best case corresponding to each legal vocabulary and its associated vocabulary.

[0006] The correlation degree is determined by the cosine similarity of the word vectors between each legal vocabulary and other vocabularies.

[0007] The determination of the correlation degree fluctuation index of each legal vocabulary and each vocabulary in the sentence to which the legal vocabulary belongs is specifically: remove the remaining vocabularies in each sentence except each legal vocabulary in turn; the sequence composed of all the correlation degrees obtained by each legal vocabulary in each sentence before removing each vocabulary is recorded as the first correlation degree sequence; the sequence composed of the remaining correlation degrees obtained by each legal vocabulary in each sentence after removing each vocabulary is recorded as the second correlation degree sequence. The negative correlation of the similarity between the first correlation degree sequence and the second correlation degree sequence of each legal vocabulary in each sentence is mapped as the correlation degree fluctuation index of each legal vocabulary and each vocabulary in the sentence to which the legal vocabulary belongs.

[0008] The specific formula of the correlation degree fluctuation index is: In the formula, represents the correlation degree fluctuation index between the legal vocabulary i and the vocabulary h in the sentence to which the legal vocabulary belongs; represents the first correlation degree sequence of the legal vocabulary i before removing the vocabulary h; represents the second correlation degree sequence of the legal vocabulary i after removing the vocabulary h; represents the exponential function with a natural constant as the base; represents the Pearson correlation coefficient function.

[0009] The associated vocabulary of each legal vocabulary is specifically the vocabulary corresponding to the maximum correlation degree fluctuation index in the sentence to which the legal vocabulary belongs.

[0010] The determination of the influence correlation degree between each legal vocabulary and each legal case is specifically: all sentences in which each legal vocabulary and its associated vocabulary in each case appear simultaneously, and calculating the similarity degree between the spliced vectors of each legal vocabulary and its associated vocabulary in each sentence, averaging all the similarity degrees, the influence correlation degree between each legal vocabulary and each legal case.

[0011] The similarity degree is obtained by cosine similarity.

[0012] The matching similarity index of each legal vocabulary and each legal case is determined, and the specific formula is: In the formula, the matching similarity of the associated combination corresponding to the legal vocabulary i to the legal case c is represented by sim (i, c), the spliced vector of the legal vocabulary i and its associated vocabulary j in the sentence is represented by v (i, j), the influence correlation degree between the associated combination corresponding to the legal vocabulary i and the legal case c is represented by sim (i, c), and the preset constant is represented by C. The matching similarity of the associated combination corresponding to the legal vocabulary i to the legal case c is represented by sim (i, c). The spliced vector of the legal vocabulary i and its associated vocabulary j in the sentence is represented by v (i, j). The influence correlation degree between the associated combination corresponding to the legal vocabulary i and the legal case c is represented by sim (i, c). The preset constant is represented by C.

[0013] The best case is specifically a legal case with the maximum matching similarity index corresponding to each legal vocabulary.

[0014] In a second aspect, the embodiments of the present application further provide a legal text intelligent analysis system based on NLP, which comprises a memory, a processor, and a computer program stored in the memory and running on the processor, and the processor implements the steps of the method according to any one of the preceding aspects when executing the computer program.

[0015] The beneficial effects of the above scheme are: the legal text data to be parsed is segmented and the part-of-speech is marked for each sentence, and through segmentation and part-of-speech marking, each word can be given grammatical and semantic information, helping the model to accurately distinguish legal terms and ordinary words, and providing a basis for subsequent semantic analysis; the legal words in each sentence are obtained, and by comparing the word vectors between each legal word and the remaining words in the sentence, the correlation degree of each legal word and other words is determined, which can effectively evaluate the semantic role of legal words in the sentence and the relationship with other words, help better understand the meaning of legal terms in the text in a specific context, and improve the accurate identification and interpretation of legal terms; the remaining words in each sentence except each legal word are removed in turn, and the change characteristics of all correlation degrees obtained by each legal word before and after the removal of the words are analyzed, and the correlation degree fluctuation index of each legal word and each word in the sentence is determined, which helps to reveal the core role and semantic changes of legal words in the sentence. By observing the fluctuation of the correlation degree before and after the removal of other words, the relationship strength between legal words and other words in the sentence can be quantified, and the understanding and processing of the semantic of legal text are improved; all correlation degree fluctuation indexes obtained by each legal word are analyzed to determine the associated words of each legal word, and by deeply analyzing the correlation degree fluctuation of each legal word, the most closely related words can be identified, helping to understand the meaning changes of the words in different contexts, and helping to build a semantic network between legal words, improving the accuracy and depth of legal text analysis; the concatenation of part-of-speech and word vectors can combine grammatical and semantic information, making the representation of words more comprehensive and accurate. This has a positive effect on subsequent similarity calculation, case matching and other tasks, and enhances the model's ability to process complex legal language; the overall similarity characteristics of the concatenated vectors of each legal word and its associated words when they appear together in each legal case are analyzed to determine the influence correlation degree between each legal word and each legal case. By analyzing the co-occurrence of legal words and their associated words in actual cases, the semantic changes and legal application scenarios of these words in different cases can be revealed, which helps to build a more accurate case matching model and improve the accuracy and efficiency of legal case retrieval; the matching similarity index of each legal word and each legal case is determined by analyzing the similarity between the concatenated vectors of each legal word and its associated words in each sentence, combined with the influence correlation degree, which can effectively evaluate the matching degree between a legal word and a specific legal case by calculating the similarity of legal words and associated words in the sentence, combined with the influence correlation degree, which helps to improve the matching accuracy of legal text and cases, and provides more accurate case support for legal professionals; by analyzing the best case corresponding to each legal word and its associated words, legal practitioners can more quickly and accurately find cases related to the current text, greatly improving the efficiency and accuracy of case handling. BRIEF DESCRIPTION OF DRAWINGS

[0016] Figure 1 A flowchart illustrating the steps of an NLP-based intelligent legal text parsing method provided in one embodiment of this application; Figure 2 This is a schematic diagram illustrating the acquisition of matching similarity metrics according to one embodiment of this application. Detailed Implementation

[0017] In the description of the embodiments in this application, the words "exemplary," "or," and "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design scheme described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design schemes. Specifically, the use of the words "exemplary," "or," and "for example" is intended to present the relevant concepts in a specific manner.

[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used in this application's specification is for the purpose of describing particular embodiments only and is not intended to be limiting of the application.

[0019] It should also be noted that the terms "first" and "second" in this application and its accompanying drawings are used to distinguish similar objects, rather than to describe a specific order or sequence. The methods disclosed in the embodiments of this application or the methods shown in the flowcharts include one or more steps for implementing the method. Without departing from the scope of protection of this application, the execution order of multiple steps can be interchanged, and some steps can also be deleted.

[0020] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0021] The following description, in conjunction with the accompanying drawings, details the specific scheme of the NLP-based intelligent legal text parsing method and system provided in this application.

[0022] Please see Figure 1 The diagram illustrates a flowchart of a method for intelligent parsing of legal text based on NLP, according to an embodiment of this application. The method includes the following steps: The first step is to segment each sentence in the legal text data to be parsed and then perform part-of-speech tagging.

[0023] In NLP-based intelligent legal text parsing, the first step is to collect legal text data, obtaining publicly available legal texts and cases from the internet, and collecting legal-related text documents and scanned copies from various legal websites. Further, the process involves obtaining corresponding definitions of legal terms from the internet, including but not limited to Chinese legal terminology compilations and other legal terminology parsing websites.

[0024] In this embodiment, OCR technology is used to extract text from the acquired legal text data, outputting the extracted text data. Simultaneously, text extraction is performed on the text file. For each sentence in the extracted text, a regular expression tool is used to remove irrelevant characters and punctuation marks, cleaning the text data. Then, the jieba word segmentation tool, combined with the legal thesaurus in the Tsinghua University Open Chinese Dictionary, is used to segment each sentence into lexical units (containing existing legal terms), and the Harbin Institute of Technology stop word list is used to remove stop words. Simultaneously, the LAC tool is used to perform part-of-speech tagging on all segmented words, enhancing the text's structure. All text processing processes use the UTF-8 encoding standard to avoid garbled characters. The use of OCR technology, the jieba word segmentation tool, and the LAC tool are all existing well-known technologies, and this application will not elaborate on them further.

[0025] The second step is to obtain legal terms in each sentence, and determine the correlation between each legal term and other terms by comparing the similarity features of word vectors between each legal term and the other terms in the sentence. Then, remove all other terms in each sentence except for each legal term, analyze the changes in the correlation of each legal term before and after word removal, and determine the correlation fluctuation index between each legal term and each term in the sentence.

[0026] The pre-processed vocabulary is matched and filtered against a custom legal terminology dictionary. This custom legal terminology dictionary is specifically a set of legal terms obtained from Baidu Wenku. The process of building a custom dictionary is well-known in this field and will not be elaborated further. Legal text terms are then extracted from the document data, and intelligent parsing is performed on these extracted terms.

[0027] When performing intelligent parsing of legal text terms, the legal meaning of a term can vary depending on the context and the surrounding words. For example, in the sentence "A third party acquires ownership of the movable property in good faith, and his rights are protected by law," the word "good faith" changes from its original meaning of "with good intentions" to "unaware and without gross negligence," significantly altering its legal connotation. Furthermore, due to the high rigor required in legal texts, when "good faith" is expressed as "unaware and without gross negligence," it is usually combined with "acquired" to highlight its specific legal meaning. Removing words that alter the semantics of the target term often results in a significant change in the target term's meaning, leading to substantial shifts in the relevance between the target term and other words in the context. Therefore, to avoid errors in interpreting legal terms, it is generally necessary to analyze the impact of surrounding words during the semantic analysis process.

[0028] Based on the above analysis, word vectors are calculated for the preprocessed vocabulary using the BERT model to obtain the word vector for each word. The BERT model is a well-known technology, and the specific operational steps will not be elaborated further. Taking a sentence containing legal term i (legal term i represents a term with legal interpretation) in a legal text as an example, the BERT model is used to obtain the word vector for each word in that sentence.

[0029] When removing words that can alter the semantics of the target word, the meaning of the target word often changes significantly. Therefore, we identify words in a sentence that have a significant impact on the target word: First, we calculate the correlation between the word vectors of each word in the sentence and the legal word i. In this embodiment, the correlation is calculated using cosine similarity.

[0030] Furthermore, when the meaning of the target word changes significantly, the degree of association between the target word and other words in the context also changes significantly. Based on this, for the sentence containing legal word i, the degree of association between each word and legal word i is obtained. Based on the order of the words in the sentence, extract the relevance of words other than word 'h'. Arranged from left to right, as the first relevance sequence of legal term i Furthermore, obtain the sentence after removing any word h except for the legal term i, and similarly obtain the relevance of each word in the current sentence to the legal term. Based on the order of the words in the sentence, determine the relevance of the sentence. Arranged from left to right, this is the second relevance sequence of legal term i after removing term h. Considering that the deletion of words that significantly impact legal term i will alter the relevance between the remaining words in the original sentence and legal term i, a relevance fluctuation index is constructed to analyze the changes in the relevance between other words in the sentence and legal term i before and after the removal of term h. Specifically, in this embodiment, the relevance fluctuation index between legal term i and term h in its sentence is defined as follows: Its formula is as follows: In the formula, This represents an exponential function with the natural constant as its base, avoiding computational complexity. The value is 0. This represents the Pearson correlation coefficient. It should be noted that, to ensure the validity of the Pearson correlation coefficient, the correlation between word h and legal word i is not included in the first correlation sequence. This is to ensure that the number of elements in the first and second correlation sequences remains consistent.

[0031] It should be understood that by calculating the correlation between the correlation sequence of the sentence after word removal and the correlation sequence of the original sentence, the more serious the impact of the removed word on the meaning of legal terms, the greater the difference between the correlation sequence after removal and the correlation sequence of the original sentence, resulting in a relatively small Pearson correlation coefficient and a larger corresponding correlation fluctuation index. Conversely, the less the removed word has on the meaning of legal terms, the less impact it usually has on the original sentence, and the relatively large correlation coefficient between its correlation sequence and the correlation sequence of the original sentence, resulting in a smaller corresponding correlation fluctuation index.

[0032] The third step is to analyze all correlation fluctuation indicators obtained for each legal term to determine the related terms for each legal term; to concatenate the part-of-speech and word vector of each term to obtain the concatenated vector of each term; to analyze the overall similarity features of the concatenated vectors when each legal term and its related terms co-occur in each legal case to determine the influence correlation between each legal term and each legal case; to analyze the similarity between the concatenated vectors of each legal term and its related terms in each sentence; and, in conjunction with the influence correlation, to determine the matching similarity index between each legal term and each legal case to obtain the best case corresponding to each legal term and its related terms.

[0033] For each legal term, all correlation fluctuation indicators were obtained. By calculating the maximum value among all correlation fluctuation indicators, the situation with the most severe impact on the legal term was identified. Furthermore, considering that legal terms form new meanings when combined with their adjuncts, their part-of-speech collocations are not random but follow the inherent high degree of standardization and logic of legal language, usually forming fixed pairs. For example, in a legal context, when "bona fide" modifies "acquisition" to form "bona fide acquisition," the part-of-speech of "acquisition" is a noun, while in "acquiring ownership," "acquisition" is a verb. The meaning represented by this term in different legal combinations will vary significantly from the meaning of the corresponding legal combination. Once this modification relationship changes, the legal meaning will change, producing different interpretations.

[0034] Based on this, taking legal term i as an example, the term removed when the correlation fluctuation index in the sentence is the largest is taken as the related term of legal term i. It should be understood that the related terms of the same legal term may be different in different sentences. The part of speech of the related term and the legal term in the original sentence are obtained respectively. The word vector of the corresponding part of speech is obtained by using the Word2Vec model. The word vector of the corresponding part of speech is concatenated with the word vector of the corresponding part of speech by using NumPy to obtain the concatenated vector of legal term i and related term j in the original sentence.

[0035] When matching legal terms to the best analytical case, there may be instances where corresponding legal terms and related terms appear in the same sentence but their meanings differ significantly due to their different positions. For example, when "good faith" and "acquire" are combined to form "acquire in good faith," it has a legal meaning. However, in the sentence "He treated others with sincere good faith, and therefore gained everyone's trust and respect," although "good faith" and "acquire" are in the same sentence, their meanings are quite different from "acquire in good faith." Therefore, when a sentence contains legal term i and related term j, then legal term i and related term j are considered a pair of key terms in that sentence.

[0036] When a sentence in a legal case contains each legal term and its associated term, the concatenated vectors of the two terms are obtained separately. Furthermore, the average similarity of the concatenated vectors of legal term i and its associated term j in all sentences where they coexist is calculated, yielding the influence correlation between each legal term and its associated term. In this embodiment, the similarity of the concatenated vectors is calculated using cosine similarity.

[0037] Specifically, legal term i is combined with its related term j to form a combination of legal term i, and the degree of influence between the combination of legal term i and legal case c is denoted as . Its specific formula is as follows: In the formula, Z represents the frequency of legal term i and its related term j appearing simultaneously in a sentence in legal case c. , Let i and j represent the concatenated vectors of legal term i and related term j in the d-th sentence of legal case c, respectively. This represents the cosine similarity function.

[0038] Furthermore, a matching similarity metric is constructed, with the following formula: In the formula, This represents the similarity between the associated combination of legal term i and legal case c. This represents the concatenated vector of legal term i and its associated term j in a sentence. This indicates the degree of influence between the associated combination of legal term i and legal case c. This represents a preset constant with a value of 0.01.

[0039] The diagram illustrating the acquisition of the matching similarity index is shown below. Figure 2 As shown.

[0040] It should be understood that when the combination of corresponding legal terms and related terms in the target case has the same meaning as the word combinations in the original sentence, the calculated and obtained... The closer the value is to 1, the better. The closer the calculated value is to 0, the better; when the combination of legal terms and related terms in the target case deviates significantly from the meaning of the word combinations in the original sentence, the calculated value... The further the value deviates from 1, the worse the calculation results. The larger the similarity, the better. Obtain the maximum matching similarity from the associated combinations corresponding to legal term i, and use its corresponding case as the best matching case to ensure that the case can effectively parse the legal term information.

[0041] The system obtains the best case analysis corresponding to the legal text to be matched, and concatenates word vectors based on the corresponding legal words and related words. It then calculates the cosine similarity between the concatenated word vectors and various interpretations of the legal words in the law, and takes the interpretation corresponding to the maximum value as the legal interpretation of the legal word. For the best case analysis and interpretation of the legal words, it outputs them as the case analysis and interpretation of the corresponding legal text words, respectively. Combined with a JavaScript event listening algorithm, the system associates and displays the corresponding text words with the corresponding case analysis and interpretation. When the user hovers the mouse over a legal text word, they can get the content of the text word (such as the word "force majeure") and its best case analysis and interpretation in the corresponding target legal text word, thus completing the intelligent parsing of legal text based on NLP.

[0042] Based on the same inventive concept as the above methods, this application also provides an NLP-based legal text intelligent parsing system, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of any one of the above-described NLP-based legal text intelligent parsing methods.

[0043] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. In some alternative implementations, the functions marked in the blocks may occur in a different order than that shown in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. In the descriptions corresponding to the flowcharts and block diagrams in the accompanying drawings, the operations or steps corresponding to different blocks may also occur in a different order than disclosed in the description; sometimes there is no specific order between different operations or steps. For example, two consecutive operations or steps may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. Each block in a block diagram and / or flowchart, and combinations of blocks in a block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0044] It will be apparent to those skilled in the art that this application is not limited to the details of the exemplary embodiments described above, and that this application can be implemented in other specific forms without departing from its essential characteristics. Therefore, the embodiments described above should be considered exemplary and non-limiting in all respects; modifications to the technical solutions described in the foregoing embodiments, or equivalent substitutions of some technical features, without causing the essence of the corresponding technical solutions to deviate from the scope of the technical solutions in the embodiments of this application, should all be included within the protection scope of this application.

Claims

1. A method for intelligent parsing of legal text based on NLP, characterized in that, The method includes the following steps: Each sentence in the legal text data to be parsed is segmented into words and then tagged with parts of speech. Legal terms in each sentence are obtained. By comparing the word vectors of each legal term with the other terms in the sentence, the correlation between each legal term and other terms is determined. All other terms in each sentence except for each legal term are removed in turn. The changes in the correlation of each legal term before and after word removal are analyzed to determine the correlation fluctuation index between each legal term and each term in the sentence. Analyze all correlation fluctuation indicators obtained for each legal term to determine the related terms for each legal term; concatenate the part-of-speech and word vector of each term to obtain the concatenated vector of each term; analyze the overall similarity features of the concatenated vectors when each legal term and its related terms co-occur in each legal case to determine the influence correlation between each legal term and each legal case; analyze the similarity between the concatenated vectors of each legal term and its related terms in each sentence; combine the influence correlation degree to determine the matching similarity index between each legal term and each legal case, and obtain the best case corresponding to each legal term and its related terms.

2. The NLP-based intelligent legal text parsing method as described in claim 1, characterized in that, The degree of relevance is determined by the cosine similarity of the word vectors of each legal term with other terms.

3. The NLP-based intelligent legal text parsing method as described in claim 1, characterized in that, The method for determining the correlation fluctuation index between each legal term and each term in its sentence is as follows: Remove all words from each sentence except for legal terms; The sequence of all relevance scores obtained for each legal term in each sentence before removing the terms is denoted as the first relevance sequence; the sequence of the remaining relevance scores obtained for each legal term in each sentence after removing the terms is denoted as the second relevance sequence. The negative correlation mapping of the similarity between the first correlation sequence and the second correlation sequence of each legal term in each sentence is used as the correlation fluctuation index between each legal term and each term in the sentence to which it belongs.

4. The NLP-based intelligent legal text parsing method as described in claim 3, characterized in that, The specific formula for the correlation fluctuation index is as follows: In the formula, This is an indicator of the fluctuation in the correlation between legal term i and term h in its sentence; This represents the first relevance sequence of legal term i before removing term h; This represents the second degree of relevance sequence of legal term i after removing term h; Represents an exponential function with the natural constant as its base; This represents the Pearson correlation coefficient function.

5. The NLP-based intelligent legal text parsing method as described in claim 1, characterized in that, The related words for each legal term are specifically the words corresponding to the sentence in which the correlation fluctuation index is the highest.

6. The NLP-based intelligent legal text parsing method as described in claim 1, characterized in that, The determination of the influence correlation between each legal term and each legal case specifically involves: Obtain all sentences in each case where each legal term and its associated terms appear simultaneously, and calculate the similarity between the concatenated vectors of each legal term and its associated terms in each sentence. Calculate the average of all obtained similarity values ​​to determine the influence correlation between each legal term and each legal case.

7. The NLP-based intelligent legal text parsing method as described in claim 6, characterized in that, The degree of similarity is obtained using cosine similarity.

8. The NLP-based intelligent legal text parsing method as described in claim 1, characterized in that, The specific formula for determining the matching similarity index between each legal term and each legal case is as follows: In the formula, This represents the similarity between the associated combination of legal term i and legal case c. This represents the concatenated vector of legal term i and its associated term j in a sentence. This indicates the degree of influence between the associated combination of legal term i and legal case c. This represents a preset constant.

9. The NLP-based intelligent legal text parsing method as described in claim 1, characterized in that, The best case is specifically the legal case with the highest matching similarity index corresponding to each legal term.

10. A legal text intelligent parsing system based on NLP, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1-9.

Images