An intelligent analysis method and system for text information based on large language models

Abstract

The present invention discloses an intelligent text information analysis method and system based on a large language model, which relates to the technical field of text analysis. By means of an algorithm for the call probability of text information and legal provisions, legal provisions similar to the text information are obtained. The legal provisions matched by the algorithm are screened manually. By means of the algorithm for the call probability, legal provisions similar to the text information are obtained. The legal provisions matched by the algorithm are screened manually. The differences between the legal provisions screened manually and the legal provisions with the highest call probability are compared, the difference features of the differences are extracted, and the recommendation ranking of the legal provisions with the same difference features in the text information is adjusted.

Description

Technical Field

[0001] The present invention relates to the technical field of text analysis, and particularly to an intelligent text information analysis method and system based on a large language model. Background Art

[0002] In order to reduce the workload in the process of consulting legal documents, a legal knowledge database is established, and a method of providing relevant legal knowledge for text content by combining natural language processing with machine learning is gradually popularized. For example, a Chinese patent with the publication number: CN111694945A and the name of "Legal Provision Association Recommendation Method and Device Based on Neural Network" discloses a method of providing legal document references based on statement call probability.

[0003] Since the target technology cannot completely get rid of the step of manually verifying the legal recommendation clauses generated by the computer, it is necessary to generalize the manually verified results of local text to similar positions in the text. In the prior art, there is a lack of means to repeat and generalize the manual verification process. When relevant personnel conduct legal clause verification, they still need to compare the calculation recommendation results item by item to obtain the most appropriate legal clause. Summary of the Invention

[0004] The purpose of the present invention is to provide an intelligent text information analysis method and system based on a large language model to solve the problems raised in the above background art.

[0005] To solve the above technical problems, the present invention provides the following technical solution: An intelligent text information analysis method based on a large language model;

[0006] Step S100: Collect the annotation records of relevant personnel on text information. The annotation records include: collecting a certain paragraph in the text information and the legal clauses cited in a certain paragraph, identifying the text elements in any paragraph, and forming a sample record by combining the text elements in the same paragraph with the corresponding legal clauses. Collect all the sample records of the first target record to obtain a comparison database;

[0007] Step S200: Identify at least two types of text elements in the text information, form a binary group by combining one of the text elements with the corresponding legal clause. In the comparison database, calculate the occurrence probability of another text element under the condition that a certain binary group appears, and record the occurrence probability as the influence coefficient of another text element on the legal clause;

[0008] Step S300: Collect a certain text information as the target text, input each paragraph of the target text into the legal clause call model respectively, and obtain the initial recommendation sequence of the legal clauses corresponding to each paragraph respectively;

[0009] Step S400: Take the paragraph corresponding to a second target record as the first target paragraph, and obtain the call probabilities of two legal provisions in the second target record;

[0010] Step S500: Identify the text elements in the first target paragraph. Respectively form binary groups with the legal provisions selected by the relevant personnel and the same type of text elements in the legal provision with the highest call probability, calculate the influence coefficients of the same type of text elements on the two legal provisions respectively, and extract the difference features between the text elements;

[0011] Step S600: Take the paragraphs that have semantic association with the first target paragraph as the second target paragraphs. In the second target paragraphs, adjust the order of the legal provisions recommended for the second target paragraphs according to the characteristics of the elements selected from the legal provisions by the relevant personnel.

[0012] Further, Step S200 includes:

[0013] Step S201: Identify two types of text elements in the same paragraph in the text information, and record them as the first text element and the second text element respectively. Form a binary group with any text element and the legal provision corresponding to the paragraph. Among them, the binary group formed by the second text element and the corresponding legal provision is denoted as (A, C), where A represents the second text element and C represents the legal provision corresponding to the same paragraph;

[0014] Step S202: In the comparison database, calculate the occurrence probability of the binary group (A, C), and calculate the occurrence probability pE of the first text element under the condition that the binary group (A, C) appears, pE = P(E|(A, C)), where E represents the first text element, and P() in the formula represents the probability calculation formula;

[0015] Step S203: Obtain the binary group (E, C) formed by the first text element and the legal provision corresponding to the first target paragraph, and calculate the occurrence probability pA of the second text element under the condition that the binary group (E, C) appears, pA = P(A|(E, C));

[0016] Step S204: Denote pE as the influence coefficient of the first text element on the legal provision corresponding to the same paragraph, and denote pA as the influence coefficient of the second text element on the legal provision corresponding to the same paragraph;

[0017] Regarding the occurrence of the binary group (A, C) as an event, P(E|(A, C)) calculates the probability of the occurrence of event E under the condition that event (A, C) occurs. Regarding the occurrence of the binary group (E, C) as an event, P(A|(E, C)) calculates the probability of the occurrence of event A under the condition that event (E, C) occurs;

[0018] Compare the values of pA and pE to obtain the correlation degrees of both the first text element and the second text element with C. When pA > pE, it indicates that in the historical record, the first text element has a higher correlation degree with legal provision C compared to the second text element. The correlation degree serves as a measure of the distance between the text element and the legal provision in this solution.

[0019] Since the dominant text elements are different, the applicable legal provisions in the text information are also different. By calculating the dominant positions of different text elements in the text relative to the legal provisions, the recommended legal provisions are optimized. For the same legal provision cla, when P(E|(A, C)) > P(A|(E, C)), it indicates that for legal provision cla, event E is in the dominant position.

[0020] Further, step S300 includes:

[0021] Step S301: Obtain several legal provisions matched with the first target paragraph and obtain the invocation probabilities of each legal provision.

[0022] Step S302: Arrange the invocation probabilities of the legal provisions and the first target paragraph from high to low, and gather to obtain the initial recommended sequence of the legal provisions corresponding to the first target paragraph.

[0023] Step S303: In the initial recommended sequence, obtain the legal provision with the highest invocation probability and denote it as the first target provision of the first target paragraph. Obtain the invocation probability of the legal provision in the first target paragraph and denote it as α.

[0024] Further, step S400 includes:

[0025] Step S401: In the second adjustment record, the legal provision selected by the relevant person is denoted as the second target provision of the first target paragraph.

[0026] Step S402: In the first target paragraph, obtain the invocation probability of the second target provision and denote it as β, satisfying the condition β < α.

[0027] Further, step S500 includes:

[0028] Step S501: Obtain the first text element and the second text element in the first target paragraph. Under the condition of the first target provision, the influence coefficient of the first text element is denoted as pE1, the influence coefficient of the second text element is denoted as pA1. Under the condition of the second target provision, the influence coefficient of the first text element is denoted as pE2, and the influence coefficient of the second text element is denoted as pA2.

[0029] Step S502: When pE1 > pA1 and pA2 > pE2, it is the first adjustment mode. In the first adjustment mode, calculate kd = pA2 - pE2.

[0030] In the first target paragraph, there is a relationship pE1 > pA1. For the first target paragraph, the legal provision matched by the computer is dominated by the first text element. After proofreading by the relevant personnel, in the adopted legal provision, there is a relationship pA2 > pE2, and it is dominated by the second text element;

[0031] It shows that there are defects in the legal provision recommendation process for the first target paragraph in the recommendation of legal provisions dominated by the first text element and the second text element. In a paragraph where both the first text element and the second text element exist, it is necessary to increase the weight of the legal provision dominated by the second text element so that the legal provision dominated by the second text element is more easily viewed by the relevant personnel;

[0032] Step S503: When pE1 > pA1, calculate k1 = pE1 - pA1, k2 = pE2 - pA2. Among them, when k1 < k2, it is the second adjustment mode, and when k1 > k2, it is the third adjustment mode;

[0033] For the first target paragraph, k1 represents the difference degree between the legal provision matched by the computer and the first text element and the second text element in the first target paragraph, and k2 represents the difference degree between the legal provision selected by the relevant personnel and the first text element and the second text element in the first target paragraph;

[0034] k1 < k2 indicates that the gap in the dominant position between the first text element and the second text element increases. Compared with the first target paragraph, the legal provisions before and after the target adjustment show that the gap in the dominant position between the first text element and the second text element increases. It can be inferred that the legal provision with a larger gap in the dominant position between the first text element and the second text element is more applicable to the target text. It is necessary to increase the weight value of the legal provision with a larger gap in the dominant position between the first text element and the second text element among the legal provisions matched by the target text;

[0035] k1 > k2 indicates that the gap in the dominant position between the first text element and the second text element decreases. Compared with the first target paragraph, the legal provisions before and after the target adjustment show that the gap in the dominant position between the first text element and the second text element decreases. It can be inferred that the legal provision with a smaller gap in the dominant position between the first text element and the second text element is more applicable to the target text. It is necessary to increase the weight value of the legal provision with a smaller gap in the dominant position between the first text element and the second text element among the legal provisions matched by the target text.

[0036] Furthermore, step S600 includes:

[0037] Step S601: Obtain the initial recommendation sequence of legal provisions corresponding to the second target paragraph, and establish an adjustment coefficient calculation function H for legal provisions, where H = μ + W. Here, μ represents the invocation probability of the legal provision and the second target paragraph, and W represents the adjustment weight;

[0038] Step S602: Obtain the influence coefficient of the first text element in the second target paragraph, denoted as pE3, and the influence coefficient of the second text element, denoted as pA3;

[0039] Step S603: According to the matching conditions of the first adjustment mode, the second adjustment mode, and the third adjustment mode respectively, obtain the corresponding legal provisions in the initial recommendation sequence of the second target paragraph;

[0040] When pE1 > pA1, pA2 > pE2, and pA3 > pE3, match the first adjustment mode. Among the legal provisions corresponding to the second target paragraph, obtain the legal provisions that meet the condition pA3 > pE3, calculate ke = pA3 - pE3, obtain the invocation probability μ1 of the legal provision and the second target paragraph, and calculate the adjustment coefficient H1 in the first adjustment mode, where H1 = μ1 + (ke / kd) × (α - β);

[0041] When pE1 > pA1, pA3 < pE3, and k1 < k2, match the second adjustment mode. Among the legal provisions corresponding to the second target paragraph, obtain the legal provisions that meet the condition pA3 < pE3, obtain the invocation probability μ2 of the legal provision and the second target paragraph, and calculate the adjustment coefficient H2 in the second matching mode, where H2 = μ2 + (k3 / k2) × (α - β), and k3 = pE3 - pA3;

[0042] When pE1 > pA1, pA3 < pE3, and k1 > k2, match the third adjustment mode. Among the legal provisions corresponding to the second target paragraph, obtain the legal provisions that meet the condition pA3 < pE3, obtain the invocation probability μ3 of the legal provision and the third target paragraph, and calculate the adjustment coefficient H3 in the third adjustment mode, where H3 = μ3 + (k2 / k3) × (α - β), and k3 = pE3 - pA3;

[0043] Step S605: In the initial recommendation sequence of legal provisions corresponding to the second target paragraph, replace the invocation probability of the corresponding legal provision with the adjustment coefficient, re - arrange the values in the initial recommendation sequence from high to low to obtain a recommended update sequence, and arrange the legal provisions according to the invocation probability or adjustment coefficient corresponding to the legal provisions in the recommended update sequence from high to low, and recommend the legal provisions to relevant personnel.

[0044] To better implement the above method, an intelligent text information analysis system based on a large language model is also proposed, including a comparison database management module, an influence coefficient calculation module, an association matching module, a target adjustment management module, a difference calculation module, and an adjustment management module. Among them, the comparison database management module is used to manage the comparison database, the influence coefficient calculation module is used to calculate the influence coefficients of text elements, the association matching module is used to obtain the legal provisions matched by each paragraph, the target adjustment management module is used to record the target adjustment, the difference calculation module is used to calculate the degree of difference of the text elements corresponding to the legal provisions before and after the target adjustment, and the adjustment management module is used to adjust the order of the legal provisions matched by the second target paragraph;

[0045] Further, the influence coefficient calculation module includes: a text element acquisition unit, a binary tuple management unit, and an influence coefficient calculation unit. Among them, the text element acquisition unit is used to acquire the text elements in the text information, the binary tuple management unit is used to manage the binary tuples composed of the first text element or the second text element and the corresponding legal provisions, and the first influence coefficient calculation unit is used to calculate the influence coefficients of the second text element of the first text element in the first target paragraph;

[0046] Further, the association matching module includes: an association matching model management unit and a first sorting unit. Among them, the association matching model management unit is used to manage the legal provision call model, and the first sorting unit is used to obtain the initial recommendation sequence of the paragraphs in the target text;

[0047] Further, the target adjustment management module includes: a target adjustment acquisition unit and a call probability comparison unit. Among them, the target adjustment acquisition unit is used to acquire the target adjustment, and the call probability comparison unit is used to compare the first target clause and the second target clause of the first target paragraph;

[0048] Further, the difference calculation module includes: an influence difference calculation unit and an adjustment mode acquisition unit. Among them, the influence difference calculation unit is used to calculate the difference in influence coefficients under different legal provisions in the first target paragraph, and the adjustment mode acquisition unit is used to obtain the adjustment mode in the second target paragraph according to the difference in influence coefficients;

[0049] Further, the adjustment management module includes: a second influence coefficient calculation unit, a pattern matching unit, an adjustment coefficient calculation unit, and a second sorting unit. Among them, the second influence coefficient calculation unit is used to calculate the influence coefficients of the second text element of the first text element in the second target paragraph, the pattern matching unit is used to match the adjustment modes corresponding to each legal provision in the legal provisions corresponding to the second target paragraph according to the adjustment mode, the adjustment coefficient calculation unit is used to calculate the adjustment coefficients under each matching mode through an adjustment coefficient calculation function, and the second sorting unit is used to update the recommendation sequence of the legal articles.

[0050] Compared with the prior art, the beneficial effects achieved by the present invention are as follows: The present invention compares the legally selected legal provisions with the computer-recommended legal provisions, analyzes the dominant text elements in the legal provisions, and determines the relative differences in the dominant text elements in the legal provisions after manual selection. The changes in the relative differences are used as modification features, and these modification features are repeated and propagated to rearrange the order of the computer-recommended legal provisions, making it easier for relevant personnel to find the appropriate legal provisions that match the text. Description of the Drawings

[0051] The drawings are used to provide a further understanding of the present invention and constitute a part of the specification. They are used in conjunction with the embodiments of the present invention to explain the present invention and do not limit the present invention. In the drawings:

[0052] Figure 1 is a schematic structural diagram of an intelligent text information analysis system based on a large language model according to the present invention;

[0053] Figure 2 is a schematic flowchart of an intelligent text information analysis method based on a large language model according to the present invention. Detailed Embodiments

[0054] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0055] Please refer to Figure 1 and Figure 2 The present invention provides a technical solution: an intelligent text information analysis method based on a large language model;

[0056] Step S100: Collect the annotation records of relevant personnel on the text information. The annotation records include: collecting a certain paragraph in the text information and the legal provisions cited in a certain paragraph, identifying the text elements in any paragraph, and forming a sample record by combining the text elements in the same paragraph with the corresponding legal provisions. All the sample records of the first target record are collected to obtain a comparison database;

[0057] Specifically, by collecting the annotations of relevant personnel on a certain paragraph, the annotations include (G, cla), where G includes the text elements that appear in a certain paragraph, and cla represents the legal provisions corresponding to a certain paragraph.

[0058] Step S200: Identify at least two types of text elements in the text information, form a binary tuple by combining one of the text elements with the corresponding legal clause. In the comparison database, calculate the occurrence probability of another text element under the condition that a certain binary tuple appears, and record the occurrence probability as the influence coefficient of the other text element on the legal clause;

[0059] In the embodiment, the text elements are, for example: the parties to the contract, the domicile of the parties, the subject matter of the contract, the place of contract performance or the time of contract performance; the manner of contract performance, the obligations specified in the contract or the limitations on breach of contract; the remuneration obtained, liability for breach of contract or the method of resolving disputes;

[0060] Among them, step S200 includes:

[0061] Step S201: Identify two text elements in the same paragraph of the text information, and record them as the first text element and the second text element respectively. Form a binary tuple by combining any text element with the legal clause corresponding to the paragraph. Among them, the binary tuple formed by the second text element and the corresponding legal clause is denoted as (A, C), where A represents the second text element and C represents the legal clause corresponding to the same paragraph;

[0062] Step S202: In the comparison database, calculate the occurrence probability of the binary tuple (A, C), and calculate the occurrence probability pE of the first text element under the condition that the binary tuple (A, C) appears, pE = P(E|(A, C)), where E represents the first text element, and P() in the formula represents the probability calculation formula;

[0063] Step S203: Obtain the binary tuple (E, C) formed by the first text element and the legal clause corresponding to the first target paragraph, and calculate the occurrence probability pA of the second text element under the condition that the binary tuple (E, C) appears, pA = P(A|(E, C));

[0064] Step S204: Record pE as the influence coefficient of the first text element on the legal clause corresponding to the same paragraph, and record pA as the influence coefficient of the second text element on the legal clause corresponding to the same paragraph.

[0065] Step S300: Collect a certain text information as the target text, input each paragraph of the target text into the legal clause invocation model respectively, and obtain the initial recommendation sequence of the legal clauses corresponding to each paragraph;

[0066] Among them, step S300 includes:

[0067] Step S301: Obtain several legal clauses matching the first target paragraph, and obtain the invocation probabilities of each legal clause;

[0068] Step S302: Arrange the calling probabilities of legal clauses and the first target paragraph from high to low, and gather them to obtain the initial recommendation sequence of legal clauses corresponding to the first target paragraph;

[0069] Step S303: In the initial recommendation sequence, obtain the legal clause with the highest calling probability and denote it as the first target clause of the first target paragraph, and obtain the calling probability of the legal clause in the first target paragraph and denote it as α.

[0070] Step S400: Take the paragraph corresponding to a second target record as the first target paragraph, and obtain the calling probabilities of two legal clauses in the second target record;

[0071] Among them, Step S400 includes:

[0072] Step S401: In the second adjustment record, the legal clause selected by the relevant person is denoted as the second target clause of the first target paragraph;

[0073] Step S402: In the first target paragraph, obtain the calling probability of the second target clause and denote it as β, satisfying the condition β < α.

[0074] Step S500: Identify the text elements in the first target paragraph, respectively form binary groups of the same type of text elements in the legal clause selected by the relevant person and the legal clause with the highest calling probability, calculate the influence coefficients of the same type of text elements on the two legal clauses respectively, and extract the difference features between the text elements;

[0075] Among them, Step S500 includes:

[0076] Step S501: Obtain the first text element and the second text element in the first target paragraph. Under the condition of the first target clause, the influence coefficient of the first text element is denoted as pE1, and the influence coefficient of the second text element is denoted as pA1. Under the condition of the second target clause, the influence coefficient of the first text element is denoted as pE2, and the influence coefficient of the second text element is denoted as pA2;

[0077] Step S502: When pE1 > pA1 and pA2 > pE2, it is the first adjustment mode. In the first adjustment mode, calculate kd = pA2 - pE2;

[0078] Step S503: When pE1 > pA1, calculate k1 = pE1 - pA1, k2 = pE2 - pA2. Among them, when k1 < k2, it matches the second adjustment mode, and when k1 > k2, it is the third adjustment mode.

[0079] Step S600: Obtain the paragraphs that have semantic associations with the first target paragraph as the second target paragraphs. In the second target paragraphs, adjust the recommended order of the legal provisions according to the characteristics of the elements selected from the legal provisions by the relevant personnel.

[0080] During the implementation process, extract the second target paragraphs related to the first target paragraph by means of extracting correlation words, text indexing, or semantic call probability calculation.

[0081] Among them, step S600 includes:

[0082] Step S601: Obtain the initial recommended sequence of the legal provisions corresponding to the second target paragraphs, and establish an adjustment coefficient calculation function H for the legal provisions, H = μ + W, where μ represents the call probability of the legal provisions and the second target paragraphs, and W represents the adjustment weight.

[0083] Step S602: Obtain the influence coefficient of the first text element in the second target paragraph, denoted as pE3, and the influence coefficient of the second text element, denoted as pA3.

[0084] Step S603: Obtain the corresponding legal provisions in the initial recommended sequence of the second target paragraphs respectively according to the matching conditions of the first adjustment mode, the second adjustment mode, and the third adjustment mode.

[0085] When pE1 > pA1, pA2 > pE2, and pA3 > pE3, match the first adjustment mode, obtain the legal provisions that meet the condition of pA3 > pE3 among the legal provisions corresponding to the second target paragraphs, calculate ke = pA3 - pE3, obtain the call probability μ1 of the legal provisions and the second target paragraphs, and calculate the adjustment coefficient H1 in the first adjustment mode, H1 = μ1 + (ke / kd) × (α - β).

[0086] When pE1 > pA1, pA3 < pE3, and k1 < k2, match the second adjustment mode, obtain the legal provisions that meet the condition of pA3 < pE3 among the legal provisions corresponding to the second target paragraphs, obtain the call probability μ2 of the legal provisions and the second target paragraphs, and calculate the adjustment coefficient H2 in the second matching mode, H2 = μ2 + (k3 / k2) × (α - β), where k3 = pE3 - pA3.

[0087] When pE1 > pA1, pA3 < pE3, and k1 > k2, match the third adjustment mode, obtain the legal provisions that meet the condition of pA3 < pE3 among the legal provisions corresponding to the second target paragraphs, obtain the call probability μ3 of the legal provisions and the third target paragraphs, and calculate the adjustment coefficient H3 in the third adjustment mode, H3 = μ3 + (k2 / k3) × (α - β), where k3 = pE3 - pA3.

[0088] For example: Obtain the first target paragraph L1, and obtain the legal clauses cla1 and cla2 that match L1. Among them, the invocation probability α of cla1 with respect to the first target paragraph is 0.75, and the invocation probability β of cla2 with respect to the first target paragraph is 0.5;

[0089] Obtain the second target paragraph L2 and the legal clauses matched by L2;

[0090] In the first adjustment mode, calculate the influence coefficients of the first text element and the second text element in the first target paragraph, and satisfy the conditions pE1 > pA1 and pA2 > pE2. In the second target paragraph, search for the legal clauses that satisfy the condition pA3 > pE3;

[0091] Obtain the legal clause cla31 that meets the conditions among the legal clauses matched by L2. The invocation probability of cla31 with respect to L2 is μ1;

[0092] Calculate the adjustment coefficient H11 = μ1 + (ke / kd)×(α - β). If it satisfies the boundary conditions that the difference in the dominant text elements in cla31 is the same as the difference in the dominant text elements in cla2, and the invocation probability with respect to the second target paragraph is equal to β, that is, μ1 = β, pA3 = pA2, pE3 = pE2, then calculate H11 = α. If ke > kd, that is, pA3 > pE3, the dominant position of the second text element in cla3 is more prominent, and calculate H12 under the condition that the invocation probability remains unchanged, and H12 > H11;

[0093] In the second adjustment mode, calculate the influence coefficients of the first text element and the second text element in the first target paragraph, and satisfy the conditions pE1 > pA1 and k1 < k2. In the second target paragraph, search for the legal clauses that satisfy the condition pE3 > pA3;

[0094] Obtain the legal clause cla32 that meets the conditions among the legal clauses matched by L2. The invocation probability of cla32 with respect to L2 is μ2;

[0095] Calculate the adjustment coefficient H21 = μ2 + (k3 / k2)×(α - β). If it satisfies the boundary conditions that the difference in the dominant text elements in cla32 is the same as the difference in the dominant text elements in cla2, and the invocation probability with respect to the second target paragraph is equal to β, that is, μ2 = β, pA3 = pA2, pE3 = pE2, then calculate H21 = α. If k3 > k2, that is, pE3 - pA3 > pE3 - pA3, the difference in the dominant positions of the first text element and the second text element in cla3 is more prominent, and calculate H22 under the condition that the invocation probability remains unchanged, and H22 > H21;

[0096] In the third adjustment mode, calculate the influence coefficients of the first text element and the second text element in the first target paragraph, and satisfy the conditions pE1 > pA1 and k1 > k2. In the second target paragraph, search for the legal provisions corresponding to the condition pE3 < pA3;

[0097] Obtain the legal provision cla33 that meets the conditions among the legal provisions matched by L2. The call probability of cla33 and L2 is μ3;

[0098] Calculate the adjustment coefficient H31 = μ3 + (k2 / k3)×(α - β). If it satisfies the boundary condition that the difference in the dominant text elements in cla32 is the same as the difference in the dominant text elements in cla2, and the call probability with the second target paragraph is equal to β, that is, μ3 = β, pA3 = pA2, pE3 = pE2, then calculate H31 = α. If k3 < k2, that is, pE3 - pA3 < pE3 - pA3, in cla3, the difference in the dominant positions of the first text element and the second text element decreases, and calculate H32 under the condition that the call probability remains unchanged. H32 > H31;

[0099] Step S605: In the initial recommendation sequence of the legal provisions corresponding to the second target paragraph, replace the call probability of the corresponding legal provisions with the adjustment coefficient, re - arrange the values in the initial recommendation sequence from high to low to obtain a recommended update sequence, and arrange the legal provisions according to the call probability or adjustment coefficient corresponding to the legal provisions in the recommended update sequence from high to low, and recommend the legal provisions to relevant personnel.

[0100] The system includes: a comparison database management module, an influence coefficient calculation module, an association matching module, a target adjustment management module, a difference calculation module, and an adjustment management module;

[0101] Among them, the comparison database management module is used to manage the comparison database;

[0102] Among them, the influence coefficient calculation module is used to calculate the influence coefficients of text elements. Among them, the influence coefficient calculation module includes: a text element acquisition unit, a binary group management unit, and an influence coefficient calculation unit. Among them, the text element acquisition unit is used to acquire text elements in text information, the binary group management unit is used to manage the binary groups composed of the first text element or the second text element and the corresponding legal provisions, and the first influence coefficient calculation unit is used to calculate the influence coefficients of the first text element and the second text element in the first target paragraph;

[0103] Among them, the association matching module includes: an association matching model management unit and a first sorting unit. Among them, the association matching model management unit is used to manage the legal provision call model, and the first sorting unit is used to obtain the initial recommendation sequence of paragraphs in the target text;

[0104] Among them, the target adjustment management module includes: a target adjustment acquisition unit and a call probability comparison unit. Among them, the target adjustment acquisition unit is used to acquire the target adjustment, and the call probability comparison unit is used to compare the first target clause and the second target clause of the first target paragraph;

[0105] Among them, the difference calculation module includes: an influence difference calculation unit and an adjustment mode acquisition unit. Among them, the influence difference calculation unit is used to calculate the difference in the influence coefficients under different legal clause conditions in the first target paragraph, and the adjustment mode acquisition unit is used to obtain the adjustment mode in the second target paragraph according to the difference in the influence coefficients;

[0106] Among them, the adjustment management module is used to adjust the order of the legal clauses matched by the second target paragraph. Among them, the adjustment management module includes: a second influence coefficient calculation unit, a pattern matching unit, an adjustment coefficient calculation unit, and a second sorting unit. Among them, the second influence coefficient calculation unit is used to calculate the influence coefficient of the second text element of the first text element in the second target paragraph, the pattern matching unit is used to match the adjustment mode corresponding to each legal clause in the legal clauses corresponding to the second target paragraph according to the adjustment mode, the adjustment coefficient calculation unit is used to calculate the adjustment coefficient under each matching mode through an adjustment coefficient calculation function, and the second sorting unit is used to update the recommended sequence of the legal articles.

[0107] 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. Moreover, the terms "include", "comprise" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or device.

[0108] Finally, it should be noted that the above are only the preferred embodiments of the present invention and are not used to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. An intelligent analysis method for text information based on large language models, characterized in that, The method comprises the following steps: Step S100: Collecting the annotation records of the text information by relevant personnel, the annotation records include: collecting a certain paragraph in the text information and the legal clauses cited in the certain paragraph, identifying the text elements in any paragraph, combining the text elements in the same paragraph and the corresponding legal clauses into a sample record, and collecting all the sample records of the first target record to obtain a comparison database; Step S200: Identify at least two text elements in the text information, combine one of the text elements with the corresponding legal clause into a tuple, calculate the occurrence probability of the other text element under the condition that a certain tuple appears in the comparison database, and record the occurrence probability as the influence coefficient of the other text element on the legal clause; Step S300: collecting a certain text information as a target text, inputting each paragraph in the target text into the legal clause calling model, and obtaining an initial recommended sequence of legal clauses corresponding to each paragraph; Step S400: taking a paragraph corresponding to a second target record as a first target paragraph, and obtaining the calling probability of two legal clauses in the second target record; Step S500: Identify the text elements in the first target paragraph, respectively form two tuples of the same text elements in the legal clause selected by the relevant personnel and the legal clause with the highest call probability, respectively calculate the influence coefficients of the same text elements on the two legal clauses, and extract the difference features between the text elements; Step S500 includes: Step S501: Obtain the first text element and the second text element in the first target paragraph. Under the condition of the first target clause, the influence coefficient of the first text element is recorded as pE1, and the influence coefficient of the second text element is recorded as pA1. Under the condition of the second target clause, the influence coefficient of the first text element is recorded as pE2. The influence coefficient of the second text element is recorded as pA2. Step S502: when pE1>pA1 and pA2>pE2, it is the first adjustment mode. In the first adjustment mode, kd=pA2-pE2 is calculated; Step S503: When pE1>pA1, calculate k1=pE1-pA1, k2=pE2-pA2, wherein when k1<k2, it matches the second adjustment mode, and when k1>k2, it matches the third adjustment mode. Step S600: A paragraph that is semantically associated with the first target paragraph is used as a second target paragraph. In the second target paragraph, the order of legal clauses recommended by the second target paragraph is adjusted according to the features of the elements in the legal clauses selected by the relevant personnel.

2. The intelligent text information analysis method based on a large language model according to claim 1, wherein: Step S200 includes: Step S201: Identify two text elements in the same paragraph in the text information, record them as the first text element and the second text element respectively, and form a tuple of any text element and the legal clause corresponding to the paragraph, wherein the tuple composed of the second text element and the corresponding legal clause is recorded as (A, C), wherein A represents the second text element, and C represents the legal clause corresponding to the same paragraph; Step S202: Calculate the occurrence probability of the binary pair (A, C) in the comparison database. Calculate the occurrence probability pE of the first text element under the condition that the binary pair (A, C) appears, where pE = P(E|(A, C)), E represents the first text element, and P() in the formula represents the probability calculation formula; Step S203: Obtain the binary pair (E, C) composed of the first text element and the legal clause corresponding to the first target segment, and calculate the occurrence probability pA of the second text element under the condition that the binary pair (E, C) appears, where pA = P(A|(E, C)); Step S204: Denote pE as the influence coefficient of the first text element on the legal clause corresponding to the same paragraph, and denote pA as the influence coefficient of the second text element on the legal clause corresponding to the same paragraph.

3. The intelligent text information analysis method based on a large language model according to claim 2, wherein: Step S300 includes: Step S301: Obtain several legal clauses matched by the first target paragraph, and obtain the call probabilities of each legal clause; Step S302: Arrange the call probabilities of the legal clauses and the first target paragraph from high to low, and gather to obtain the initial recommendation sequence of the legal clauses corresponding to the first target paragraph; Step S303: Obtain the legal clause with the highest call probability in the initial recommendation sequence as the first target clause of the first target paragraph, and obtain the call probability of the legal clause in the first target paragraph as α.

4. The intelligent text information analysis method based on a large language model according to claim 3, wherein: Step S400 includes: Step S401: In the second adjustment record, the legal clause selected by the relevant person is denoted as the second target clause of the first target paragraph; Step S402: In the first target paragraph, obtain the call probability of the second target clause as β, satisfying the condition β < α.

5. The intelligent text information analysis method based on a large language model according to claim 4, wherein: Step S600 includes: Step S601: Obtain the initial recommendation sequence of the legal clauses corresponding to the second target paragraph, and establish the adjustment coefficient calculation function H of the legal clause, where H = μ + W, μ represents the call probability of the legal clause and the second target paragraph, and W represents the adjustment weight; Step S602: Obtain the influence coefficient of the first text element in the second target paragraph as pE3, and the influence coefficient of the second text element as pA3; Step S603: Obtain the corresponding legal clauses in the initial recommendation sequence of the second target paragraph according to the matching conditions of the first adjustment mode, the second adjustment mode, and the third adjustment mode respectively; Step S604: When pE1 > pA1, pA2 > pE2 and pA3 > pE3, match the first adjustment mode, obtain the legal clauses that meet the condition pA3 > pE3 among the legal clauses corresponding to the second target paragraph, calculate ke = pA3 - pE3, obtain the call probability μ1 of the legal clause and the second target paragraph, and calculate the adjustment coefficient H1 in the first adjustment mode, where H1 = μ1 + (ke / kd) × (α - β); When pE1 > pA1, pA3 < pE3, and k1 < k2, match the second adjustment mode, obtain the legal provisions in the legal provisions corresponding to the second target paragraph that meet the condition of pA3 < pE3, obtain the invocation probability μ2 of the legal provisions and the second target paragraph, and calculate the adjustment coefficient H2 in the second matching mode. H2 = μ2 + (k3 / k2) × (α - β), where k3 = pE3 - pA3; When pE1 > pA1, pA3 < pE3, and k1 > k2, match the third adjustment mode, obtain the legal provisions in the legal provisions corresponding to the second target paragraph that meet the condition of pA3 < pE3, obtain the invocation probability μ3 of the legal provisions and the third target paragraph, and calculate the adjustment coefficient H3 in the third matching mode. H3 = μ3 + (k2 / k3) × (α - β), where k3 = pE3 - pA3; Step S605: In the initial recommendation sequence of the legal provisions corresponding to the second target paragraph, replace the invocation probability of the corresponding legal provisions with the adjustment coefficient, re-arrange the values in the initial recommendation sequence from high to low to obtain a recommended updated sequence, and recommend the legal provisions to relevant personnel according to the invocation probability or adjustment coefficient corresponding to the legal provisions in the recommended updated sequence from high to low.

6. An intelligent text information analysis system based on a large language model, which is used to execute an intelligent text information analysis method based on a large language model according to any one of claims 1-5, and is characterized in that The system includes the following modules: a comparison database management module, an influence coefficient calculation module, an association matching module, a target adjustment management module, a difference calculation module, and an adjustment management module. Among them, the comparison database management module is used to manage the comparison database, the influence coefficient calculation module is used to calculate the influence coefficient of text elements, the association matching module is used to obtain the legal provisions matched by each paragraph, the target adjustment management module is used to record the target adjustment, the difference calculation module is used to calculate the degree of difference of the text elements corresponding to the legal provisions before and after the target adjustment, and the adjustment management module is used to adjust the order of the legal provisions matched by the second target paragraph.

7. An intelligent text information analysis system based on a large language model according to claim 6, characterized in that: The influence coefficient calculation module includes: a text element acquisition unit, a binary tuple management unit, and an influence coefficient calculation unit. Among them, the text element acquisition unit is used to acquire the text elements in the text information, the binary tuple management unit is used to manage the binary tuples composed of the first text element or the second text element and the corresponding legal provisions, and the first influence coefficient calculation unit is used to calculate the influence coefficient of the second text element of the first text element in the first target paragraph; The association matching module includes: an association matching model management unit and a first sorting unit. Among them, the association matching model management unit is used to manage the legal provision invocation model, and the first sorting unit is used to obtain the initial recommendation sequence of the paragraphs in the target text.

8. The intelligent text information analysis system based on a large language model according to claim 6, characterized in that: The target adjustment management module includes: a target adjustment acquisition unit and an invocation probability comparison unit. Among them, the target adjustment acquisition unit is used to acquire the target adjustment, and the invocation probability comparison unit is used to compare the first target clause and the second target clause of the first target paragraph; The difference calculation module includes: an impact difference calculation unit and an adjustment mode acquisition unit. Among them, the impact difference calculation unit is used to calculate the difference in impact coefficients under different legal clause conditions in the first target paragraph, and the adjustment mode acquisition unit is used to obtain the adjustment mode in the second target paragraph according to the difference in impact coefficients; The adjustment management module includes: a second impact coefficient calculation unit, a pattern matching unit, an adjustment coefficient calculation unit, and a second sorting unit. Among them, the second impact coefficient calculation unit is used to calculate the impact coefficient of the second text element on the first text element in the second target paragraph. The pattern matching unit is used to match the adjustment mode corresponding to each legal clause in the legal clause corresponding to the second target paragraph according to the adjustment mode. The adjustment coefficient calculation unit is used to calculate the adjustment coefficient under each matching mode through an adjustment coefficient calculation function, and the second sorting unit is used to update the recommended sequence of legal articles.

Images