AI mediation method based on large data model

Abstract

The invention discloses an AI mediation method based on a large data model, and the method comprises an interaction system which is used for receiving the multi-modal information of a party, and feeding back the multi-modal dialogue information outputted by a mediation engine system to the party, so as to complete the multi-modal dialogue interaction between an AI mediator and the party; the mediation engine system is used for initiating one or more rounds of AI mediation of one or more AI mediators, and generating multi-modal dialogue information by combining a multi-modal mediation template of the mediation large model system and a context of a dialogue in the interaction system, so as to send the multi-modal dialogue information to the interaction system; and the mediation large model system is used for generating the multi-modal mediation template according to the multi-modal mediation data and the multi-dimensional intelligent labeling result of the AI mediator so as to send the multi-modal mediation template to the mediation engine system. According to the scheme, the mediation efficiency can be improved, and the mediation quality can be enhanced.

Description

[0001] This application is a divisional application of application number CN202411286603.2, entitled "An AI Mediation System Based on a Large Data Model," with the parent application filed on September 13, 2024. Technical Field

[0002] This application relates to the field of artificial intelligence, and in particular to an AI mediation method based on a large data model. Background Technology

[0003] Mediation plays a vital role in the modern judicial system, aiming to resolve disputes through non-litigation means, reduce the burden on courts, and improve judicial efficiency. With the continuous development and increasing complexity of society, the number of cases has increased dramatically. Traditional litigation methods are time-consuming, labor-intensive, and costly. Mediation offers an efficient and economical alternative, quickly resolving disputes, reducing the time and financial burden on parties involved, and promoting social harmony. Existing mediation systems primarily rely on the experience and judgment of human mediators. Mediators are typically professionals with legal knowledge and mediation skills who facilitate agreements between parties by communicating with them, analyzing the case, and proposing solutions. While effective to some extent, this approach also has significant drawbacks: First, the quality and ability of mediators vary, making it difficult to guarantee the quality of mediation. Second, human mediation is inefficient, especially with a large volume of cases, which mediators struggle to handle. Finally, human mediation is susceptible to subjective influences, potentially leading to unjust outcomes. At the same time, there is a large backlog of various disputes and a rapid increase in cases. Judicial resources are insufficient, and the courts face problems such as difficulty in reminding, collecting debts, and litigating. For example, due to high time costs, mediation organizations and mediators represented by lawyers lack motivation to participate and are inefficient. Grassroots people's courts are overwhelmed by the large number of cases and the limited number of staff.

[0004] Currently, artificial intelligence technologies based on commercial large-scale models, such as the judicial large-scale model and the Pangu large-scale model, have been widely utilized in many fields, driving the rapid development of the industry. However, in the field of mediation, it remains largely unexplored. With the rapid development of the internet, social conflicts and disputes in the digital age exhibit the characteristics of being "numerous, fast, widespread, complex, and difficult." This brings the following problems to the mediation industry: Mediation lacks standardization and industry standards have not been established. The number of mediators is limited, and their experience is also limited, leading to insufficient professionalism. The number of mediated disputes is large and growing rapidly, making timely and effective resolution difficult. Even the most basic intelligent outbound calling can no longer meet the needs of mediation.

[0005] Therefore, there is an urgent need for a technical solution that can improve mediation efficiency and enhance mediation quality. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this application provides an AI-based mediation method based on a large data model. This application solves the problems of insufficient number of mediators and long training periods.

[0007] This application provides an AI mediation method based on a large data model, employing an AI mediation system, which includes an interaction system, a mediation engine system, and a large mediation model system. The interactive system includes an interactive subsystem, a digital human subsystem, and a human mediation subsystem. The interactive subsystem is used to respond to the multimodal information of the parties involved, so as to drive the AI ​​mediator to conduct multimodal dialogue and interaction with the parties involved. The digital human subsystem is used to receive the multimodal dialogue information from the interactive subsystem, so as to conduct multimodal dialogue and interaction with the parties involved. The human mediation subsystem is used to receive the instruction to switch from AI mediation to human mediation, so as to assign the mediation case to the corresponding human mediator. The mediation engine system includes: a mediation initiation subsystem, a mediation dialogue subsystem, an intelligent annotation subsystem, and an intelligent monitoring subsystem. The mediation initiation subsystem imports cases into the system and selects a target AI mediator to transmit AI mediation parameters to the mediation dialogue subsystem. The mediation dialogue subsystem generates multimodal dialogue information based on the AI ​​mediation parameters from the mediation initiation subsystem, the multimodal mediation template from the mediation library subsystem in the large-scale mediation model system, the multidimensional intelligent annotation from the intelligent annotation subsystem, and the context of the current mediation dialogue. This information is then transmitted to the interaction subsystem within the interaction system. The intelligent annotation subsystem receives text, audio, and video text elements from the interaction subsystem in the interaction system and generates corresponding data tags. The results of the multidimensional intelligent annotation are then transmitted to the mediation dialogue subsystem and the mediation library subsystem in the large-scale mediation model system, respectively. The intelligent monitoring subsystem monitors the manual mediation process and checks the mediation results. The mediation big data model system includes: a mediation library subsystem, a mediation training subsystem, a mediation modeling subsystem, and a third-party resource library subsystem. The mediation library subsystem is used to construct core mediation data based on the outputs of one or more subsystems. The mediation training subsystem is used to optimize multimodal mediation templates using AI strategy feature sets and target machine learning models. The mediation modeling subsystem is used to integrate target algorithms through a multi-core algorithm structure and generate multimodal mediation algorithm models for use by the mediation training subsystem. The third-party resource library subsystem is used to acquire the target database and the target big data model. Cases requiring mediation are imported into the AI ​​mediation system via API or files. The AI ​​mediation system analyzes the cases and invokes an AI mediation robot trained on a large-scale mediation model. Utilizing ICT, AR / VR, big data, and blockchain technologies, the AI ​​mediation robot engages in mediation with the parties based on the content of the mediation application. The AI ​​mediation robot reaches the parties via phone, SMS, video, AR / VR, and digital human. It creates a profile of the parties, analyzes their communication intentions, and engages in continuous, interactive, and layered mediation communication and negotiation, guiding the parties to reach a mediation agreement, sign the legal documents (mediation agreement / judicial confirmation application), and fulfill the legal obligations stipulated in the mediation agreement. Mediation supervisors monitor the AI ​​mediator's operations in real time, identifying business problems during the mediation process, adjusting mediation strategies, and maintaining the stable operation of the mediation robot.

[0008] Preferably, the execution steps of the interactive subsystem include: The system collects and extracts audio information from the parties involved, converts continuous analog audio signals into discrete digital signals, performs frame segmentation, extracts MFCC features, finds the optimal probability path, uses an acoustic model to identify phonemes, and uses a vocabulary model and a language model to identify words and sentences. The system identifies and organizes the audio signals of the parties involved during the mediation process by the mediator and mediation assistant into short audio segments and corresponding text element information, and transmits them to the AI ​​intelligent annotation subsystem. The system collects and extracts video information from the parties involved, using computer vision technology to detect the position and contour of faces; it uses deep learning models to identify key points on faces, and machine learning models, convolutional neural networks, and recurrent neural networks to identify micro-expressions and movements to form video element slices. Local video element slices include micro-movement videos of the eyes, ears, mouth, nose, eyebrows, hair, wrinkles, cheeks, and facial muscles, while overall video element slices include videos of movements of the head, shoulders, neck, hands, elbows, arms, torso, hips, and limbs. The system identifies and organizes the emotional text elements of the parties involved from the local and overall video element slices: happiness, surprise, fear, anger, sadness, nausea, contempt, and interest. The identified emotions are then transmitted to the AI ​​intelligent annotation subsystem to form video text element information. The system collects and extracts real-time messages from the parties involved. It analyzes the text, punctuation marks, and emoticons sent by the parties in the real-time messages, identifies online slang, analyzes the context, and organizes the text elements of the parties involved. The identified and organized text elements of the parties involved are then transmitted to the AI ​​intelligent annotation subsystem. The system receives multimodal dialogues from the AI ​​mediation dialogue subsystem, invokes the AI ​​digital human subsystem to complete a mediation dialogue with the parties involved, and terminates the AI ​​mediation session if the parties do not wish to mediate or explicitly require human mediation. The AI ​​mediation record is then transferred to the AI ​​human mediation subsystem, which then transfers the mediation to a human mediator. Based on the parties' original text, voice, and video records, as well as the identified and organized text, audio, and video elements, multimodal mediation data is generated and transmitted to the AI ​​mediation library subsystem of the AI ​​mediation big model system for multimodal storage. The text elements include: keywords indicating legal knowledge (knowledge of the law, lack of knowledge of the law); keywords indicating the parties' living conditions (income, work, food, clothing, housing, transportation, consumption, family, social activities); keywords indicating positive, pessimistic, or depressed mental states; keywords indicating willingness to fulfill obligations, delays in fulfilling obligations, or refusal to fulfill obligations; emotionally sensitive words indicating joy, anger, sorrow, happiness, intellect, rudeness, aversion, dislike, or gratitude; negative keywords related to pornography, terrorism, violence, or insults; and the tone of voice corresponding to the keywords in the audio.

[0009] Preferably, the AI ​​digital human subsystem receives multimodal dialogue information from the AI ​​interaction subsystem and uses core AI technologies such as speech synthesis, speech recognition, semantic understanding, image processing, machine translation, and virtual avatar driving to enable digital mediators to conduct legal education, provide legal consultation, and interact with parties involved, thereby guiding parties to have a willingness to mediate and reach a settlement.

[0010] Preferably, the AI ​​mediation initiation subsystem automatically analyzes cases and, based on different causes of action and the parties' household registration, age, and occupation information, automatically and / or manually selects the AI ​​mediation robot corresponding to the mediator with a high success rate in mediation; sets AI mediation parameters: number of mediation robots, mediation robot, start time, termination conditions, and initiates AI mediation, transmitting the AI ​​mediation parameters to the AI ​​mediation dialogue subsystem.

[0011] Preferably, the mediation dialogue subsystem generates dialogue content between the AI ​​mediator and the parties in real time, receives AI mediation parameters from the AI ​​mediation initiation subsystem, calls the mediator's multimodal mediation template in the AI ​​mediation library subsystem, generates a multimodal dialogue to be communicated with the parties, and feeds it back to the AI ​​interaction subsystem. It also receives multi-dimensional intelligent annotations from the AI ​​intelligent annotation subsystem, and based on the multi-dimensional intelligent annotations from the AI ​​intelligent annotation subsystem and the mediator's multimodal mediation template in the AI ​​mediation library subsystem, combined with the context of the current mediation dialogue, generates a multimodal dialogue to be communicated with the parties, and feeds it back to the AI ​​interaction subsystem.

[0012] Preferably, the intelligent annotation subsystem performs multimodal intelligent annotation on the parties involved. The AI ​​intelligent annotation subsystem receives message text elements, audio text elements, and video text elements from the AI ​​interaction subsystem, combines and analyzes the message text elements, audio text elements, and video text elements of the parties involved, and when a keyword is matched in an element, the parties involved are intelligently annotated in multiple dimensions and given preset data tags. The multi-dimensional intelligent annotation includes the parties' original files, audio, and video information, analyzed and organized into corresponding message / audio / video text elements, and corresponding data tags. For example, the relationship between elements and annotations is given. The multi-dimensional intelligent annotation of the parties involved is fed back to the AI ​​mediation dialogue subsystem, and the multimodal intelligent annotation and analysis results of the parties involved are transmitted to the AI ​​mediation library subsystem for multimodal storage.

[0013] Preferably, the intelligent monitoring subsystem retrieves multimodal mediation supervision data from the mediation database subsystem within the large-scale mediation model system, and acquires real-time multimodal mediation dialogue data from human mediators; identifies whether the mediation behavior of human mediators conforms to the mediation process specifications, and prompts mediators to make real-time improvements; generates mediation analysis reports for human mediators, and combines the mediation supervision process data to form multimodal mediation supervision data, which is then transmitted to the mediation database subsystem within the large-scale mediation model system.

[0014] Preferably, during the monitoring and mediation process, the mediation results are checked. Multimodal mediation supervision data, such as mediation process standardization data, mediation compliance data, and mediation supervision cases, are retrieved from the AI ​​mediation database subsystem. Real-time multimodal mediation dialogue data from human mediators is obtained, and text, voice, and video elements, as well as tone of voice, facial expressions, and gestures, are processed in real time. This analysis examines whether the human mediators adhered to the mediation process standards, including identifying themselves, verifying the identities of the parties, explaining the reasons for mediation, and listening to the parties' voices. The analysis also examines whether the human mediators complied with regulations, such as whether they used sensitive words, maintained neutrality, and respected the parties' wishes. Whether there is emotional instability; analysis reveals that the mediation behavior of the human mediator does not conform to the mediation process norms, prompting the mediator to make real-time improvements; analysis reveals that the mediation of the human mediator is non-compliant, and real-time monitoring actions are taken, such as blocking the call and recording and reminding the mediator if content from the sensitive word library is detected; if an overly aggressive dialogue or emotional instability is detected, the human mediation administrator or AI mediation supervisor is notified to take over the mediation or immediately end the mediation; a mediation analysis report of the human mediator is generated, analyzing the mediator's good performance and areas for improvement; the process data and analysis report of mediation supervision form multimodal mediation supervision data, which is output to the AI ​​mediation library subsystem.

[0015] Preferably, the mediation database subsystem obtains information on different causes of action and related laws, regulations, and precedents from a third-party resource database; receives and stores multimodal mediation data from the manual mediation subsystem, including mediation process data and anonymized case materials; receives and stores multimodal mediation data of the parties involved collected, analyzed, and organized during AI mediation from the AI ​​interaction subsystem; and receives and stores multimodal intelligent annotation and analysis results of the parties involved collected and analyzed by the mediators during a single mediation session from the AI ​​intelligent annotation subsystem. It also analyzes and organizes multimodal mediation data of disputes where mediation has been reached and where not. The system collects multimodal mediation data, including mediation scripts, techniques, and dialogues with high success rates for different types of cases. This data is then output to the AI ​​mediation training subsystem. The AI ​​mediation training subsystem receives multimodal mediation templates from trained mediators. These templates include the mediator's experience, scripts, techniques, and cases in various disputes, their knowledge of relevant laws and regulations, successful third-party mediation cases, and litigation precedents. The system also includes the mediator's thought process and speaking logic, their rating of the parties' willingness to mediate, and their typical tone of voice, facial expressions, and body language data, information, and procedures.

[0016] Preferably, the mediation training subsystem retrieves multimodal mediation data of mediators mediating disputes of different causes of action stored in the AI ​​mediation library subsystem; the AI ​​mediation training subsystem retrieves multimodal mediation algorithm models generated and published by the AI ​​mediation modeling subsystem; based on the existing multimodal mediation templates and multimodal mediation data of a specified mediator, the AI ​​mediation training subsystem calls, debugs, and runs the multimodal mediation algorithm models of the AI ​​mediation modeling subsystem, constructing a complete set of model training tools such as entity recognition models, intent classification models, sentiment classification models, ASR-related models, and TTS-related models, and constructing AI strategy feature sets and machine learning models. The AI ​​strategy engine is self-iterative and continuously updated and improved, outputting AI strategies. Using the AI ​​strategy engine and model training tools, the multimodal mediation template of the mediator is continuously iterated and optimized. The multimodal mediation template includes the mediator's mediation process flow diagram for various disputes, including: mediation context and path, mediation nodes, mediation skills, successful mediation cases, unsuccessful mediation cases, relevant laws and regulations, successful mediation cases by third parties, litigation precedents, the mediator's thinking logic, speaking logic, and data and information on the mediator's usual tone of voice, facial expressions and gestures. The multimodal mediation template is then output to the AI ​​mediation library subsystem.

[0017] In the AI-based mediation method based on a large data model provided above, this application embodiment empowers mediation by leveraging a large model and combining it with advanced technologies such as natural language processing, speech conversion, and semantic recognition. This allows for the training of AI mediators by aggregating various case types, relevant laws and regulations, points of contention, precedents, successful mediation cases, mediation techniques, and mediation scripts, equipping them with the mediation capabilities of top-tier mediators and improving mediation efficiency. Furthermore, in some embodiments, training AI mediation quality inspectors to monitor the mediation process in real time enhances mediation quality and ensures efficiency, compliance, and security. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 A system architecture diagram of an AI mediation system provided in this application embodiment; Figure 2 An AI mediation flowchart of an AI mediation system provided in this application embodiment; Figure 3 This application provides an AI mediation system training flowchart for an embodiment of the present application. Figure 4 This is a flowchart illustrating the AI ​​mediation supervision process of an AI mediation system provided in this application embodiment. Detailed Implementation

[0020] Various exemplary embodiments of the present application will now be described in detail with reference to the accompanying drawings. It should be noted that, unless otherwise specifically stated, the relative arrangement, numerical expressions, and values ​​of the components and steps set forth in these embodiments do not limit the scope of the present application.

[0021] Those skilled in the art will understand that the terms "first," "second," etc., in the embodiments of this application are only used to distinguish different steps, devices, or subsystems, and do not represent any specific technical meaning, nor do they indicate a necessary logical order between them. It should also be understood that in the embodiments of this application, "multiple" can refer to two or more, and "at least one" can refer to one, two, or more. It should also be understood that any component, data, or structure mentioned in the embodiments of this application can generally be understood as one or more unless explicitly defined or given a contrary indication in the context. Furthermore, the term "and / or" in this application is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this application generally indicates that the preceding and following related objects have an "or" relationship. It should also be understood that the descriptions of the various embodiments in this application emphasize the differences between them; their similarities or commonalities can be referred to mutually, and for the sake of brevity, they will not be elaborated upon one by one.

[0022] Furthermore, it should be understood that, for ease of description, the dimensions of the various parts shown in the accompanying drawings are not drawn to actual scale. The following description of at least one exemplary embodiment is merely illustrative and is in no way intended to limit this application or its application or use. Techniques, methods, and devices known to those skilled in the art will not be discussed in detail, but where appropriate, such techniques, methods, and devices should be considered part of the specification. It should be noted that similar reference numerals and letters in the following drawings denote similar items; therefore, once an item is defined in one drawing, it need not be further discussed in subsequent drawings.

[0023] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0024] Figure 1This is a system architecture diagram of an AI mediation system provided for embodiments of this application. It should be understood that the system shown in the diagram is exemplary and not restrictive. This means that the system architecture involved is not limited to a specific form or design, but is presented as an example. In other words, the architecture shown in the diagram can be considered as a way of expressing related concepts and relationships clearly, and does not exclude other forms of architecture. Therefore, when interpreting the architecture in the diagram, it should be understood that the model is flexible and diverse, and its purpose is to provide an exemplary description, not a restrictive provision on a specific form.

[0025] like Figure 1 As shown, the interaction system 101 receives multimodal information from the parties involved and feeds back the multimodal dialogue information output by the mediation engine system to the parties, thereby completing the multimodal dialogue interaction between the AI ​​mediator and the parties. It should be noted that the interaction system includes: an interaction subsystem, a digital human subsystem, and a human mediation subsystem. The AI ​​interaction system is responsible for the interaction between the AI ​​mediator and the parties. On the one hand, it receives multimodal dialogue information (including text, voice, tone, actions, facial expressions, etc.) from the AI ​​mediation engine system and completes a multimodal dialogue with the parties; on the other hand, it collects multimodal information (text, images, audio, and video) from the parties during a multimodal dialogue.

[0026] Further (such as) Figure 2 As shown, the cases requiring mediation are first imported into the AI ​​mediation system via API or files (such as forms). Then, the AI ​​mediation system analyzes the cases and invokes an AI mediation robot trained on a large-scale mediation model. Next, the AI ​​mediation robot utilizes technologies such as ICT, AR / VR, big data, and blockchain to mediate with the parties based on the content of the mediation application. Following this, the AI ​​mediation robot reaches the parties via telephone, SMS, video, AR / VR, and digital avatars. Subsequently, the AI ​​mediation robot profiles the parties, analyzes their communication intentions, and engages in continuous mediation communication, conducting interactive and layered negotiations to guide the parties to reach a mediation agreement, sign legal documents such as mediation agreements / judicial confirmation applications, and fulfill the legal obligations stipulated in the mediation agreement. Finally, mediation supervisors monitor the AI ​​mediator's operations in real time, identifying business problems during the mediation process, adjusting mediation strategies, and maintaining the stable operation of the mediation robot.

[0027] Specifically, the interaction subsystem is used to respond to the multimodal information of the parties involved, driving the AI ​​mediator to conduct multimodal dialogue and interaction with the parties. This includes: collecting the parties' audio information and transmitting the target audio segment and corresponding text element information to the intelligent annotation subsystem in the mediation engine system; collecting the parties' video information and recognizing their actions and micro-expressions, transmitting the target video segment and corresponding emotional text elements to the intelligent annotation subsystem in the mediation engine system; collecting the parties' real-time messages during the interaction and transmitting the corresponding text elements to the intelligent annotation subsystem in the mediation engine system; driving the digital human subsystem to conduct multimodal dialogue with the parties based on the multimodal dialogue information received from the mediation dialogue subsystem in the mediation engine system, and transferring the connection to a human mediator based on preset conditions; and constructing multimodal mediation data for the parties based on their text records, voice records, video records, and identified and organized text, audio, and video elements, transmitting this data to the mediation library subsystem in the large-scale mediation model system for multimodal storage.

[0028] In one embodiment, the AI ​​interaction subsystem is responsible for the AI ​​mediator's interaction with the parties involved via voice, video, and text (such as SMS, instant messaging, etc.).

[0029] First, the audio information of the parties involved is collected and extracted. Continuous analog audio signals are converted into discrete digital signals, framed, and MFCC features are extracted. The optimal probability path is found, and an acoustic model identifies phonemes, while a lexical model and a language model identify words and sentences. The audio signals of the parties involved during the mediation process, recorded by the mediator and mediation assistant, are organized into short audio segments and corresponding textual elements (keywords indicating legal knowledge, income, work, daily life, consumption, family, social activities, etc.; keywords indicating mental states such as positive, pessimistic, and depressed; keywords indicating willingness to fulfill obligations, delays in fulfilling obligations, refusal to fulfill obligations, etc.; emotionally sensitive words such as joy, anger, sorrow, intellect, rudeness, aversion, dislike, gratitude, etc.; negative keywords such as pornography, terrorism, violence, and insults; the tone of voice corresponding to the keywords, etc.), and transmitted to the AI ​​intelligent annotation subsystem.

[0030] Next, video information of the parties involved is collected and extracted. Computer vision technology is used to detect the position and contour of the face; deep learning models are used to identify key points on the face. Machine learning models, such as convolutional neural networks (CNN) and recurrent neural networks (RNN), are used to identify micro-expressions and movements to form video element slices. Local video element slices include micro-movement videos of the eyes, ears, mouth, nose, eyebrows, hair, wrinkles, cheeks, and facial muscles, while overall video element slices include movement videos of the head, shoulders, neck, hands, elbows, arms, torso, hips, and limbs. The emotional text elements of the parties involved are identified and sorted from the local and overall video element slices: happiness, surprise, fear, anger, sadness, nausea, contempt, interest, etc. The identified emotional text element information is then transmitted to the AI ​​intelligent annotation subsystem.

[0031] Secondly, the system collects and extracts real-time messages from the parties involved. It collects and extracts real-time messages (such as SMS, WeChat, and Fetion) between the parties and the mediator; analyzes the text, punctuation, and emoticons sent by the parties in these messages, identifies online slang, analyzes the context, and organizes the parties' textual elements (keywords indicating legal awareness, income, work, daily life, consumption, family, and social activities; keywords indicating mental states such as positive, pessimistic, and depressed; keywords indicating willingness to fulfill obligations, delays in fulfilling obligations, or refusal to fulfill obligations; emotionally sensitive words such as joy, anger, sorrow, intellect, rudeness, aversion, dislike, and gratitude; and negative keywords such as pornography, terrorism, violence, and insults). The identified and organized textual elements are then transmitted to the AI ​​intelligent annotation subsystem.

[0032] Next, the system receives multimodal dialogue (including text, voice, tone, actions, and facial expressions) from the AI ​​mediation dialogue subsystem, invokes the AI ​​digital human subsystem, and completes a mediation dialogue with the parties involved. Subsequently, if the parties do not wish to mediate or explicitly require human mediation, the AI ​​mediation ends, the AI ​​mediation record is transferred to the AI-human mediation subsystem, and the mediation is transferred to a human mediator. Then, based on the parties' original text, voice, and video records, as well as the identified and processed text, audio, and video elements, multimodal mediation data is generated and transmitted to the AI ​​mediation library subsystem of the AI ​​mediation big data model system for multimodal storage.

[0033] The digital human subsystem is used to receive multimodal dialogue information from the interaction subsystem to conduct multimodal dialogue interactions with the parties involved. Specifically, the AI ​​digital human subsystem receives multimodal dialogue information from the AI ​​interaction subsystem and utilizes multiple core AI technologies such as speech synthesis, speech recognition, semantic understanding, image processing, machine translation, and virtual avatar driving to enable digital mediators to conduct legal education, provide legal consultation, and interact with the parties involved, guiding them to have a willingness to mediate and reach a settlement.

[0034] The aforementioned human mediation subsystem receives instructions to switch from AI-based mediation to human mediation, assigning cases to appropriate human mediators. Specifically, it receives these instructions and assigns cases to the corresponding human mediators. The human mediators utilize the ICT technologies integrated into the AI-based human mediation subsystem, along with cloud computing, blockchain, big data, and other digital intelligence technologies, to conduct multimodal mediation with the parties involved via telephone, SMS, and video. The process and results of the multimodal human mediation (no willingness to mediate, no settlement agreement reached, settlement agreement reached, case closed, etc.) are fed back to the AI ​​mediation database subsystem.

[0035] The mediation engine system 102 initiates one or more rounds of AI mediation by one or more AI mediators. It generates multimodal dialogue information by combining the multimodal mediation templates from the large-scale mediation model system and the context of the dialogue in the interactive system, and sends this information to the interactive system. It should be noted that the mediation engine system includes: a mediation initiation subsystem, a mediation dialogue subsystem, an intelligent annotation subsystem, and an intelligent monitoring subsystem. The AI ​​mediation engine system is responsible for initiating one or more rounds of AI mediation by one or more AI mediators. Based on the trained AI multimodal mediation templates received from the large-scale mediation model system, combined with the multi-dimensional intelligent annotation results formed during the interaction between the AI ​​interactive system and the parties involved, and considering the context prior to the current dialogue, it outputs the intelligently generated scripts to the AI ​​interactive system, driving the AI ​​interactive system to mediate with the parties involved.

[0036] Specifically, the mediation initiation subsystem is used to import mediation cases into the system and select target AI mediators to transmit AI mediation parameters to the mediation dialogue subsystem. Further, cases are imported into the AI ​​mediation system via API or files (such as tables). The AI ​​mediation initiation subsystem automatically analyzes cases and, based on different causes of action and information such as the parties' household registration, age, and occupation, automatically (or manually) selects the AI ​​mediation robot corresponding to a mediator with a high success rate; it sets AI mediation parameters: the number of mediation robots, the number of mediation robots, the start time, and termination conditions such as time, keywords, and manual intervention. The AI ​​mediation initiation subsystem then initiates AI mediation and transmits the AI ​​mediation parameters to the AI ​​mediation dialogue subsystem.

[0037] The mediation dialogue subsystem generates multimodal dialogue information based on the AI ​​mediation parameters from the mediation initiation subsystem, the multimodal mediation template from the mediation library subsystem in the large-scale mediation model system, the multidimensional intelligent annotation from the intelligent annotation subsystem, and the context of the current mediation dialogue. This information is then transmitted to the interaction subsystem within the interaction system. In one embodiment, the dialogue content between the AI ​​mediator and the parties involved is generated in real time. The system receives the AI ​​mediation parameters from the AI ​​mediation initiation subsystem and invokes the mediator's multimodal mediation template from the AI ​​mediation library subsystem. It generates a multimodal dialogue (text, voice, tone, actions, facial expressions, etc.) to be communicated with the parties involved and feeds it back to the AI ​​interaction subsystem. It also receives the multidimensional intelligent annotation from the AI ​​intelligent annotation subsystem. Based on the multidimensional intelligent annotation from the AI ​​intelligent annotation subsystem and the mediator's multimodal mediation template from the AI ​​mediation library subsystem, combined with the context of the current mediation dialogue, it generates a multimodal dialogue (text, voice, tone, actions, facial expressions, etc.) to be communicated with the parties involved and feeds it back to the AI ​​interaction subsystem.

[0038] The intelligent annotation subsystem receives text, audio, and video text elements from the interactive subsystem within the interactive system and generates corresponding data tags. The results of this multi-dimensional intelligent annotation are then transmitted to the mediation dialogue subsystem and the mediation database subsystem within the large-scale mediation model system, respectively. Specifically, multi-modal intelligent annotation is performed on the parties involved. The AI ​​intelligent annotation subsystem receives message text, audio text, and video text elements from the AI ​​interactive subsystem. It combines and analyzes these elements, and when keywords are matched, it performs multi-dimensional intelligent annotation on the parties (life status, personality, honesty coefficient, repayment ability, willingness to mediate, mediation success rate, etc.), affixing preset data tags. Multi-dimensional intelligent annotation includes the parties' original files, audio, and video information, analyzed and organized into corresponding message / audio / video text elements, and corresponding data tags. An example of the relationship between elements and annotations is provided. The multi-dimensional intelligent annotation of the parties is fed back to the AI ​​mediation dialogue subsystem. The multi-modal intelligent annotation and analysis results of the parties are transmitted to the AI ​​mediation database subsystem for multi-modal storage.

[0039] The intelligent monitoring subsystem is used to monitor the human mediation process and check the mediation results. Specifically, it retrieves multimodal mediation supervision data from the mediation database subsystem of the large-scale mediation model system and obtains multimodal mediation dialogue data of human mediators in real time; it identifies whether the mediation behavior of human mediators conforms to the mediation process specifications and prompts mediators to make improvements in real time; it generates mediation analysis reports for human mediators and combines them with the mediation supervision process data to form multimodal mediation supervision data, which is then transmitted to the mediation database subsystem of the large-scale mediation model system.

[0040] In one embodiment, the mediation process is monitored, and the mediation outcome is examined (e.g., ...). Figure 4 (As shown). The AI ​​mediation database subsystem retrieves multimodal mediation supervision data, such as mediation process standard data, mediation compliance data, and mediation supervision cases. It acquires real-time multimodal mediation dialogue data from human mediators, and organizes text, voice, and video elements, as well as tone of voice, facial expressions, and other multimodal mediation elements. It analyzes whether human mediators follow the mediation process standards of "identifying the mediator's identity, verifying the parties' identities, explaining the reasons for mediation, and listening to the parties' voices," and whether the mediation is compliant, such as whether sensitive words are used, whether neutrality is maintained, whether the parties' wishes are respected, and whether emotions are out of control. If the analysis finds that the human mediator's mediation behavior does not conform to the mediation process standards, it prompts the mediator to improve in real time. If the analysis finds that the human mediator's mediation is non-compliant, it takes real-time supervisory actions, such as blocking the call and recording and reminding the mediator if sensitive words are detected; if an overly aggressive dialogue or emotional outburst is detected, it alerts the human mediation administrator or AI mediation supervisor to take over the mediation or immediately end the mediation. It generates a mediation analysis report for the human mediator, analyzing the mediator's strengths and areas for improvement. The process data and analysis reports of mediation supervision are used to form multimodal mediation supervision data, which is then output to the AI ​​mediation library subsystem.

[0041] The mediation big data model system 103 generates the multimodal mediation template based on the multimodal mediation data and multidimensional intelligent annotation results of the AI ​​mediator, and sends it to the mediation engine system. It should be noted that the mediation big data model system includes: a mediation library subsystem, a mediation training subsystem, a mediation modeling subsystem, and a third-party resource library subsystem. The mediation big data model system receives multimodal mediation data and multidimensional intelligent annotation results from the AI ​​mediation engine system, as well as laws, regulations, and case law from the third-party resource library. It performs AI training according to certain modeling and algorithms, and outputs AI multimodal mediation templates for the mediators, which the AI ​​mediation engine system can call when initiating AI mediation.

[0042] Specifically, the mediation database subsystem is used to construct core mediation data based on the outputs of one or more subsystems. In one embodiment, it acquires target legal information from a third-party resource database subsystem; acquires multimodal mediation data from a human mediation subsystem in an interactive system; acquires multimodal mediation data from an interactive subsystem in an interactive system; acquires multimodal intelligent annotation results from an intelligent annotation subsystem in a mediation engine system; generates multimodal mediation data based on multimodal mediation process data for transmission to a mediation training subsystem; and receives multimodal mediation templates from trained mediators in the mediation training subsystem.

[0043] In one implementation scenario, information such as different causes of action and related laws, regulations, and precedents is obtained from a third-party resource library. The system receives and stores multimodal mediation data from the manual mediation subsystem, including mediation process data and anonymized case materials. It also receives and stores multimodal mediation data collected, analyzed, and organized by the parties involved during AI-driven mediation. Furthermore, it receives and stores multimodal intelligent annotations and analysis results collected and organized by the mediators during a single mediation session from the AI ​​intelligent annotation subsystem. Finally, it analyzes and organizes multimodal mediation process data for disputes where mediation has been reached and unresolved, including high-success-rate mediation techniques, skills, and dialogues for different causes of action, and outputs this data to the AI ​​mediation training subsystem. The AI ​​mediation training subsystem receives multimodal mediation templates of trained mediators. These templates include the mediator's experience in mediating various disputes, their mediation techniques, skills, and cases, as well as their knowledge of relevant laws and regulations, successful third-party mediation cases, and litigation precedents. The mediator's thought process and speaking logic are also included, along with their rating of the parties' willingness to mediate, and data, information, and procedures such as their typical tone of voice, facial expressions, and gestures.

[0044] The mediation training subsystem is used to optimize multimodal mediation templates using AI policy feature sets and target machine learning models. Specifically, it retrieves multimodal mediation data from the mediation library subsystem and multimodal mediation algorithm models generated from the mediation modeling subsystem; based on existing multimodal mediation templates and multimodal mediation data, it trains the multimodal mediation algorithm models to optimize the multimodal mediation templates.

[0045] In one embodiment (such as) Figure 3As shown, the AI ​​mediation training subsystem retrieves multimodal mediation data stored in the AI ​​mediation database subsystem, representing mediators' work on disputes of different causes. The AI ​​mediation training subsystem also retrieves multimodal mediation algorithm models generated and published by the AI ​​mediation modeling subsystem. Based on existing multimodal mediation templates and data for specified mediators, the AI ​​mediation training subsystem calls, debugs, and runs the multimodal mediation algorithm models from the AI ​​mediation modeling subsystem, constructing a complete set of model training tools such as entity recognition models, intent classification models, sentiment classification models, ASR-related models, and TTS-related models, achieving technological innovation in AI and efficiency in the training process. Finally, it constructs an AI strategy feature set, a machine learning model engine, and self-iterates to continuously update and improve the AI ​​strategy engine, outputting AI strategies. Using an AI strategy engine and model training tools, the mediator's multimodal mediation template is continuously iterated and optimized. The multimodal mediation template includes the mediator's mediation process flow diagram for various disputes, including: mediation context and path, mediation nodes (multi-dimensional intelligent annotation of parties, mediation scripts, etc.), mediation skills, successful mediation cases, unsuccessful mediation cases, the laws and regulations involved in such disputes, successful mediation cases by third parties, litigation precedents, etc., the mediator's thinking logic, speaking logic, etc., as well as data and information such as the mediator's usual tone of voice, facial expressions and actions; the multimodal mediation template is then output to the AI ​​mediation library subsystem.

[0046] The mediation modeling subsystem integrates the target algorithm through a multi-core algorithm structure and generates a multimodal mediation algorithm model for use by the mediation training subsystem. Specifically, the AI ​​mediation modeling subsystem effectively integrates mainstream and specialized algorithms (LR - Linear Regression, Log-LR, LDA - Linear Discriminant Analysis, Cost-sensitive learning, DAG, Adaboost, GBDT - Gradient Boosting Decision Tree, XGBoost - eXtreme Gradient Boosting, KNN (K-Nearest Neighbor), RNN - Recurrent Neural Network, AGNES - AGglomerative NESting, etc.) to form a multimodal mediation algorithm model. This ensures accuracy while providing mutual verification for use by the AI ​​mediation training subsystem.

[0047] The third-party resource library subsystem is used to acquire target databases and target large models, including but not limited to legal and regulatory databases, judicial large models, government affairs large models, and consumption large models.

[0048] It is important to note that the entire mediation process is documented in real time, including text, images, audio, and video, ensuring traceability and preventing unnecessary disputes. AI mediators possess the mediation capabilities of top-tier mediators, significantly alleviating the current shortage of mediators in terms of both quantity and professionalism. AI quality inspectors monitor the mediation process in real time, effectively guaranteeing compliance. A comprehensive mediation model analyzes case information from multiple dimensions, combining it with a mediation experience database to provide character profiles, mediation reports, and mediation plans, effectively improving the success rate of mediation.

[0049] Furthermore, the design of the AI ​​mediation system framework, through the efficient operation and coordination of various AI mediation subsystems, enables AI mediators to automatically mediate, learn, iteratively optimize and evolve, and interact with human mediators. The large-scale mediation model analyzes, organizes, and models multimodal mediation data from mediators, training to form multimodal mediation templates. The AI ​​interaction system collects, analyzes, and organizes real-time multimodal mediation data for the parties involved. The AI ​​mediation engine achieves intelligent multimodal and multidimensional labeling of parties during a mediation dialogue, calls upon multimodal mediation templates, and autonomously conducts mediation. The AI ​​mediation engine calls upon the large-scale mediation model to self-learn and iteratively optimize the multimodal mediation templates. The large-scale mediation model continuously incorporates laws, regulations, and mediation-related legal data, and through modeling and training, it continuously optimizes the multimodal mediation templates.

[0050] Furthermore, this application also provides a computer program product that, when run on a terminal device, causes the terminal device to execute any of the methods described above.

[0051] As can be seen from the above description of the embodiments, those skilled in the art can clearly understand that all or part of the steps in the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, a server, or a network communication device such as a media gateway, etc.) to execute the methods described in various embodiments or some parts of the embodiments of this application.

[0052] It should be noted that the various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to in the method section.

[0053] It should also be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0054] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An AI mediation method based on a large data model, characterized in that, An AI-powered mediation system is employed, comprising an interaction system, a mediation engine system, and a large-scale mediation model system. The interactive system includes an interactive subsystem, a digital human subsystem, and a human mediation subsystem. The interactive subsystem is used to respond to the multimodal information of the parties involved, so as to drive the AI ​​mediator to conduct multimodal dialogue and interaction with the parties involved. The digital human subsystem is used to receive the multimodal dialogue information from the interactive subsystem, so as to conduct multimodal dialogue and interaction with the parties involved. The human mediation subsystem is used to receive the instruction to switch from AI mediation to human mediation, so as to assign the mediation case to the corresponding human mediator. The mediation engine system includes: a mediation initiation subsystem, a mediation dialogue subsystem, an intelligent annotation subsystem, and an intelligent monitoring subsystem. The mediation initiation subsystem imports cases into the system and selects a target AI mediator to transmit AI mediation parameters to the mediation dialogue subsystem. The mediation dialogue subsystem generates multimodal dialogue information based on the AI ​​mediation parameters from the mediation initiation subsystem, the multimodal mediation template from the mediation library subsystem in the large-scale mediation model system, the multidimensional intelligent annotation from the intelligent annotation subsystem, and the context of the current mediation dialogue. This information is then transmitted to the interaction subsystem within the interaction system. The intelligent annotation subsystem receives text, audio, and video text elements from the interaction subsystem in the interaction system and generates corresponding data tags. The results of the multidimensional intelligent annotation are then transmitted to the mediation dialogue subsystem and the mediation library subsystem in the large-scale mediation model system, respectively. The intelligent monitoring subsystem monitors the manual mediation process and checks the mediation results. The mediation big data model system includes: a mediation library subsystem, a mediation training subsystem, a mediation modeling subsystem, and a third-party resource library subsystem. The mediation library subsystem is used to construct core mediation data based on the outputs of one or more subsystems. The mediation training subsystem is used to optimize multimodal mediation templates using AI strategy feature sets and target machine learning models. The mediation modeling subsystem is used to integrate target algorithms through a multi-core algorithm structure and generate multimodal mediation algorithm models for use by the mediation training subsystem. The third-party resource library subsystem is used to acquire the target database and the target big data model. Cases requiring mediation are imported into the AI ​​mediation system via API or files. The AI ​​mediation system analyzes the cases and invokes an AI mediation robot trained on a large-scale mediation model. Utilizing ICT, AR / VR, big data, and blockchain technologies, the AI ​​mediation robot engages in mediation with the parties based on the content of the mediation application. The AI ​​mediation robot reaches the parties via phone, SMS, video, AR / VR, and digital human. It creates a profile of the parties, analyzes their communication intentions, and engages in continuous, interactive, and layered mediation communication and negotiation, guiding the parties to reach a mediation agreement, sign the legal documents (mediation agreement / judicial confirmation application), and fulfill the legal obligations stipulated in the mediation agreement. Mediation supervisors monitor the AI ​​mediator's operations in real time, identifying business problems during the mediation process, adjusting mediation strategies, and maintaining the stable operation of the mediation robot.

2. The AI ​​mediation method according to claim 1, characterized in that, The execution steps of the interactive subsystem include: The system collects and extracts audio information from the parties involved, converts continuous analog audio signals into discrete digital signals, performs frame segmentation, extracts MFCC features, finds the optimal probability path, uses an acoustic model to identify phonemes, and uses a vocabulary model and a language model to identify words and sentences. The system identifies and organizes the audio signals of the parties involved during the mediation process by the mediator and mediation assistant into short audio segments and corresponding text element information, and transmits them to the AI ​​intelligent annotation subsystem. The system collects and extracts video information from the parties involved, using computer vision technology to detect the position and contour of faces; it uses deep learning models to identify key points on faces, and machine learning models, convolutional neural networks, and recurrent neural networks to identify micro-expressions and movements to form video element slices. Local video element slices include micro-movement videos of the eyes, ears, mouth, nose, eyebrows, hair, wrinkles, cheeks, and facial muscles, while overall video element slices include videos of movements of the head, shoulders, neck, hands, elbows, arms, torso, hips, and limbs. The system identifies and organizes the emotional text elements of the parties involved from the local and overall video element slices: happiness, surprise, fear, anger, sadness, nausea, contempt, and interest. The identified emotions are then transmitted to the AI ​​intelligent annotation subsystem to form video text element information. The system collects and extracts real-time messages from the parties involved. It analyzes the text, punctuation marks, and emoticons sent by the parties in the real-time messages, identifies online slang, analyzes the context, and organizes the text elements of the parties involved. The identified and organized text elements of the parties involved are then transmitted to the AI ​​intelligent annotation subsystem. The system receives multimodal dialogues from the AI ​​mediation dialogue subsystem, invokes the AI ​​digital human subsystem to complete a mediation dialogue with the parties involved, and terminates the AI ​​mediation session if the parties do not wish to mediate or explicitly require human mediation. The AI ​​mediation record is then transferred to the AI ​​human mediation subsystem, which then transfers the mediation to a human mediator. Based on the parties' original text, voice, and video records, as well as the identified and organized text, audio, and video elements, multimodal mediation data is generated and transmitted to the AI ​​mediation library subsystem of the AI ​​mediation big model system for multimodal storage. The text elements include: keywords indicating legal knowledge (knowledge of the law, lack of knowledge of the law); keywords indicating the parties' living conditions (income, work, food, clothing, housing, transportation, consumption, family, social activities); keywords indicating positive, pessimistic, or depressed mental states; keywords indicating willingness to fulfill obligations, delays in fulfilling obligations, or refusal to fulfill obligations; emotionally sensitive words indicating joy, anger, sorrow, happiness, intellect, rudeness, aversion, dislike, or gratitude; negative keywords related to pornography, terrorism, violence, or insults; and the tone of voice corresponding to the keywords in the audio.

3. The AI ​​mediation method according to claim 1, characterized in that, The AI ​​digital human subsystem receives multimodal dialogue information from the AI ​​interaction subsystem and utilizes core AI technologies such as speech synthesis, speech recognition, semantic understanding, image processing, machine translation, and virtual avatar-driven systems to enable digital mediators to conduct legal education, provide legal consultation, and interact with parties involved, guiding them to have a willingness to mediate and reach a settlement.

4. The AI ​​mediation method according to claim 1, characterized in that, The AI ​​mediation initiation subsystem automatically analyzes cases and, based on different causes of action and the parties' household registration, age, and occupation information, automatically and / or manually selects the AI ​​mediation robot corresponding to the mediator with a high success rate. AI mediation parameters are set: number of mediation robots, number of robots, start time, and termination conditions. The AI ​​mediation initiation subsystem then initiates AI mediation and transmits the AI ​​mediation parameters to the AI ​​mediation dialogue subsystem.

5. The AI ​​mediation method according to claim 1, characterized in that, The mediation dialogue subsystem generates real-time dialogue content between the AI ​​mediator and the parties involved, receives AI mediation parameters from the AI ​​mediation initiation subsystem, calls the mediator's multimodal mediation template in the AI ​​mediation library subsystem, generates a multimodal dialogue to be communicated with the parties involved, and feeds it back to the AI ​​interaction subsystem. It also receives multi-dimensional intelligent annotations from the AI ​​intelligent annotation subsystem, and based on the multi-dimensional intelligent annotations from the AI ​​intelligent annotation subsystem and the mediator's multimodal mediation template in the AI ​​mediation library subsystem, combined with the context of the current mediation dialogue, generates a multimodal dialogue to be communicated with the parties involved, and feeds it back to the AI ​​interaction subsystem.

6. The AI ​​mediation method according to claim 1, characterized in that, The intelligent annotation subsystem performs multimodal intelligent annotation on the parties involved. The AI ​​intelligent annotation subsystem receives message text elements, audio text elements, and video text elements from the AI ​​interaction subsystem, combines and analyzes the message text elements, audio text elements, and video text elements of the parties involved, and when a keyword is matched in an element, the parties involved are intelligently annotated in multiple dimensions and given preset data tags. The multi-dimensional intelligent annotation includes the parties' original files, audio, and video information, analyzed and organized into corresponding message / audio / video text elements, and corresponding data tags. For example, the relationship between elements and annotations is given. The multi-dimensional intelligent annotation of the parties involved is fed back to the AI ​​mediation dialogue subsystem, and the multimodal intelligent annotation and analysis results of the parties involved are transmitted to the AI ​​mediation library subsystem for multimodal storage.

7. The AI ​​mediation method according to claim 1, characterized in that, The intelligent monitoring subsystem retrieves multimodal mediation supervision data from the mediation database subsystem within the large-scale mediation model system, and acquires real-time multimodal mediation dialogue data from human mediators. The system identifies whether the mediation behavior of human mediators conforms to the mediation process norms and prompts mediators to make improvements in real time; it generates mediation analysis reports for human mediators and combines them with process data of mediation supervision to form multimodal mediation supervision data, which is then transmitted to the mediation database subsystem of the mediation big data model system.

8. The AI ​​mediation method according to claim 7, characterized in that, During the monitoring and mediation process, the mediation results are checked. Multimodal mediation supervision data, such as mediation process standard data, mediation compliance data, and mediation supervision cases, are retrieved from the AI ​​mediation database subsystem. Multimodal mediation dialogue data of human mediators are obtained in real time. Text, voice, and video elements, as well as the multimodal mediation elements of mediators such as tone of voice, facial expressions, and actions, are sorted out in real time. It is analyzed whether the human mediators follow the mediation process standard of identifying themselves as mediators, verifying the identities of the parties, explaining the reasons for mediation, and listening to the voices of the parties. It is also analyzed whether the human mediators' mediation is compliant, such as whether they have used sensitive words, whether they have maintained the neutrality of mediation, whether they have respected the parties' mediation wishes, and whether they have lost control of their emotions. Analysis revealed that the mediation behavior of human mediators did not conform to the mediation process standards, prompting mediators to make real-time improvements. Analysis also detected non-compliant mediation by human mediators, triggering real-time monitoring actions, such as blocking calls and recording and alerting the mediator if sensitive vocabulary was detected; alerting human mediation administrators or AI mediation supervisors to take over the mediation or immediately terminate the mediation if overly aggressive dialogue or emotional outbursts were detected; generating mediation analysis reports for human mediators, identifying strengths and areas for improvement; and combining mediation monitoring process data and analysis reports to form multimodal mediation monitoring data, which is then output to the AI ​​mediation database subsystem.

9. The AI ​​mediation method according to claim 1, characterized in that, The mediation database subsystem obtains information on different causes of action and related laws, regulations, and precedents from a third-party resource database. It receives and stores multimodal mediation data from the manual mediation subsystem, including mediation process data and anonymized case materials. It also receives and stores multimodal mediation data collected, analyzed, and organized by the parties during AI mediation from the AI ​​interaction subsystem. Furthermore, it receives and stores multimodal intelligent annotations and analysis results collected and organized by the mediators during a single mediation session from the AI ​​intelligent annotation subsystem. Finally, it analyzes and organizes multimodal mediation data for both mediated and unmediated disputes. The data collected during the mediation process includes multimodal mediation data such as mediation scripts, techniques, and dialogues with high success rates for different types of cases. This data is then output to the AI ​​mediation training subsystem. The AI ​​mediation training subsystem receives multimodal mediation templates from the trained mediators. These templates include the mediator's experience, scripts, techniques, and cases in various disputes, their knowledge of the relevant laws and regulations, successful third-party mediation cases, and litigation precedents, as well as the mediator's thought process, speaking logic, rating of the parties' willingness to mediate, and data on the mediator's typical tone of voice, facial expressions, and body language.

10. The AI ​​mediation method according to claim 1, characterized in that, The mediation training subsystem retrieves multimodal mediation data of mediators mediating disputes of different causes of action stored in the AI ​​mediation database subsystem. It also retrieves multimodal mediation algorithm models generated and published by the AI ​​mediation modeling subsystem. Based on existing multimodal mediation templates and data for a specified mediator, the AI ​​mediation training subsystem calls, debugs, and runs the multimodal mediation algorithm models from the AI ​​mediation modeling subsystem. This constructs a complete set of model training tools, such as entity recognition models, intent classification models, sentiment classification models, ASR-related models, and TTS-related models, and builds AI strategy feature sets and machine learning model engines. The system iterates and continuously updates its AI strategy engine, outputting AI strategies. Using the AI ​​strategy engine and model training tools, it continuously optimizes the mediator's multimodal mediation template. This template includes a flowchart of the mediator's mediation process in various disputes, encompassing: mediation context and path, mediation nodes, mediation techniques, successful mediation cases, unsuccessful mediation cases, relevant laws and regulations, successful third-party mediation cases, litigation precedents, the mediator's thought process and speaking logic, and data and information on the mediator's typical tone of voice, facial expressions, and gestures. This multimodal mediation template is then output to the AI ​​mediation library subsystem.

Images