Method and system for automatically generating individualized mediation agreement based on mediation case characteristics

Abstract

The application provides a kind of based on mediation case feature self-generation personalized mediation agreement method and system, it is related to intelligent legal technical field, including: obtaining historical mediation case information;According to historical mediation case information is structured to obtain historical mediation case feature data set;According to historical mediation case feature data set is handled, obtains mediation agreement template library;Obtain the information of mediation case to be mediated and carry out feature extraction processing to obtain input feature data;According to input feature data is carried out pattern recognition and fuzzy inference processing to obtain fuzzy inference result;Based on fuzzy inference result is matched and screened to obtain preliminary mediation agreement;According to input feature data, preliminary mediation agreement is optimized to obtain personalized mediation agreement.The application greatly reduces mediation agreement writing time by dynamically constructing and optimizing mediation agreement template library, and combining the automatic matching and personalized generation method of mediation case features.

Description

Technical Field

[0001] The present invention relates to the field of intelligent legal technology, and in particular to a method and system for automatically generating personalized mediation agreements based on mediation case characteristics. Background Art

[0002] With the continued development of the social economy and the popularization of legal awareness, the number of social disputes has shown a trend of increasing year by year. Mediation, as a non-litigation dispute resolution method, is gaining increasing attention and recognition from all sectors of society due to its rapidity, efficiency, cost-effectiveness, and flexibility. In particular, mediation has gradually become an important solution for a wide range of social issues, including civil and commercial disputes, family conflicts, and labor disputes, playing a positive role in maintaining social harmony and stability. However, in traditional mediation practice, the drafting of mediation agreements primarily relies on the mediator's personal experience and manual writing, which has many limitations.

[0003] First, the manual process of drafting mediation agreements is cumbersome and time-consuming, increasing the mediator's workload and potentially leading to inefficiencies, making it difficult to meet the growing demand for dispute mediation. Second, due to human negligence or insufficient legal expertise, the terms of mediation agreements may be imprecise or incomplete, potentially leading to enforcement disputes and further damaging the rights and interests of the parties. Furthermore, manually drafted mediation agreements are often based on templates, making it difficult to accurately capture the specific characteristics of the case and the individual needs of the parties. This results in a lack of specificity in the agreement content, which in turn affects the success rate of mediation and the effectiveness of the agreement's implementation. Faced with rapidly changing social needs and legal environments, traditional manual drafting methods are clearly unable to meet the complex and ever-changing nature of mediation scenarios. Summary of the Invention

[0004] The present invention aims to provide a method and system for automatically generating personalized mediation agreements based on mediation case characteristics to address the aforementioned issues. To achieve this objective, the present invention employs the following technical solutions:

[0005] In a first aspect, the present application provides a method for automatically generating a personalized mediation agreement based on mediation case characteristics, comprising:

[0006] Obtaining historical mediation case information, including case type, disputed amount, party behavior patterns, case background, mediation needs, and mediation results;

[0007] Performing structural processing on the historical mediation case information, and converting various types of information in the case into a standardized data format, thereby obtaining a historical mediation case feature data set;

[0008] construct a semantic relationship network among case features according to the historical mediation case feature dataset, and process the semantic relationship network by using a machine learning algorithm to obtain a mediation agreement template library;

[0009] acquire mediation case information to be processed, and perform feature extraction processing on the mediation case information to be processed to obtain input feature data;

[0010] perform pattern recognition and fuzzy reasoning processing according to the input feature data, predict the behavior reaction of parties in a case by using a preset sociological model, and perform reasoning analysis on a dispute focus, a behavior pattern and an emotional feature of the case by using a fuzzy reasoning algorithm to obtain a fuzzy reasoning result;

[0011] perform matching processing based on the fuzzy reasoning result, and screen a preliminary mediation agreement from the mediation agreement template library;

[0012] perform optimization processing on the preliminary mediation agreement according to the input feature data, automatically generate a mediation agreement text by using a generator part of a generative adversarial network model, and evaluate the legality and rationality of generated agreement terms by using a discriminator part to obtain an individualized mediation agreement.

[0013] In a second aspect, the present application further provides a system for automatically generating an individualized mediation agreement based on mediation case features, comprising:

[0014] an acquisition module configured to acquire historical mediation case information, wherein the historical mediation case information comprises a case type, a disputed amount, a behavior pattern of parties, a case background, mediation requirements and a mediation result;

[0015] a conversion module configured to perform structural processing on the historical mediation case information, convert various types of information in a case into a standardized data format, and obtain a historical mediation case feature dataset;

[0016] a construction module configured to construct a semantic relationship network among case features according to the historical mediation case feature dataset, and process the semantic relationship network by using a machine learning algorithm to obtain a mediation agreement template library;

[0017] an input module configured to acquire mediation case information to be processed, and perform feature extraction processing on the mediation case information to be processed to obtain input feature data;

[0018] an analysis module configured to perform pattern recognition and fuzzy reasoning processing according to the input feature data, predict the behavior reaction of parties in a case by using a preset sociological model, and perform reasoning analysis on a dispute focus, a behavior pattern and an emotional feature of the case by using a fuzzy reasoning algorithm to obtain a fuzzy reasoning result;

[0019] The matching module performs matching processing based on the fuzzy reasoning result, and filters a preliminary mediation agreement from the mediation agreement template library;

[0020] The output module is configured to optimize the preliminary mediation agreement according to the input feature data, automatically generate a mediation agreement text by using a generator part of a generative adversarial network model, and evaluate the legality and rationality of the generated agreement terms by using a discriminator part, so as to obtain an individualized mediation agreement.

[0021] The present application has the following advantages:

[0022] The present application can greatly reduce the mediation agreement writing time by dynamically constructing and optimizing the mediation agreement template library and combining the automatic matching and individualized generation method according to the mediation case characteristics; and the present application can intelligently generate a mediation agreement with rigorous terms, clear logic and meeting the needs of the parties, and avoid the inaccuracy problem caused by negligence in the traditional manual writing. BRIEF DESCRIPTION OF DRAWINGS

[0023] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments. It should be understood that the following drawings only show some embodiments of the present application, and therefore should not be regarded as a limitation to the scope. For those skilled in the art, other related drawings can also be obtained without creative labor.

[0024] Figure 1 The method flowchart for automatically generating an individualized mediation agreement based on mediation case characteristics in the embodiments of the present application;

[0025] Figure 2 The system structure diagram for automatically generating an individualized mediation agreement based on mediation case characteristics in the embodiments of the present application

[0026] Figure 3 The device structure diagram for automatically generating an individualized mediation agreement based on mediation case characteristics in the embodiments of the present application.

[0027] In the figure, 800 is a device for automatically generating an individualized mediation agreement based on mediation case characteristics; 801 is a processor; 802 is a memory; 803 is a multimedia component; 804 is an I / O interface; 805 is a communication component; 901 is an acquisition module; 902 is a conversion module; 903 is a construction module; 904 is an input module; 905 is an analysis module; 906 is a matching module; and 907 is an output module. DETAILED DESCRIPTION

[0028] In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some but not all of the embodiments of the present application. The components of the embodiments of the present application described and shown in the drawings can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the claimed present application, but only represents selected embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the scope of protection of the present application.

[0029] It should be noted that: similar reference numerals and letters represent similar items in the following drawings, therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings. Meanwhile, in the description of the present application, the terms "first", "second" and the like are only used to distinguish description, and cannot be understood as indicating or implying relative importance.

[0030] Embodiment 1

[0031] The embodiment provides a method for automatically generating a personalized mediation agreement based on mediation case characteristics.

[0032] Referring to Figure 1 , the method includes steps S100 to S700.

[0033] Step S100, obtaining historical mediation case information, the historical mediation case information including case type, dispute amount, party behavior pattern, case background, mediation demand and mediation result;

[0034] It can be understood that the focus of this step is to comprehensively and accurately obtain case information, to ensure the integrity and authenticity of the data. Each type of information has a unique role: the case type is used for classification, the dispute amount provides a numerical basis, the party behavior pattern reveals the dynamic characteristics of the case, the case background and the mediation demand determine the specific characteristics and goals of the case, and the mediation result provides feedback for the optimization of the subsequent model and the dynamic update of the template. The collection of these data is the basis for automatic processing, and also directly affects the reliability and efficiency of the entire system.

[0035] Step S200, structuring processing according to the historical mediation case information, obtaining a historical mediation case feature data set by converting various types of information in the case into a standardized data format;

[0036] It should be noted that in the implementation process, first, different types of data are classified and processed: for text information (such as case background, mediation needs, etc.), natural language processing technology is used to extract key information; for numerical information (such as the amount of dispute), regular expressions or rule matching are used to realize numerical processing; for category information (such as case type, mediation result), classification or coding technology is applied to standardize the representation. Then, through data cleaning, redundant or incorrect information is removed to ensure data consistency.

[0037] Step S300, constructing a semantic relationship network between case features according to a historical mediation case feature data set, and processing the semantic relationship network using a machine learning algorithm to obtain a mediation agreement template library;

[0038] It should be noted that this step uses the construction and analysis of the semantic relationship network to effectively integrate the associated information between the case features, generating a mediation agreement template library with high adaptability and accuracy. This not only improves the intelligent level of template generation, but also ensures that the template library can respond to changes in laws and regulations and the diversification of case needs through a dynamic learning mechanism.

[0039] Step S400, obtaining mediation case information and performing feature extraction processing on the mediation case information to obtain input feature data;

[0040] It can be understood that this step structures and quantifies the information of the new mediation case, extracting core features that are helpful for matching and generating agreements.

[0041] Step S500, performing pattern recognition and fuzzy reasoning processing according to the input feature data, predicting the behavior reaction of the parties in the case through a preset sociological model, and using a fuzzy reasoning algorithm to analyze the dispute focus, behavior pattern and emotional characteristics of the case, obtaining a fuzzy reasoning result;

[0042] It should be noted that this step uses pattern recognition, sociological models and fuzzy reasoning in combination, which can deeply analyze the complex characteristics of the case, convert unstructured or fuzzy information into clear reasoning results, provide scientific basis for accurate matching and generation of agreements, and significantly improve the system's ability to handle complex cases.

[0043] Step S600, based on the fuzzy reasoning result, performing matching processing to select a preliminary mediation agreement from the mediation agreement template library;

[0044] It can be understood that by using the matching mechanism of fuzzy reasoning results and template features, this step can quickly and efficiently select the most suitable mediation agreement template, dynamically adjust the clauses in combination with the individual needs of the case, significantly improve the accuracy and relevance of agreement generation, and reduce the workload of manual participation.

[0045] Step S700, according to the input feature data, the preliminary mediation agreement is optimized, the generator part of the generative adversarial network model is used to automatically generate the mediation agreement text, the discriminator part evaluates the legality and rationality of the generated agreement terms, and the personalized mediation agreement is obtained.

[0046] It can be understood that this step realizes the dynamic optimization and adaptive generation of the mediation agreement through the generative adversarial network, which not only meets the individual needs of the case, but also ensures the legality and rationality of the agreement text.

[0047] Further, step S200 includes step S210 to step S250.

[0048] Step S210, according to the case type, the standardization processing is carried out, the case type is extracted from the case description through text analysis and classification algorithm, and the case category is converted into numerical category data, and the standardized case type data is obtained;

[0049] Preferably, this step uses TF-IDF feature extraction combined with support vector machine (SVM) classification method to realize the standardization processing of case type. Specifically, improvements are made according to the specific characteristics of case text to improve the accuracy and adaptability of classification. Mediation case text has significant legal terminology and structured patterns, such as high-frequency words like "contract dispute" and "both parties agree". These words may be given a lower weight in conventional algorithms. Therefore, by constructing a custom stop word list, high-frequency but low-discriminatory words such as "both" and "agree" are filtered out, and industry-specific glossary is introduced to give higher weight to specific case types (such as "penalty" in "contract dispute" or "compensation amount" in "property dispute"), which can significantly improve the recognition ability of the classification model to key features. In addition, by combining numerical features such as dispute amount and party behavior patterns with text features, a multi-modal feature vector is formed, further enhancing the discriminant ability of the classifier in distinguishing similar but different feature cases. In view of the phenomenon that case categories may exist in cross (such as "contract dispute" and "property dispute"), a multi-label classification framework is adopted, so that the case feature vector is mapped to multiple categories at the same time, solving the problem of fuzzy class boundary, and modeling the relevance between multi-labels to improve the classification effect of complex cases. In model training, by adjusting the regularization parameter of support vector machine, the generalization ability and overfitting risk of the classifier are balanced, and the weight adjustment strategy is introduced for the class imbalance problem to ensure sufficient recognition ability for small category cases. These improvements not only solve the problems of high-frequency interference of legal terms and fuzzy cross-border boundaries, but also significantly improve the robustness, accuracy and practical application value of the mediation case classification model.

[0050] Step S220, numerical processing according to the disputed amount, extracting the disputed amount information in the case through regular expression and converting it into a numerical variable to obtain standardized disputed amount data;

[0051] It should be noted that, for the text data in the mediation case, since the disputed amount often exists in an unstructured form, traditional methods cannot directly use these information for numerical processing. Therefore, this step realizes the automatic extraction and standardization of the amount through regular expression matching and conversion rules. Specifically, first, regular expressions are used to identify the amount information in the text. The regular expression design needs to adapt to the characteristics of the case text and can flexibly match multiple formats, such as: numerical format (e.g. "500 million yuan"), by matching consecutive numbers and combining unit words. Descriptive text (e.g. "one thousand dollars"), by combining natural language processing techniques, the text amount is converted to a numerical form. With currency symbols (e.g. "$1,000"), by extending the matching rules for multiple currency symbols that may be involved in international cases. Secondly, after extracting the amount of text, unit standardization processing is performed. For example, "ten thousand yuan" is converted to the base unit "yuan", and "dollars" or other currencies are converted to a predefined base currency (such as the RMB). This process requires the use of exchange rate conversion tools or rule-based unit mapping tables for processing. Finally, the cleaned and converted amount information is encoded as a numerical variable, and the original amount and converted numerical representation are retained to form standardized disputed amount data. These data are not only used for subsequent classification and matching, but also directly participate in the quantitative analysis of the dispute focus, improving the sensitivity and processing capacity of the algorithm to economic factors. This step automatically extracts and standardizes the disputed amount data, greatly reducing the need for manual intervention, while ensuring the accuracy and consistency of the amount information. In mediation cases, standardized disputed amount data provides accurate numerical input for the model, significantly improving the system's modeling ability for economic characteristics of the case and the quality of subsequent mediation agreement clauses.

[0052] Step S230, sentiment analysis and classification processing according to the behavior pattern of the parties, sentiment analysis and classification of the behavior description of the parties in the case through the BERT algorithm, and label processing combined with the mediation result to obtain structured behavior pattern data;

[0053] In this step, the sentiment analysis and classification technology is used to mine the sentiment tendency and behavior type of the description of the parties' behavior in the case text, such as "actively cooperate with mediation" and "refuse to compromise". Specifically, first, the pre-trained BERT (Bidirectional Encoder Representations from Transformers) model is used to perform semantic encoding on the case description text to generate high-dimensional semantic vector representation. This process can fully capture the context-dependent information, especially suitable for complex behavior expression in mediation case text. For example, for "Party A's attitude is firm and does not accept the other party's conditions", BERT can capture the semantic relationship between "firm" and "accept" and the context, and extract the party's resistance attitude to mediation. Secondly, a classification model (such as a deep learning-based fully connected neural network) is used to analyze the sentiment tendency and behavior classification of the semantic vector. Sentiment tendency analysis distinguishes the emotional polarity of the parties (positive, neutral or negative), and behavior classification divides the behavior pattern of the parties into cooperative, confrontational, neutral and other categories. Combined with the emotion-related adjectives and verbs (such as "angry" and "accept") in the case text, the model can further improve the accuracy of classification. Finally, the results of sentiment analysis and behavior classification are combined with the mediation results for labeling. Through statistical analysis of the mediation results in historical case data, an association model between behavior patterns and mediation success rate is constructed, so as to generate corresponding labels for each behavior pattern. For example, "cooperative type" may be strongly related to "mediation success", while "confrontational type" is related to "mediation failure". These labeled behavior pattern data are stored in a structured form to support subsequent fuzzy reasoning and protocol generation.

[0054] Step S240, according to the case background and mediation needs, feature extraction is performed, the dispute reasons and appeal points in the case background are extracted, and the TF-IDF algorithm is used to analyze the keywords and sentiment distribution of the case description to obtain structured mediation demand data;

[0055] In practical applications, the case description text is first preprocessed, including word segmentation, stop word removal, and part-of-speech tagging. Then, for the disputed reasons and points of contention, descriptive sentence fragments are extracted by constructing key sentence extraction rules (such as sentences containing the words "dispute focus", "core contention", or "main problem"). For example, "Party A believes that Party B has not fulfilled the delivery obligations under the contract" can extract the dispute focus as "not fulfilling the delivery obligations under the contract". Next, the TF-IDF algorithm is applied to calculate the weight of the key words in the case description to capture the core semantic information in the case. Finally, combined with sentiment analysis technology, the extracted key sentences and key words are classified according to their sentiment tendencies to determine the attitudes of the parties in the case towards specific contentions (such as positive, negative, or neutral). For example, "Party B firmly denies responsibility" indicates a negative attitude towards the disputed point, while "Party A is willing to compromise" shows a positive inclination. Through the above steps, structured mediation demand data is generated, which includes disputed reasons, points of contention, and sentiment tendencies. These data provide a clear semantic and emotional basis for the subsequent mediation agreement generation.

[0056] Step S250, integrating the standardized case type data, the standardized disputed amount data, the structured behavior pattern data, and the structured mediation demand data to obtain a historical mediation case feature data set.

[0057] In the integration process, the various feature data need to be aligned in a unified dimension. For this purpose, numerical data (such as disputed amount) can be adjusted to the same scale range through feature standardization or normalization techniques. For example, using min-max normalization to normalize the amount value to the [0, 1] interval, thereby avoiding the weight imbalance caused by the large difference in numerical values between different features. For categorical and textual features (such as case type and mediation demand), they are combined with numerical data into a complete feature vector through feature splicing. The final historical mediation case feature data set is stored in the form of a high-dimensional vector, where each case's feature vector contains complete multi-dimensional information such as case type, disputed amount, behavior pattern, and mediation demand, providing comprehensive data support for subsequent model training, feature correlation analysis, and agreement generation.

[0058] Further, step S300 includes steps S310 to S340.

[0059] Step S310, according to the historical mediation case feature data set, feature correlation is performed by constructing a co-occurrence matrix and applying a word embedding algorithm to vectorize the case features, converting non-numerical data into numerical vectors, obtaining semantic vector representation of the case features;

[0060] Specifically, we first construct a co-occurrence matrix based on the historical mediation case feature dataset. Assume that there is a set of case feature datasets, and set the co-occurrence matrix C to represent the co-occurrence relationship between features. There are N case features (f1, f2, ..., f N ), then the co-occurrence matrix C is an N×N matrix, where each element c ij Represents feature f i and feature f j The co-occurrence frequency in the case dataset. The calculation formula is:

[0061]

[0062] Where M is the number of cases; k is the case number; D k represents the kth case; f i represents the i-th feature; f j represents the jth feature; C ij Represents feature f i and feature f j In case D k The co-occurrence frequency in I(f i ,f j ,D k ) represents the indicator function, if in case D k In the feature f i and f j If they appear at the same time, then I(f i ,f j ,D k )=1, otherwise it is 0.

[0063] Next, we apply word embedding algorithms to vectorize the co-occurrence matrix, with the goal of learning a low-dimensional vector representation of each case feature. Make the features with high co-occurrence frequency have a closer distance in the vector space. Assume that feature f i The word embedding vector is represented as v i , the co-occurrence matrix C is combined with the word embedding algorithm for training in the following way:

[0064]

[0065] in, Represents feature f i The word embedding vector of dimension d; p(C ij |v i ,v j )) represents feature f i and feature f j The co-occurrence probability in the co-occurrence matrix C is modeled using the softmax function:

[0066]

[0067] where V denotes the word embedding matrix of all features, and is the parameter to be optimized during the training process; v k denotes the word embedding vector of feature f k .

[0068] Through training, each non-numeric feature is mapped to a semantic vector space, generating a semantic vector representation of the feature. This vectorization process not only captures the co-occurrence relationship between features, but also learns the potential semantic association of the features through word embedding technology. This semantic vector representation provides high-quality input for subsequent feature clustering and model training.

[0069] Step S320, clustering processing is performed according to the semantic vector representation, and at least two case feature clusters are obtained by aggregating case features with similar semantics;

[0070] It can be understood that this step aggregates scattered feature vectors into feature clusters with consistent semantics through clustering processing. This process reduces the complexity of feature dimensionality and reveals the semantic structure between case features, laying an important foundation for subsequent semantic relationship network construction and mediation agreement template design

[0071] Step S330, network construction processing is performed according to all case feature clusters, and a semantic relationship network is established through a graph neural network to establish semantic connection relationships between features, wherein each node in the semantic relationship network represents a case feature, and each edge represents the association information between features;

[0072] Specifically, first, the nodes in the semantic relationship network represent case features. For example, each node can correspond to a specific case feature such as “contract breach” or “rent arrears”. These nodes come from the feature clusters in the clustering result, and each feature in the feature cluster has similar semantic information. Then, by calculating the similarity within and between feature clusters, the edge weights between nodes are determined. The edge weight calculation can use semantic similarity (such as cosine similarity) or a weighting method based on co-occurrence relationship. The directionality and weight of the edge represent the association strength between features, for example, a high weight may indicate that two features frequently co-occur in case descriptions.

[0073] Then, the semantic relationship network is processed using a graph neural network. The graph neural network updates the feature representation of each node to reflect the semantic information of its neighbor nodes through a message passing mechanism. The node update formula is:

[0074]

[0075] where denotes the feature representation of node u after the t+1 iteration. denotes the neighbor set of node u; d u denotes the number of neighbors of node u; W t denotes the weight matrix of the t-th layer; denotes the activation function, such as the ReLU function; v denotes the neighbor node of node u.

[0076] Through multi-layer iteration, the node feature gradually fuses the information of its neighbor nodes, realizing the deep expression of the semantic association between features. The finally generated semantic relationship network can not only represent the direct connection between case features, but also capture the implicit high-order semantic relationship.

[0077] Step S340, template generation according to the semantic relationship network, processing the semantic relationship network of the case feature by using the convolutional neural network, automatically identifying the key clauses and template structure in the mediation agreement through the training model, generating the standardized mediation agreement template suitable for different case types, and obtaining the final mediation agreement template library.

[0078] Specifically, the feature nodes and edges in the semantic relationship network are converted into tensor input form of graph data. The feature vector of each node contains semantic features, and the edge weight represents the semantic association between nodes. The input data is processed through the graph convolution layer, and the graph convolution operation can capture the local semantic structure between nodes and their neighbors. Through multi-layer graph convolution, the feature representation of each node in the semantic relationship network is further fused and optimized, and finally the global semantic relationship of the key features is captured. Next, through the feature extraction ability of the convolutional neural network, the key clauses of the mediation agreement template are identified. The convolutional neural network uses the processed feature map in the network input to automatically generate a template suggestion containing clause content, clause order and structural relationship. For example, different case types may require specific clauses (such as compensation clauses, behavior commitments), which are identified and extracted through weight learning of the convolutional neural network layer. Finally, the generated template is normalized to form a standardized protocol template library. The template library generates a dynamic structured framework for different types of cases, which can be quickly matched and applied according to the case features.

[0079] Further, step S400 includes steps S410 to S430.

[0080] Step S410, text processing and feature extraction are performed on the mediation case information based on a preset natural language processing model to obtain semantic feature data of the case;

[0081] Specifically, first, the case text information is preprocessed, including word segmentation, removal of stop words, morphological reduction, and uniform formatting operations. Preprocessing ensures the consistency and efficiency of the text data, laying the foundation for subsequent feature extraction. Next, the preprocessed text is subjected to deep semantic analysis through a pre-set natural language processing model. Common models such as the BERT model based on the Transformer architecture generate dynamic word vectors for each word through context-dependent mechanisms. This method can capture the implicit semantic relationships in case descriptions, and through these vectorized semantic features, it can comprehensively represent the key semantic information in the case text. Subsequently, the model uses named entity recognition (NER) to extract specific case entities. These entities are further mapped to fixed feature labels, such as "contract dispute" and "compensation amount," adding structured information to the text analysis. Finally, the extracted semantic features are subjected to dimensionality reduction processing to reduce redundant information and optimize feature representation, generating semantic feature data. This data is stored in the form of high-dimensional vectors, which can comprehensively reflect the text features and semantic connotations of the case.

[0082] Step S420, standardizing the quantitative features in the mediation case information to obtain numerical feature data;

[0083] It can be understood that by normalizing the units and scaling the dimensions of the quantitative features, the influence caused by the dimensional differences between the features is effectively reduced. The standardized data not only improves the stability of the model processing, but also enhances the expressiveness of the features in analysis and prediction, providing a solid foundation for the system to handle cases of different scales and complexity.

[0084] Step S430, integrating the semantic feature data and the numerical feature data to obtain integrated feature data, and using principal component analysis algorithm to reduce the dimension of the integrated feature data to obtain input feature data.

[0085] Further, step S500 includes steps S510 to S530.

[0086] Step S510, performing pattern recognition processing according to the input feature data, using a deep neural network to train the reduced input features, extracting the potential patterns in the data to obtain a pattern recognition result;

[0087] It should be noted that the semantic feature data is usually a high-dimensional vector representing the semantic content in the case text, and the numerical feature data is a low-dimensional numerical variable, such as the amount of dispute and the time of the case. During integration, the two types of features need to be normalized to maintain the balance of the importance of the features, for example, the semantic features and the numerical features are uniformly mapped to the same scale interval (such as [0, 1]). The integrated features are represented as a high-dimensional joint feature vector, which contains both the text semantic information of the case and the numerical feature information. Next, the principal component analysis (PCA) algorithm is used to reduce the dimension of the integrated feature data. The goal of principal component analysis is to project high-dimensional data into a space with lower dimension but retains as much original information as possible through linear transformation. Specifically, principal component analysis calculates the covariance matrix of the feature data, and decomposes its eigenvalues and eigenvectors, and selects the first few principal components with higher cumulative variance proportion as the new feature representation. This not only reduces the redundancy of the features, but also reduces the computational complexity of the model. The input feature data after dimension reduction is a set of low-dimensional vectors, which not only retains the core information of the case features, but also removes noise and redundant features, providing efficient input for subsequent model processing.

[0088] Step S520, according to the mode recognition result, behavior prediction processing is carried out, time series modeling is carried out through the long short-term memory network, the attitude and cooperation willingness of the parties in the case change with time are analyzed, and the future behavior trend is predicted, and the behavior prediction result is obtained;

[0089] Specifically, first, the mode recognition result provides the key features of the case, such as the emotional tendency of the parties (positive, neutral or negative), the behavior type (cooperative or confrontational), and the evolution of the dispute points of the case. These features are represented in the form of time series, forming input data. For example, the description "Party A is more cooperative at the beginning of mediation, but disputes arise over compensation" can be quantified as a time series: [cooperation, cooperation, confrontation]. Next, the long short-term memory network is used to model these time series data. The long short-term memory network is an extension of recurrent neural network (RNN) and is good at processing time series data, which can capture long-distance dependencies between features. For input features such as current behavior features of the parties, the long short-term memory network selectively updates and retains feature memory through the gating mechanism (input gate, forget gate and output gate), thereby generating the hidden state of each time step. This mechanism allows the model to infer the future trend of the behavior of the parties based on historical features. During training, the long short-term memory network minimizes the difference between predicted behavior and real behavior through a loss function, and gradually learns the dynamic change pattern of the case features. For example, through time step prediction, the behavior trend "when the compensation amount dispute continues to expand, the confrontational emotion of the parties rises" can be obtained.

[0090] Step S530, fuzzy reasoning is performed on the behavior prediction result based on a preset fuzzy rule, and an inference result is obtained in combination with a dispute focus in the case and a behavior mode of the parties.

[0091] It can be understood that, based on the variable input (such as the cooperation probability and the confrontation probability) provided by the behavior prediction result, the dispute focus (such as “compensation amount” or “performance mode”) in the case and the behavior mode (such as “cooperative type” or “confrontation type”) of the parties, the input variables are converted into fuzzy sets. Then, according to the preset fuzzy rule base (for example, “if the behavior of the parties is confrontation type and the dispute focus is compensation amount, then the mediation difficulty is high”), the input variables are inferred. The fuzzy reasoning includes steps such as fuzzification, rule evaluation, inference calculation and defuzzification. The activation strength of the rule is calculated through the fuzzy set calculation, and the inference output is generated. Finally, the fuzzy result is interpreted as a clear mediation strategy suggestion or mediation difficulty evaluation. This step flexibly handles the uncertainty and complex characteristics in the case through fuzzy reasoning, and obtains key conclusions such as the priority of the dispute focus and the success possibility of mediation.

[0092] Further, step S600 includes step S610 to step S640.

[0093] Step S610, according to the fuzzy reasoning result, the weighting processing of the case characteristics is performed, the weighted case characteristic vector is obtained by weighting various fuzzy characteristics in the case and combining the inference confidence corresponding to the fuzzy characteristics;

[0094] This step combines the fuzzy reasoning result to perform weighting processing on the case characteristics, and enhances the accuracy of case analysis. The weighted characteristic vector not only comprehensively reflects the importance of the semantic characteristics and the dispute focus of the case, but also effectively improves the analysis ability of the system to the complex information of the case.

[0095] Step S620, similarity calculation is performed according to the weighted case characteristic vector, similarity score results are obtained by calculating the similarity between the case characteristic vector and the template characteristic vector, and the similarity score results include the matching degree of each mediation agreement template and the current case;

[0096] It should be noted that, in this step, the similarity calculation of the characteristic vector quantifies the semantic consistency between the case and the template, providing an intuitive and operable basis for template selection. This process combines the priority of the weighted characteristics, ensures the dominant role of the key characteristics on the matching degree, and effectively improves the accuracy and pertinence of the system in template recommendation

[0097] Step S630, according to the similarity score results, the templates in the mediation agreement template library are sorted, and the optimal mediation agreement template is obtained based on the sorting results;

[0098] Specifically, similarity score results are first received, which indicate the degree of adaptation of the current case to each template in the template library. The score results are presented in numerical form, with higher scores indicating that the template is closer to the characteristics of the case. According to the scores, the templates are sorted in descending order, generating an ordered template list. Then, the template with the highest matching degree is selected according to the sorting results. Further, additional constraints are combined in the selection process, such as the legal applicability of the template, regional characteristics or historical usage. If the case involves multiple points of contention, the system can also integrate the similarity of multiple feature dimensions and use a weighted scoring method to comprehensively sort the templates, ensuring that the selected template can fully adapt to the needs of the case. Finally, the highest ranked template is selected as the optimal mediation agreement template for the current case, for use in subsequent steps. This template not only makes full use of the relevance of historical data, but also dynamically adapts to the individual characteristics of the case.

[0099] Step S640, according to the optimal mediation agreement template, an agreement generation process is performed, and the clause content in the template is adjusted based on the specific needs of the case to generate a preliminary mediation agreement.

[0100] It should be noted that this step first loads the selected optimal template, which contains general mediation agreement clauses and framework structures. The optimal template provides the basic skeleton of the agreement, but still needs to be further adjusted according to the actual needs of the case. For example, the template may have preset content for compensation clauses, performance clauses or confidentiality clauses, which need to be refined or modified according to the dispute focus of the case or the wishes of the parties. Then, the specific content of the template clauses is automatically adjusted according to the characteristics of the case, such as the dispute amount, the behavior pattern of the parties and the mediation needs. For example, for cases with a large dispute amount, the compensation clause in the template may need a more detailed installment payment plan; and for cases where the parties tend to cooperate, flexibility provisions can be added to the performance clause. This dynamic adjustment process can be achieved through a rule engine or data-driven algorithm, which maps the characteristics of the case to the modification logic of the template clauses. Finally, a preliminary mediation agreement is generated based on the adjusted template clauses, and the agreement text retains legal rigor while fully reflecting the specific needs of the case and the intentions of the parties.

[0101] Further, step S700 includes steps S710 to S730.

[0102] Step S710, according to the input feature data, a content generation process is performed, and based on a pre-set generative adversarial network, the semantic information of the characteristics of the case is captured to individually adjust the clauses of the preliminary agreement, generating a mediation agreement draft;

[0103] First, input feature data (e.g., case type, dispute amount, party behavior pattern, and mediation needs) are input as the input of the generative network. After preprocessing, these features are mapped to semantic vector representations to guide the generation process. The generative adversarial network includes two main parts: the generator and the discriminator. The generator receives the input features and generates the content of the clauses, including compensation clauses, performance clauses, etc., through a deep neural network, ensuring that the generated agreement content meets the basic characteristics and needs of the case.

[0104] Next, the discriminator evaluates the clauses generated by the generator to determine whether they are reasonable and reasonable. The discriminator receives the generated draft agreement and compares it with the actual historical agreement clauses, learns the semantic and logical differences between the two, and continuously adjusts the output of the generator. In the adversarial training process of generation and discrimination, the generator gradually improves the quality and authenticity of the generated clauses, and the discriminator helps to exclude clauses that do not meet the actual needs. In this process, personalized adjustments are mainly reflected in semantic consistency, flexibility, and situational relevance. Semantic consistency means that the generator ensures that the clause content is consistent with the semantic characteristics of the case based on the input features. For example, a case with a large dispute amount generates more rigorous compensation clauses. Flexibility means that the generated clauses can automatically adjust the wording according to the party behavior pattern (e.g., cooperative or confrontational) to adapt to the wishes of the parties. Situational relevance is reflected in dynamically adding or adjusting special clauses according to the case type, such as adding a confidentiality agreement in intellectual property cases. Finally, the generated mediation agreement draft contains a complete clause structure, and the content has been optimized to match the characteristics of the case, providing a high-quality basis for subsequent legality and reasonableness evaluation.

[0105] Step S720, according to the mediation agreement draft, the generated protocol clauses are evaluated using the discriminant network of the generative adversarial network to determine whether the protocol clauses are legal and reasonable, and based on pre-set legal rules, industry standards and historical cases, the protocol clauses are checked to determine whether the protocol meets the legal requirements and is reasonable, and a legality-reasonability evaluation result is obtained;

[0106] It can be understood that this step first generates a discriminator network in the generative adversarial network, which is trained as an evaluator to receive the content of the clauses of the mediation agreement draft, and compare it with the legal agreement clauses in the historical cases in terms of semantics and structure. The output value of the discriminator network represents the probability that the agreement clauses conform to the historical agreement mode, which is used as a preliminary basis for judging the legality and reasonableness of the agreement. For example, if the language structure of the clauses is consistent with the successfully executed agreements in history, a higher credibility score is given. Then the agreement clauses are checked by pre-set legal rules and industry standards. This process includes matching the key elements in the clauses (such as the upper limit of the compensation amount, the performance period, the dispute resolution method, etc.) with the compliance of relevant legal provisions. For example, for compensation clauses with large dispute amounts, the system checks whether there are enforcement clauses and whether they comply with the legal provisions of the compensation range. At the same time, industry standards can provide domain-specific supplementary rules, such as whether appropriate overtime pay compensation clauses are included in labor dispute agreements. Finally, the reasonableness of the agreement clauses is judged in combination with the analysis of historical cases. This includes a balance check of the wording of the clauses, such as whether it is too biased towards the interests of one party, whether there are potential difficulties in execution, etc. Reasonableness assessment not only focuses on legal compliance, but also considers fairness and operability in actual execution.

[0107] Step S730, feedback adjustment according to the legality-reasonableness evaluation result, adjusting the generation strategy of the generative adversarial network through the reinforcement learning algorithm, and outputting the final personalized mediation agreement.

[0108] In practical application, first, the legality-reasonableness evaluation result output is received, which contains the feedback score of the agreement clauses in terms of legal compliance and actual reasonableness. The reinforcement learning algorithm uses this evaluation result as a reward signal to guide the optimization process of the generative network. Specifically, the generative network is regarded as an "agent" in reinforcement learning, and its goal is to generate an agreement clause that obtains higher rewards in the discriminator network and rule evaluation. The reward function is designed according to the legality and reasonableness score of the clauses, for example: clauses with high legality weight are given positive rewards; clauses with legal conflicts or logical loopholes are punished.

[0109] Then the generative network uses policy gradient algorithm or Q-learning algorithm to update the generation strategy. The generative network adjusts the parameters of its generation model through iterative training to gradually improve the quality of the generated content. For example, by learning the characteristics of clauses in specific types of cases, such as the legal limits of compensation amounts in labor dispute agreements, or the general wording of performance clauses in contract dispute agreements. After multiple rounds of feedback optimization, a fully adjusted and more personalized mediation agreement text can be generated. The final output mediation agreement not only fully complies with the legal requirements of the case, but also accurately reflects the unique needs of the case and the wishes of the parties.

[0110] Embodiment 2

[0111] As Figure 2 shown, the embodiment provides a system for automatically generating a personalized mediation agreement based on mediation case characteristics, the system comprising:

[0112] The acquisition module 901 is configured to acquire historical mediation case information, the historical mediation case information including case type, dispute amount, party behavior pattern, case background, mediation demand and mediation result.

[0113] The conversion module 902 is configured to perform structured processing according to the historical mediation case information, to convert various types of information in the case into a standardized data format, and to obtain a historical mediation case feature data set.

[0114] The construction module 903 is configured to construct a semantic relationship network between case features according to the historical mediation case feature data set, and to process the semantic relationship network using a machine learning algorithm to obtain a mediation agreement template library.

[0115] The input module 904 is configured to acquire mediation case information and perform feature extraction processing on the mediation case information to obtain input feature data.

[0116] The analysis module 905 is configured to perform pattern recognition and fuzzy reasoning processing according to the input feature data, to predict the behavior of the parties in the case using a pre-set sociological model, and to perform reasoning analysis on the dispute focus, behavior pattern and emotional characteristics of the case using a fuzzy reasoning algorithm to obtain a fuzzy reasoning result.

[0117] The matching module 906 is configured to perform matching processing based on the fuzzy reasoning result to filter a preliminary mediation agreement from the mediation agreement template library.

[0118] The output module 907 is configured to optimize the preliminary mediation agreement according to the input feature data, to automatically generate a mediation agreement text using the generator part of a generative adversarial network model, and to evaluate the legality and rationality of the generated agreement terms using the discriminator part to obtain a personalized mediation agreement.

[0119] In one specific embodiment of the present disclosure, the conversion module 902 comprises:

[0120] The first processing unit is configured to perform standardized processing according to the case type, to extract the case type from the case description using a text analysis and classification algorithm, and to convert the case type into numerical category data to obtain standardized case type data.

[0121] The second processing unit is configured to perform numerical processing according to the dispute amount, to extract the dispute amount information in the case using a regular expression and to convert it into a numerical variable to obtain standardized dispute amount data.

[0122] The first analysis unit is used for emotional analysis and classification processing according to the behavior mode of the parties, performs emotional tendency analysis and classification on the behavior description of the parties in the case through the BERT algorithm, and performs label processing in combination with the mediation result to obtain structured behavior mode data;

[0123] The first extraction unit is used for feature extraction according to the case background and mediation demand, extracts the dispute reason and appeal point in the case background, and analyzes the keywords and emotional distribution of the case description by using the TF-IDF algorithm to obtain structured mediation demand data;

[0124] The first integration unit is used for integrating the standardized case type data, the standardized disputed amount data, the structured behavior mode data and the structured mediation demand data to obtain the historical mediation case feature data set.

[0125] In one specific embodiment disclosed in the present application, the construction module 903 comprises:

[0126] The first association unit is used for feature association according to the historical mediation case feature data set, performs vectorization processing on the case features by constructing a co-occurrence matrix and applying a word embedding algorithm to convert non-numerical data into numerical vectors to obtain semantic vector representation of the case features;

[0127] The first clustering unit is used for clustering processing according to the semantic vector representation, aggregates case features with similar semantics to obtain at least two case feature clusters;

[0128] The first construction unit is used for network construction processing according to all case feature clusters, establishes semantic connection relationships between features by a graph neural network to obtain a semantic relationship network, each node in the semantic relationship network represents a case feature, and each edge represents association information between features;

[0129] The first generation unit is used for template generation according to the semantic relationship network, processes the semantic relationship network of the case features by using a convolutional neural network, automatically identifies key clauses and template structures in the mediation agreement by training a model, generates standardized mediation agreement templates suitable for different case types, and obtains a final mediation agreement template library.

[0130] Embodiment 3:

[0131] Corresponding to the method embodiments above, the present embodiment also provides a device for automatically generating a personalized mediation agreement based on mediation case features, which can be mutually referred to as described below and as described above.

[0132] Figure 3 FIG. 8 is a block diagram of a device 800 for automatically generating a personalized mediation agreement based on mediation case features according to an exemplary embodiment. As shown in Figure 3 FIG. 8, the device 800 for automatically generating a personalized mediation agreement based on mediation case features can include a processor 801, a memory 802. The device 800 for automatically generating a personalized mediation agreement based on mediation case features can also include one or more of a multimedia component 803, an I / O interface 804, and a communication component 805.

[0133] The processor 801 is configured to control overall operations of the device 800 for generating a personalized mediation agreement based on mediation case characteristics, so as to complete all or part of the steps in the method for generating a personalized mediation agreement based on mediation case characteristics. The memory 802 is configured to store various types of data to support the operations of the device 800 for generating a personalized mediation agreement based on mediation case characteristics. For example, the data can include instructions for any application or method operating on the device 800 for generating a personalized mediation agreement based on mediation case characteristics, and application-related data, such as contact data, sent and received messages, pictures, audio, video, and the like. The memory 802 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk or optical disk. The multimedia component 803 can include a screen and an audio component. The screen can be a touch screen, for example, and the audio component is configured to output and / or input audio signals. For example, the audio component can include a microphone configured to receive external audio signals. The received audio signals can be further stored in the memory 802 or transmitted through the communication component 805. The audio component also includes at least one speaker configured to output audio signals. The I / O interface 804 provides an interface between the processor 801 and other interface modules, which can be a keyboard, a mouse, a button, and the like. The buttons can be virtual buttons or physical buttons. The communication component 805 is configured to perform wired or wireless communication between the device 800 for generating a personalized mediation agreement based on mediation case characteristics and other devices. The wireless communication, such as Wi-Fi, Bluetooth, near field communication (NFC), 2G, 3G or 4G, or a combination of one or more of them, so the corresponding communication component 805 can include a Wi-Fi module, a Bluetooth module, an NFC module.

[0134] In an exemplary embodiment, the device 800 for automatically generating a personalized mediation agreement based on mediation case characteristics can be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors or other electronic components to execute the above-mentioned method of automatically generating a personalized mediation agreement based on mediation case characteristics.

[0135] In another exemplary embodiment, a computer-readable storage medium including program instructions is further provided. When executed by a processor, the program instructions implement the steps of the aforementioned method for automatically generating a personalized mediation agreement based on mediation case characteristics. For example, the computer-readable storage medium may be the aforementioned memory 802 including the program instructions. The program instructions may be executed by the processor 801 of the device 800 for automatically generating a personalized mediation agreement based on mediation case characteristics to implement the aforementioned method for automatically generating a personalized mediation agreement based on mediation case characteristics.

[0136] The above description is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with this technical field can easily think of changes or replacements within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A method for automatically generating a personalized mediation agreement based on mediation case characteristics, characterized in that: include: Obtaining historical mediation case information, including case type, disputed amount, party behavior patterns, case background, mediation needs, and mediation results; Performing structural processing on the historical mediation case information, and converting various types of information in the case into a standardized data format, thereby obtaining a historical mediation case feature data set; Constructing a semantic relationship network between case features based on the historical mediation case feature dataset, and processing the semantic relationship network using a machine learning algorithm to obtain a mediation agreement template library; Acquiring information about cases to be mediated, and performing feature extraction processing on the information about the cases to be mediated to obtain input feature data; Performing pattern recognition and fuzzy reasoning processing based on the input feature data, predicting the behavioral responses of the parties in the case through a preset sociological model, and using a fuzzy reasoning algorithm to reason and analyze the dispute focus, behavioral patterns, and emotional characteristics of the case to obtain fuzzy reasoning results; Performing matching processing based on the fuzzy reasoning result, and screening the mediation agreement template library to obtain a preliminary mediation agreement; The preliminary mediation agreement is optimized according to the input feature data, and the mediation agreement text is automatically generated by utilizing the generator part of the generative adversarial network model. The discriminator part evaluates the legality and rationality of the generated mediation agreement text to obtain a personalized mediation agreement.

2. The method for automatically generating a personalized mediation agreement based on mediation case characteristics according to claim 1 is characterized in that: Based on the structured processing of the historical mediation case information, various types of information in the case are converted into a standardized data format to obtain a historical mediation case feature dataset, including: Standardization is performed based on the case type, extracting the case type from the case description through text analysis and classification algorithms and converting the case type into numerical data to obtain standardized case type data; Numerical processing is performed based on the disputed amount. The disputed amount information in the case is extracted using regular expressions and converted into numerical variables to obtain standardized disputed amount data. Conduct sentiment analysis and classification based on the parties' behavior patterns. Use the BERT algorithm to analyze and classify the emotional tendencies of the parties' behavior descriptions in the case, and label them in combination with the mediation results to obtain structured behavior pattern data. Feature extraction is performed based on the case background and mediation needs. By extracting the causes of dispute and appeals from the case background and using the TF-IDF algorithm to analyze the keywords and sentiment distribution of the case description, structured mediation demand data is obtained. The standardized case type data, the standardized dispute amount data, the structured behavior pattern data, and the structured mediation demand data are integrated to obtain a historical mediation case feature data set.

3. The method for automatically generating a personalized mediation agreement based on mediation case characteristics according to claim 1 is characterized in that: A semantic relationship network between case features is constructed based on the historical mediation case feature dataset, and the semantic relationship network is processed using a machine learning algorithm to obtain a mediation agreement template library, including: Based on the historical mediation case feature dataset, we perform feature association, construct a co-occurrence matrix, and apply a word embedding algorithm to vectorize the case features, converting non-numeric data into numeric vectors to obtain a semantic vector representation of the case features. Performing clustering processing based on the semantic vector representation to obtain at least two case feature clusters by aggregating case features with similar semantics; Performing network construction based on all the case feature clusters, establishing semantic connection relationships between features through a graph neural network, and obtaining a semantic relationship network, wherein each node in the semantic relationship network represents a case feature, and each edge represents association information between features; Templates are generated based on the semantic relationship network, and the semantic relationship network of case features is processed using a convolutional neural network. The key clauses and template structure in the mediation agreement are automatically identified through the training model, and standardized mediation agreement templates suitable for different case types are generated to obtain the final mediation agreement template library.

4. The method for automatically generating a personalized mediation agreement based on mediation case characteristics according to claim 1 is characterized in that: Acquiring information about cases to be mediated, and performing feature extraction processing on the information about cases to be mediated to obtain input feature data, including: Performing text processing and feature extraction on the case information to be mediated based on a preset natural language processing model to obtain semantic feature data of the case; Performing standardization processing on the quantitative features in the information of the case to be mediated to obtain numerical feature data; The semantic feature data and the numerical feature data are integrated to obtain integrated feature data, and a principal component analysis algorithm is used to perform dimensionality reduction processing on the integrated feature data to obtain input feature data.

5. The method for automatically generating a personalized mediation agreement based on mediation case characteristics according to claim 1 is characterized in that: Pattern recognition and fuzzy reasoning are performed based on the input feature data. The behavioral responses of the parties in the case are predicted using a preset sociological model. Fuzzy reasoning algorithms are used to analyze the dispute focus, behavioral patterns, and emotional characteristics of the case to obtain fuzzy reasoning results, including: Performing pattern recognition processing on the input feature data, using a deep neural network to train the reduced-dimensional input features, extracting potential patterns in the data to obtain pattern recognition results, which include the emotional tendencies and behavioral types of the parties involved, as well as the evolution of controversial points in the case; Conducting behavior prediction based on the pattern recognition results, performing time series modeling through a long short-term memory network, analyzing how the attitudes and willingness to cooperate of the parties involved in the case change over time, and predicting future behavior trends to obtain behavior prediction results; Perform fuzzy reasoning on the behavior prediction results based on preset fuzzy rules, and obtain reasoning results based on the dispute focus and behavior patterns of the parties in the case; The fuzzy reasoning of the behavior prediction result is performed based on preset fuzzy rules, including: Based on the variable inputs provided by the behavior prediction results, combined with the dispute focus of the case and the behavior patterns of the parties, these input variables are converted into fuzzy sets, and then the input variables are reasoned according to the preset fuzzy rule base. Fuzzy reasoning includes fuzzification, rule evaluation, reasoning calculation and defuzzification steps. The activation strength of the rules is calculated through fuzzy sets and the reasoning output is generated. Finally, the fuzzy results are interpreted as clear mediation strategy recommendations or mediation difficulty assessments.

6. The method for automatically generating a personalized mediation agreement based on mediation case characteristics according to claim 1 is characterized in that: Matching is performed based on the fuzzy reasoning result, and a preliminary mediation agreement is obtained by screening from the mediation agreement template library, including: Performing weighted processing on the case features according to the fuzzy reasoning results, by weighting various fuzzy features in the case and combining the reasoning confidence corresponding to the fuzzy features to obtain a weighted case feature vector; Performing similarity calculation based on the weighted case feature vectors, and obtaining a similarity score result by calculating the similarity between the case feature vector and the template feature vector, wherein the similarity score result includes a matching degree between each mediation agreement template and the current case; Sorting the templates in the mediation agreement template library according to the similarity scoring results, and screening the optimal mediation agreement template based on the sorting results; An agreement is generated based on the optimal mediation agreement template, and the terms and conditions in the template are adjusted based on the specific needs of the case to generate a preliminary mediation agreement.

7. The method for automatically generating a personalized mediation agreement based on mediation case characteristics according to claim 1 is characterized in that: The preliminary mediation agreement is optimized based on the input feature data, and a mediation agreement text is automatically generated by using the generator part of the generative adversarial network model. The discriminator part evaluates the legality and rationality of the generated mediation agreement text to obtain a personalized mediation agreement, including: performing content generation processing based on the input feature data, capturing semantic information of case features based on a preset generative adversarial network, and personalizing the terms of the preliminary mediation agreement to generate a mediation agreement text; Performing an evaluation process based on the mediation agreement text, using the discriminant network of the generative adversarial network to judge the legality and rationality of the generated mediation agreement text, checking the terms of the agreement based on preset legal rules, industry standards, and historical cases, and judging whether the agreement meets the legality requirements and is reasonable, thereby obtaining a legality-rationality evaluation result; Feedback adjustment is performed based on the legality-rationality evaluation results, the generation strategy of the generative adversarial network is adjusted through a reinforcement learning algorithm, and the final personalized mediation agreement is output.

8. A system for automatically generating personalized mediation agreements based on mediation case characteristics, characterized in that: include: An acquisition module is used to obtain historical mediation case information, including case type, dispute amount, party behavior pattern, case background, mediation needs, and mediation results; A conversion module is used to perform structured processing based on the historical mediation case information, and obtain a historical mediation case feature data set by converting various types of information in the case into a standardized data format; A construction module is used to construct a semantic relationship network between case features based on the historical mediation case feature data set, and process the semantic relationship network using a machine learning algorithm to obtain a mediation agreement template library; An input module is used to obtain information about cases to be mediated, and perform feature extraction processing on the information about cases to be mediated to obtain input feature data; An analysis module is used to perform pattern recognition and fuzzy reasoning based on the input feature data, predict the behavioral responses of the parties in the case through a preset sociological model, and use a fuzzy reasoning algorithm to perform reasoning and analysis on the dispute focus, behavioral patterns, and emotional characteristics of the case to obtain fuzzy reasoning results; a matching module, performing matching processing based on the fuzzy reasoning result, and screening and obtaining a preliminary mediation agreement from the mediation agreement template library; The output module is used to optimize the preliminary mediation agreement based on the input feature data, automatically generate the mediation agreement text by utilizing the generator part of the generative adversarial network model, and the discriminator part evaluates the legality and rationality of the generated mediation agreement text to obtain a personalized mediation agreement.

9. The system for automatically generating personalized mediation agreements based on mediation case characteristics according to claim 8 is characterized in that: The conversion module comprises: A first processing unit is configured to perform standardization processing according to the case type, extract the case type from the case description through text analysis and classification algorithms, and convert the case type into numerical data to obtain standardized case type data; The second processing unit is used to perform numerical processing based on the dispute amount, extract the dispute amount information in the case through regular expressions and convert it into a numerical variable to obtain standardized dispute amount data; The first analysis unit is used to perform sentiment analysis and classification based on the behavior patterns of the parties involved. The BERT algorithm is used to analyze and classify the sentiment tendency of the descriptions of the behavior of the parties involved in the case, and the mediation results are combined for labeling to obtain structured behavior pattern data. The first extraction unit is used to extract features based on the case background and mediation needs, extract the causes of dispute and appeal points in the case background, and use the TF-IDF algorithm to analyze the keywords and sentiment distribution of the case description to obtain structured mediation demand data; The first integration unit is used to integrate the standardized case type data, the standardized dispute amount data, the structured behavior pattern data and the structured mediation demand data to obtain a historical mediation case feature data set.

10. The system for automatically generating personalized mediation agreements based on mediation case characteristics according to claim 8 is characterized in that: The building blocks include: The first association unit is used to associate features based on the historical mediation case feature dataset. By constructing a co-occurrence matrix and applying a word embedding algorithm to vectorize the case features, non-numeric data is converted into numeric vectors to obtain a semantic vector representation of the case features. a first clustering unit, configured to perform clustering processing based on the semantic vector representation, and obtain at least two case feature clusters by aggregating case features with similar semantics; A first construction unit is configured to perform network construction processing based on all the case feature clusters, establish semantic connection relationships between features through a graph neural network, and obtain a semantic relationship network, wherein each node in the semantic relationship network represents a case feature, and each edge represents association information between features; The first generation unit is used to generate templates based on the semantic relationship network, use a convolutional neural network to process the semantic relationship network of case features, automatically identify key clauses and template structures in the mediation agreement through a training model, generate standardized mediation agreement templates suitable for different case types, and obtain a final mediation agreement template library.

Images