Dispute handling and defense document generation method, system and device, storage medium and program product

Abstract

The embodiment of the invention provides a dispute handling and defense document generation method, system and device, a storage medium and a program product, and relates to the technical field of computers. The dispute handling method comprises the steps of obtaining material information provided by at least one party of a dispute event; determining at least one information acquisition channel based on the material information by using a preset model, and acquiring at least one piece of supplementary information through the at least one information acquisition channel; and outputting a processing result for the dispute event by utilizing a preset model according to the material information and the at least one piece of supplementary information. According to the technical scheme provided by the embodiment of the invention, the advantages of an artificial intelligence technology are utilized instead of a preset dispute handling rule, differentiated solutions can be given for different dispute events, and the method has certain flexibility.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a dispute resolution and defense document generation method, system, device, storage medium, and program product. Background Art

[0002] With the rapid development of internet technology, people are becoming more and more accustomed to shopping through e-commerce platforms. After a buyer purchases a product on an e-commerce platform, the seller can then ship the product to the buyer through logistics. After receiving the product, the buyer can pay the seller for the corresponding fee.

[0003] However, during the transaction process, due to various reasons, a dispute may arise between the buyer and the seller regarding the purchased item. For example, the buyer may request a refund or refuse payment for the purchased item, but the seller may reject the request. The two parties may have different claims regarding the same item, thus causing a dispute.

[0004] These disputes can be handled manually or through pre-set dispute resolution rules. However, manual judgment is slow, resulting in low dispute resolution efficiency. Furthermore, manual processing requires highly specialized staff, resulting in high labor costs and the potential for human error. Pre-set dispute resolution rules, on the other hand, make it difficult to provide differentiated solutions for different dispute scenarios. Summary of the Invention

[0005] In view of the above problems, the embodiments of the present application provide a dispute resolution and defense document generation method, system, device, storage medium and program product that can improve the above problems.

[0006] In a first embodiment of the present application, a dispute resolution method is provided. The method includes: obtaining material information provided by at least one party to a dispute; determining at least one information acquisition channel based on the material information using a preset model, and obtaining at least one supplementary information through the at least one information acquisition channel; and outputting a resolution result for the dispute based on the material information and the at least one supplementary information using the preset model.

[0007] In a second embodiment of the present application, a dispute resolution method is provided. The method includes: obtaining material information provided by at least one party to a dispute; determining the intention of the at least one party to the dispute based on the material information; determining, when it is determined based on the intention that supplementary information is required, at least one information acquisition channel, and obtaining at least one supplementary information through the at least one information acquisition channel; and resolving the dispute based on the material information and the at least one supplementary information.

[0008] In a third embodiment of the present application, a method for generating a defense document is provided. The method includes: obtaining material information provided by a user regarding a dispute; determining the user's defense intent; when determining, based on the defense intent, that supplementary information is required, determining at least one information acquisition channel, and obtaining at least one supplementary information through the at least one information acquisition channel; and generating a defense document based on the defense intent, the material information, and the at least one supplementary information.

[0009] In a fourth embodiment of the present application, a method for generating a defense document is provided. The method includes: obtaining material information provided by a user regarding a dispute event; displaying a defense document generated by a preset model in response to a document generation instruction triggered by the user through a client interface; and displaying interactive controls so that the user can confirm and / or edit the defense document through the interactive controls. The defense document is generated by the preset model based on the material information, or by the preset model based on the material information and at least one supplementary information; the at least one supplementary information is related to the dispute event and is obtained by the preset model through at least one information acquisition channel.

[0010] In a fifth embodiment of the present application, a service system is provided. The service system includes: The client is used to respond to user operations, obtain material information corresponding to the dispute event, and send the material information to the server; A server, configured to implement the steps in each embodiment corresponding to the above dispute resolution method; The client is also used to display the processing results returned by the server.

[0011] In a sixth embodiment of the present application, a service system is provided. The service system includes: The client is used to obtain the material information provided by the user in response to the dispute event, respond to the document generation instruction triggered by the user through the client interface, and send a request containing the material information to the server; A server, configured to implement the steps in the above-mentioned method for generating a defense document and send the defense document to the client; The client is further used to display the defense document and interactive controls; The user can confirm the defense document and / or edit the defense document through the interactive control.

[0012] In a seventh embodiment of the present application, an electronic device is provided. The electronic device includes: a memory and a processor, wherein the memory is configured to store a program; and the processor, coupled to the memory, is configured to execute the program stored in the memory to implement the steps of the dispute resolution method provided in the above embodiments, or the steps of the defense document generation method provided in the above embodiments.

[0013] In an eighth embodiment of the present application, a computer-readable storage medium is provided. The computer-readable storage medium stores a computer program or instructions, which, when executed, implement the steps of the dispute resolution method provided in the above embodiments, or the steps of the defense document generation method provided in the above embodiments.

[0014] In a ninth embodiment of the present application, a computer program product is provided. The computer program product includes a computer program that, when executed, causes a computer to execute the steps of the dispute resolution method provided in the above embodiments, or to implement the steps of the defense document generation method provided in the above embodiments.

[0015] The technical solution provided in the embodiments of this application utilizes artificial intelligence technology (i.e., a preset model) to empower dispute scenarios, replacing or assisting manual dispute handling. This improves dispute resolution speed, improves efficiency, reduces labor costs, and ensures a certain degree of accuracy in dispute resolution. Furthermore, by leveraging the advantages of artificial intelligence technology rather than relying on preset dispute resolution rules, the present application solution can provide differentiated solutions for different dispute cases, offering a certain degree of flexibility. Furthermore, the preset model in the embodiments of this application can also determine at least one information acquisition channel based on material information provided by at least one party to the dispute, and obtain supplementary information through the determined information acquisition channel to enrich the evidentiary information, thereby improving the preset model's ability to produce more reasonable resolutions for dispute cases.

[0016] Furthermore, other embodiments of the present application provide technical solutions that, after obtaining material information provided by at least one party to a dispute, determine the intent of at least one party to the dispute based on the material information and then, based on that intent, determine whether supplemental information is necessary. For example, in some scenarios, the buyer's intent is simply to obtain a refund. In such cases, it's difficult to arrive at a fair dispute resolution based solely on the material information provided by the buyer (e.g., product images, order information, etc.). Therefore, it's necessary to obtain additional supplemental information to assist in decision-making (e.g., prior high-quality precedents, more supplemental information that can assist in decision-making). When it's determined that supplemental information is necessary, the solutions provided by the embodiments of the present application can identify at least one information acquisition channel for the required information and obtain at least one supplemental information through the at least one information acquisition channel. Finally, the dispute is resolved based on the material information and the at least one supplemental information. Thus, the technical solutions provided by the embodiments of the present application can replace or assist manual dispute resolution, improving dispute resolution speed, efficiency, labor costs, and accuracy. Furthermore, the solutions provided by the embodiments of the present application can also obtain additional supplemental information to assist in decision-making based on the intent of at least one party to the dispute, helping to increase user confidence in dispute resolution information.

[0017] In addition, some embodiments of this application provide a technical solution for automatically generating a defense document. Specifically, after obtaining the material information of the dispute event provided by the user, the user's defense intention is determined; when it is determined that supplementary information needs to be obtained based on the defense intention, this solution can determine at least one information acquisition channel and obtain supplementary information through the at least one determined information acquisition channel to enrich the content of the defense document; finally, a defense document is generated based on the defense intention, the material information and the at least one supplementary information. During the entire process, the user only needs to provide the material information, and the defense document can be automatically generated, which provides convenience for the user's defense and reduces the difficulty of defense. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings: Figure 1 A schematic diagram of a service system provided in an embodiment of the present application; Figure 2 A schematic diagram of a service system provided in another embodiment of the present application; Figure 3 A flowchart of a dispute resolution method provided in one embodiment of the present application; Figure 4 A schematic diagram illustrating the principles of the dispute resolution method for the specific scenario provided in this application; Figure 5A flowchart of a dispute resolution method provided in another embodiment of the present application; Figure 6 A flowchart of a method for generating a defense document provided in one embodiment of the present application; Figure 7 A flowchart of a method for generating a defense document provided in another embodiment of the present application; Figure 8 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0019] To make the purpose, technical solutions, and advantages of this application more clear, the technical solutions of this application will be clearly and completely described below in conjunction with the specific embodiments of this application and the corresponding drawings. Obviously, the embodiments described are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0020] It should be noted that when the embodiments of this application involve user information, the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in the embodiments of this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and provide corresponding operation portals for users to choose to authorize or refuse. In addition, the various models involved in this application (including but not limited to language models or large models) are in compliance with relevant laws and standards.

[0021] First, the terms used in the embodiments of the present application are explained. It should be understood that this explanation is for a clearer understanding of the embodiments of the present application and does not necessarily constitute a limitation on the embodiments of the present application.

[0022] A chargeback, also known as a refund or cancellation, occurs when a credit card holder (the buyer) files a request with the bank (the credit card issuer) to refuse payment for an order within a certain period of time after payment. A chargeback is essentially a right granted by the bank to the buyer, and regardless of the e-commerce platform used by merchants (the sellers), this risk is inherent. When a buyer pays for an online transaction on an e-commerce platform using a credit card, they can file a chargeback with the bank according to the bank's prescribed rules and timeframes.

[0023] The concept of multimodality originates from the study of information representation in the field of computer-human interaction. The term "modality" is defined as the way information is expressed and interacted on a specific physical medium. Images, videos, audio, text, animated images, and so on are examples of modalities. The "preset model" mentioned in this article can be a "multimodal model." Key features of multimodal models include cross-modal learning, joint understanding and generation, and enhanced task adaptability. Multimodal models can convert and generate information between different modalities, such as generating image information from text descriptions and text descriptions from image information. The advantage of multimodal models lies in their ability to capture the inherent connections between different data types, thereby improving the expressive power of information understanding. Multimodal models have a wide range of applications in various fields, such as natural language processing, computer vision, and speech processing. The multimodal model is pre-trained on a diverse dataset, including image-text pairs, optical character recognition (OCR) data, interwoven image-text articles, visual question-answering datasets, video conversations, and image knowledge datasets. Data sources can include cleaned web pages, open-source datasets, and synthetic data. Training with image-document pairs focuses on learning image-text relationships, fostering a more nuanced understanding of the interplay between visual and textual information. OCR is used to identify textual content in images, as well as image classification tasks. The addition of visual question-answering datasets improves the model's ability to respond to image-related queries.

[0024] Multimodal Large Language Models (MLLMs): An artificial intelligence (AI) model capable of processing and generating multiple types of data, not just text. Evolved from Large Language Models (LLMs), MLLMs are capable of receiving and reasoning about multimodal information. They can not only understand and generate text, but also process data in multiple modalities, such as images, audio, and video, achieving cross-modal understanding and generation.

[0025] Preset Model: The preset model mentioned in this article refers to a machine learning model with a large number of parameters generated by training with a moderate amount of data using self-supervised or unsupervised methods. It provides excellent distributed feature representation and model generalization capabilities for downstream tasks. The preset model can be called a generalized large model. This article does not limit the specific structure of the preset model. For example, it can be a multimodal model, more specifically, the multimodal large language model mentioned above.

[0026] API: Application Programming Interface, application programming interface, is a set of well-defined rules and specifications for communication and interaction between different software applications. The functions of an application or service can be called through the API.

[0027] An intelligent agent (or agent for short) is a core concept in artificial intelligence and computer science, referring to an entity capable of perceiving its environment, making autonomous decisions, and executing actions to achieve specific goals or tasks. Based on their application areas, intelligent agents can be categorized as software agents (such as virtual assistants and intelligent customer service), physical agents (such as robots), and hybrid agents (combining software and hardware, such as smart home systems).

[0028] Chargeback Dispute: In e-commerce or cross-border transactions, chargeback dispute refers to the process in which when a consumer initiates a chargeback through a credit card company or payment institution, the merchant provides evidence to refute the chargeback in an effort to reverse the chargeback and recover financial losses.

[0029] In e-commerce transactions, when a merchant faces a complaint or claim related to a product issue (e.g., platform arbitration, consumer rights protection, or legal action), the merchant can submit a letter of rebuttal (or appeal letter) providing evidence and reasons to refute the claim if the merchant believes they are not responsible or need to mitigate their liability. The "rebuttal document" mentioned in the following examples can refer to the rebuttal letter, but can also include written defense materials, rebuttal opinions, objections, response letters, and more. Rebuttal documents can include text, images, links, and more.

[0030] During the transaction process of purchasing goods through e-commerce platforms, due to various reasons, disputes may arise between buyers and sellers regarding the purchased goods.

[0031] One approach to these dispute scenarios is to use manual judgment. For example, a buyer or seller can initiate a dispute resolution request to the e-commerce platform regarding the dispute. The e-commerce platform staff will intervene and make a judgment based on the evidence provided by the buyer or seller, thereby obtaining a resolution that allows the conflict between the buyer and seller to be resolved fairly. However, manual judgment is slow, resulting in low dispute resolution efficiency. Furthermore, the labor cost of dispute resolution is high, and when the number of disputes is large, the labor cost of dispute resolution will increase significantly. Furthermore, manual judgment relies on the experience and capabilities of the e-commerce platform staff. In the absence of these capabilities, the accuracy of dispute resolution may be low.

[0032] Another approach to these dispute scenarios is to use pre-set dispute resolution rules. For example, in a refund scenario, certain types of disputes can be directly refunded in full, or small-value items can be directly refunded based on the amount. However, using pre-set dispute resolution rules requires fixed rules. In practice, disputes arise from a variety of causes, types, and degrees of severity. Pre-set dispute resolution rules are difficult to handle, making it difficult to provide differentiated solutions for different dispute scenarios.

[0033] Leveraging AI to improve dispute resolution efficiency and build a highly intelligent, low-cost, and highly accurate dispute resolution system can address these issues. This will not only significantly increase dispute resolution efficiency but also enhance the fairness of dispute resolution, further advancing the field of e-commerce dispute resolution towards a higher level of intelligence.

[0034] For this purpose, this application is proposed. The technical solutions of each embodiment of this application apply artificial intelligence technology to the dispute event handling scenario. Figure 1 As shown, the service system provided by the present application. The service system includes: a client 102 and a server 101. Among them, the client 102 is used to respond to the user's operation, obtain the material information corresponding to the dispute event, and send the material information to the server. The server 101 is used to obtain the material information provided by at least one party to the dispute event; use a preset model to determine at least one information acquisition channel based on the material information, and obtain at least one supplementary information through the at least one information acquisition channel; use the preset model to output the processing result made for the dispute event based on the material information and the at least one supplementary information. The client 102 is also used to display the processing result returned by the server 101.

[0035] It's important to note that when a user is a buyer, their corresponding client is the buyer's client; when a user is a seller, their corresponding client is the seller's client. Different users can upload materials related to the same dispute through their respective clients. Of course, in some specific scenarios, only one party to the dispute, such as the buyer, provides material information.

[0036] Furthermore, the service system provided in this embodiment may also include another client, which may be a client corresponding to a manual auditor. Figure 1 The client corresponding to the manual reviewer can be used to receive the material information of the dispute event, the at least one supplementary information and the processing result sent by the server, and display them. The manual reviewer can use the interactive controls displayed on the client, such as Figure 1The "Edit" control and "Approved" control shown in the figure are used to manually review the handling result of the dispute event. For example, if the manual reviewer believes that the handling result of the dispute event is compliant, the "Approved" control can be clicked. The client corresponding to the manual reviewer responds to the manual reviewer's approval operation and sends a confirmation instruction to the server 101. After receiving the confirmation instruction sent by the client corresponding to the manual reviewer regarding the dispute event, the server 101 sends the handling result to the client 102 corresponding to the user for display on the interface of the client 102.

[0037] If the manual review deems that the handling result of the dispute event is not compliant, the handling result can be modified by clicking the "Edit" control. After the manual reviewer completes the modification and clicks to confirm the submission, the client corresponding to the manual reviewer sends the modified handling result to the server 101. After the server 101 receives the modified handling result for the dispute event, it sends the modified handling result to the client 102 corresponding to the user. In order to retain historical data, the server 101 can associate and store the handling result generated by the preset model corresponding to the dispute event, the handling result modified by the manual reviewer, the material information of the dispute event, and at least one supplementary information in a designated database. These retained historical data can provide data support for training or optimizing preset models.

[0038] If there is no manual review process, see Figure 1 In the first approach, the server 101 directly sends the processing result generated by the preset model to the client 102. In the process of adding manual review, after the preset model generates the processing result, the server 101 takes the second approach and sends the processing result generated by the preset model to the client corresponding to the manual reviewer (i.e. Figure 1 The client corresponding to the manual reviewer sends the processing result of the manual review to the server 101, and the server 101 then sends the processing result of the manual review to the client 102 corresponding to the user.

[0039] The server can deploy a pre-set model. When deploying a model, one can directly select a pre-set model designed and trained for dispute resolution scenarios. Alternatively, one can select an existing base model, construct training samples suitable for dispute resolution scenarios, and then fine-tune and train the base model. This allows the output of the fine-tuned and trained model to be more suitable for dispute resolution scenarios. The base model (such as a deep learning model) is also trained using a suitable amount of data using self-supervised or unsupervised methods. This embodiment does not specifically limit the specific structure of the base model or the fine-tuning and training methods.

[0040] The technical solutions provided in the above embodiments, including determining information acquisition channels, obtaining supplementary information through the channels, and generating processing results based on the material information and supplementary information of the dispute incident, can all be implemented using pre-set models. In fact, in another possible embodiment, each of the above steps (determining information acquisition channels, obtaining supplementary information, and generating processing results) can be partially implemented using pre-set models, or each of the above steps can be implemented using different pre-set models, or none of the above steps can be implemented using pre-compiled program code, search engines, expert systems, etc.

[0041] It should be noted here that different preset models refer to models with different model structures, or different models obtained using different training algorithms and / or adjustment algorithms.

[0042] Furthermore, information acquisition channels can be planned and analyzed based on the intentions of at least one party to the dispute and the materials provided by at least one party. These channels can include at least one knowledge base and at least one callable API. Each API corresponds to a functional module. Calling the API activates the corresponding functional module, and the content serving as the supplemental information is obtained through the functional model.

[0043] That is, another embodiment of the present application provides a service system, which includes a client 102 and a server 101. The client 102 is used to respond to user operations, obtain material information corresponding to the dispute event, and send the material information to the server. The server 101 is used to obtain material information provided by at least one party to the dispute event; based on the material information, determine the intention of at least one party to the dispute event; when it is determined that supplementary information needs to be obtained based on the intention, determine at least one information acquisition channel, and obtain at least one supplementary information through the at least one information acquisition channel; and handle the dispute event based on the material information and the at least one supplementary information.

[0044] In this embodiment, at least one of the following steps implemented by the server 101 may be implemented by a preset model, or by at least two different preset models: Determining the intention of at least one party to the dispute based on the material information; When it is determined that supplementary information needs to be obtained according to the intention, determining at least one information acquisition channel, and obtaining at least one supplementary information through the at least one information acquisition channel; The dispute event is handled according to the material information and the at least one supplementary information.

[0045] Of course, the service system provided in this embodiment may also include a client corresponding to a manual auditor (such as Figure 1103 in ). For details, please refer to the above text and will not be repeated here.

[0046] The technical solutions provided by the embodiments of this application not only provide dispute resolution services but also offer users other dispute-related services, such as the automatic generation of dispute-related documents (defense documents). Users only need to provide relevant material information. Alternatively, users can provide partial material information, and the server can obtain the required supplementary information from various information acquisition channels based on the partial material information provided by the user, and then generate a defense document based on the material information and supplementary information provided by the user. The defense document may include, but is not limited to, text, images, links, and so on.

[0047] In the e-commerce sector, for example, buyers may request a payment rejection for various reasons. For example, they may not have received the item or may not have received it within the agreed timeframe. Another example is when the item they received is not what they requested. Alternatively, they may receive the item but claim they did not receive it. In these cases, buyers may request a payment rejection. In these cases, sellers can initiate a chargeback process to counter the payment rejection.

[0048] In other scenarios, buyers may request a refund for a variety of reasons. For example, if the item they received was damaged, or the size, color, or style of the item they received didn't meet their expectations, or if they discovered the quantity wasn't what they expected after receiving the item, they might request a refund for the item they purchased.

[0049] Therefore, when the seller user initiates a defense against payment, the buyer user initiates a refund request, or in other dispute scenarios, the user can operate on the client 102 to initiate a dispute handling request for the commodity transaction order on the client 102. The client can send the dispute handling request to the server 101 with which the client has been pre-established.

[0050] In actual applications, for buyer users, the client can be a client for buyer users. For seller users, the client can be a client for seller users. The server 101 can be a server of an e-commerce platform, or other server that can provide dispute resolution functions.

[0051] The dispute resolution request includes material information related to the commodity transaction order, which may include but is not limited to at least one of the following: text, pictures, audio, video, animated images, and web page snapshots.

[0052] It should be noted that the web page snapshot can be understood as the current page information of the product when the buyer places an order to purchase the product, which includes but is not limited to the product name, product price (including the original price of the product, and the discounted price or promotional price of the product when the buyer places an order to purchase the product), product description, product pictures and other information.

[0053] For example, if a buyer receives a damaged product, they can enter text, multiple images, and videos related to the product transaction order on the client. The text can be a refund reason, such as "The product is damaged and cannot be used"; the image can be a picture of the product taken by the buyer; and the video can be a video of the product taken by the buyer.

[0054] Specifically, such as Figure 2 As shown, another embodiment of the present application provides a service system, which includes a server 101 and a client 102. The client 102 is used to obtain the material information provided by the user for the dispute event, and responds to the document generation instruction triggered by the user through the client interface, and sends a request containing the material information to the server 101. The server 101 obtains the material information provided by the user for the dispute event; determines the user's defense intention; when it is determined that supplementary information needs to be obtained based on the defense intention, determines at least one information acquisition channel, and obtains at least one supplementary information through the at least one information acquisition channel; generates a defense document based on the defense intention, the material information and the at least one supplementary information. The client 102 is also used to display the defense document and interactive controls. Among them, the user can use the interactive controls (such as Figure 2 "Confirm" control, "Edit" control in the dialog box) to confirm the defense document and / or edit the defense document.

[0055] In this embodiment, at least one of the following steps implemented by the server 101 may be implemented by a preset model, or by at least two different preset models: Determine the user's defense intent; When it is determined that supplementary information is required based on the defense intention, determining at least one information acquisition channel, and obtaining at least one supplementary information through the at least one information acquisition channel; A defense document is generated according to the defense intention, the material information and at least one supplementary information.

[0056] If the user's objection intent is determined by a preset model, the preset model can obtain or clarify the user's objection intent through interaction with the user (such as dialogue). Alternatively, the preset model can directly identify the user's objection intent based on the material information provided by the user.

[0057] The client 102 can be specifically used to: obtain the material information provided by the user regarding the dispute event; respond to the document generation instruction triggered by the user through the client interface, and display the defense document generated by the preset model; display interactive controls (such as Figure 2 ) so that the user can confirm the defense document and / or edit the defense document through the interactive controls.

[0058] In the above-mentioned embodiments, the server 101 can be a server, a service cluster, a virtual server, or a cloud, etc., and this embodiment does not specifically limit this. The server 101 provides corresponding functional services to the client, such as artificial intelligence services (dispute resolution functions), the generation of documents involved in dispute scenarios (such as defense documents), etc. Users can trigger the server to provide corresponding functional services through the browser, client application (APP), web application H5 (HyperText Markup Language 5, the fifth generation of HTML, Hypertext Markup Language), light application (also known as mini-program, a lightweight application program), or cloud application on the client device. The corresponding device on the client side can be, but is not limited to: a smartphone, a smart wearable device, a tablet computer, a laptop computer, a desktop computer, etc.

[0059] The technical solution provided in this application will be described below in the form of a method embodiment.

[0060] Figure 3 The following is a flow chart of a dispute resolution method provided by an embodiment of the present application. The execution subject of the method provided by this embodiment may be the server in the above-mentioned service system. Of course, in the case where a local preset model is deployed on the client, the execution subject of each step in the following method may also be the client. Alternatively, the execution subject of some steps in the following method is the client, and the execution subject of other steps is the server. Specifically, the method includes: 201. Obtaining material information provided by at least one party to the dispute; 202. Determine at least one information acquisition channel based on the material information using a preset model, and acquire at least one supplementary information through the at least one information acquisition channel; 203. Output a processing result for the dispute event according to the material information and the at least one supplementary information using the preset model.

[0061] The preset model may be a multimodal model, and the material information and / or supplementary information may be multimodal information.

[0062] In the above 201, the parties involved in the dispute may be more than two, but may be three, or even four, in more complex cases. This embodiment of the present application does not specifically limit this. When providing material information, only one party may provide the material information (or evidence information) that it can provide, or both parties may provide the material information they have.

[0063] Taking e-commerce as an example, a dispute event may be related to a product transaction order. The parties involved in the dispute may include the seller and buyer, as well as the logistics service provider, etc. For example, disputes related to product transaction orders may include disputes in the smart refund scenario and disputes in the chargeback scenario.

[0064] The material information provided can be input and / or uploaded by the user through the client. Figure 1 and Figure 2 In the example shown, users can enter text in the "Please describe the dispute in detail" input box and can also click the "Upload" control to upload images (such as captured images, page screenshots, webpage snapshots, etc.), videos, audio, etc. In other words, the material information may include, but is not limited to, at least one of the following: text, images, audio, video, etc.

[0065] In a specific scenario, the material information may include at least two items of text, picture, audio, and video. The preset model is a multimodal model that can support multimodal information input.

[0066] The at least one information acquisition channel in 202 may be selected from a plurality of preset information acquisition channels. For example, the plurality of preset information acquisition channels may include: a knowledge base, a plurality of APIs (by calling different APIs, corresponding functional modules can be activated to acquire information), and the like.

[0067] The functional modules corresponding to multiple APIs may include but are not limited to: search engines, voice translation tools, and data processing tools (such as automated template evidence generation tools).

[0068] Starting the search engine can retrieve the target information in the database on the server side. Taking the e-commerce scenario as an example, the product information of the goods in the commodity transaction order involved in the dispute event, the messages left by the buyer and seller during the transaction, screenshots related to the goods, product usage videos, other attached files (such as pictures captured from the web page), order information related to the goods, order information related to logistics, logistics status information (such as whether it has been delivered, shipped, in transit, etc.), etc. can be retrieved.

[0069] Starting the language translation tool can unify the language of the text in the material information into the same language, such as translating the English text in the material information into Chinese text.

[0070] Starting the data processing tool can process the material information and / or at least one acquired supplementary information, such as extracting key fields and logical relationships between key fields from the material information and / or supplementary information, identifying the type of each key field, etc.; assembling the extracted multiple key fields according to the type of each key field and the logical relationship between the key fields to generate data of a preset structure.

[0071] Pre-structured data takes different forms in different application scenarios. For example, in e-commerce scenarios, for disputes related to product transaction orders, the generated pre-structured data could be a defense document that conforms to the style of dispute defense. In legal litigation, the generated pre-structured data could be a complaint letter or statement of defense that conforms to legal documents. In the finance / payment sector, the generated pre-structured data could be a chargeback defense or dispute rebuttal letter that conforms to financial / payment requirements.

[0072] The technical solution provided in this embodiment utilizes artificial intelligence technology (i.e., a pre-set model) to empower dispute scenarios, replacing or assisting manual dispute handling. This improves dispute resolution speed, efficiency, labor costs, and accuracy. Furthermore, by leveraging the advantages of artificial intelligence technology rather than relying on pre-set dispute resolution rules, this solution can provide differentiated solutions for different dispute cases, offering a degree of flexibility. Furthermore, the pre-set model in this embodiment can also determine at least one information acquisition channel based on material information provided by at least one party to the dispute, and obtain supplementary information through the determined information acquisition channel to enrich the evidentiary information, thereby enhancing the ability of the pre-set model to produce more reasonable dispute resolution outcomes.

[0073] In step 202 above, after identifying the intention of at least one party to the dispute, the information acquisition channels and the required supplementary information can be analyzed and planned based on the intention. Once the intention is clear, the supplementary information related to the intention can be identified as supporting evidence. In a more specific embodiment, the "using a preset model to determine at least one information acquisition channel based on the material information" in step 202 above may include: 2021. Identifying the intention of at least one party to the dispute based on the material information using a preset model; 2022. Analyze and plan at least one information acquisition channel based on the intention of at least one party and the material information.

[0074] In one possible implementation example, step 2022 may include: Analyze and plan required knowledge items and / or functional modules based on the intention of the at least one party and the material information; If the required knowledge items are analyzed and planned, one of the information acquisition channels is determined to be the target knowledge base containing the knowledge items; If the required functional modules are analyzed and planned, one of the information acquisition channels is determined to be the application program interface API for calling the functional module.

[0075] In a feasible specific implementation, in step 202 of this embodiment, “obtaining at least one supplementary information through the at least one information acquisition channel” may include: 2023. Retrieve knowledge information corresponding to the knowledge item in the target knowledge base using the preset model; 2024. Activate the function module by calling the API using the preset model to retrieve target information or process the information to be processed to obtain result information; The information to be processed includes, but is not limited to, at least one of the following: at least part of the material information, at least part of the retrieved knowledge information, and at least part of the retrieved target information. The at least one supplementary information includes at least one of the following: the knowledge information, the target information, and the result information.

[0076] For example, in an e-commerce scenario, one party (e.g., a buyer) purchases a product and demands compensation from the merchant because it doesn't match the product's promotional information. The buyer's intent is to have the merchant refund the purchase price and pay penalties for breach of contract. Based on this intent, analysis can be performed to determine the need for relevant compensation rules (such as those established by the merchant, by the e-commerce platform, or agreed upon between the merchant and the buyer), so that the dispute can be resolved in accordance with these rules. If, based on the intent, the material information is determined to be insufficient to prove that the product doesn't match the merchant's promotional information, supplementary information can be developed to best meet the user's intent and provide a complete chain of evidence. Once the supplementary information required is determined, the required functional modules can be mapped out and activated by calling the corresponding APIs to retrieve this supplementary information. For example, calling the corresponding search engine API can trigger a search to retrieve product promotional images, screenshots of the merchant's promotional webpage, and live streams of the merchant's product from a historical period.

[0077] Taking the e-commerce scenario as an example, the dispute event in this embodiment is related to a product transaction order. Accordingly, the target knowledge base may include, but is not limited to, at least one of the following: a product information library, a rule library, and a sample library. The product information library stores product information for various products. The rule library stores intents and dispute resolution rules associated with those intents. The sample library stores intents and historical dispute resolution examples associated with those intents.

[0078] For example, consider a dispute where a buyer claims product A is counterfeit and requests compensation. The buyer provides material information (such as descriptions, product photos, and third-party inspection reports). The pre-set model can determine the user's intent (compensation) based on the material information provided by the user and, based on the user's intent and the material information provided, determine the need to retrieve counterfeit compensation rules from the knowledge base. Furthermore, the pre-set model can first determine the category to which product A belongs (such as clothing, luggage, or electronics). Different product categories have different corresponding counterfeit compensation rules. Based on the category to which product A belongs, the pre-set model can determine the need to retrieve the corresponding counterfeit compensation rules from the knowledge base.

[0079] Continuing with the above-mentioned dispute incident as an example, the preset model determines, based on the user's intent and the material information provided by the user, that the third-party inspection report needs to be verified for credibility, and that the merchant's promotional page information for the product needs to be obtained from the network, etc. Accordingly, the preset model will retrieve the API corresponding to the credibility verification function module from multiple preset APIs, call the API, and start the credibility verification function module to verify the credibility of the third-party experience report provided by the user. The preset model will also retrieve the API corresponding to the search engine from multiple preset APIs, call the API to start the search engine, and search the network or target database for the merchant's promotional page information for the product.

[0080] If the description and / or third-party inspection report provided by the buyer is in English, the preset model will also determine that the description and / or third-party inspection report needs to be translated into Chinese. At this time, the preset model will retrieve the corresponding API of the translation function module from multiple preset APIs, call the API to activate the translation function module, and translate the description and / or third-party inspection report into Chinese.

[0081] Among them, multiple preset APIs can be stored in a list format.

[0082] That is, in one feasible implementation, in step 2024 of this embodiment, "using the preset model to start the functional module by calling the API to retrieve target information or process the information to be processed to obtain result information" may specifically include at least one of the following: The preset model calls the first API to start the search function module to retrieve the target information to provide supplementary explanation and evidence for the material information; The preset model calls a second API to start a processing module to process the information to be processed and generate data of a preset structure; The preset model calls a third API to start a text translation module to perform language conversion on the information to be processed.

[0083] The specific form of the above-mentioned retrieval function module is a search engine. The preset model can provide the search keywords to the search engine, and the search engine can search for target information related to the keywords.

[0084] The processing module can be set based on scenario requirements. The processing module can process the information to be processed including but not limited to: verifying the authenticity of the information content, identifying and extracting key content from the information content, converting the data format of the information content, etc.

[0085] For example, the processing module may include a structured template understanding submodule and a template content assembly submodule. Accordingly, in the above step, "the processing module processes the information to be processed and generates data of a preset structure" may include: The structured template understanding submodule determines a template according to the intent, identifies fixed and variable parts in the template, the hierarchical structure of the template, dynamic fields to be filled in the template, and constraints corresponding to the dynamic fields, to obtain a template recognition result; The template content assembly submodule performs template content assembly based on the template recognition result and the information to be processed to generate data of a preset structure.

[0086] The fixed portion of a template is immutable, while the variable portion can be modified based on the actual content (this modification can be made by the user or by a pre-set model based on the actual text (e.g., sentence flow and word accuracy). The template's hierarchical structure, for example, includes the number of paragraphs, paragraph types (e.g., explanatory paragraphs, clause paragraphs (e.g., legal documents), data presentation paragraphs (e.g., photos, videos, screenshots), and conclusion paragraphs). Understanding and correctly defining these hierarchical relationships is key to achieving high-quality automated template generation. In practical systems, these structures are typically represented and stored using XML, JSON, or a custom DSL (domain-specific language). Dynamic fields that need to be populated in the template include information such as the cause, evidence, date and time, amount, and the rules and regulations that the claim is based on. In automated template generation, constraints on dynamic fields are a key control mechanism for ensuring the quality and compliance of generated content. Constraints on dynamic fields may include, but are not limited to, data type constraints (e.g., string, number, date and time), format constraints (e.g., text format, number format, date format), logical relationships between dynamic fields, and dependencies.

[0087] The template content assembly submodule identifies and extracts the content in the information to be processed based on the template recognition results; then, based on the hierarchical structure of the template, the constraints corresponding to the dynamic fields, etc., the content in the information to be processed is assembled with the template to obtain data with a preset structure.

[0088] It should be added here that: this article only exemplifies an implementation scheme of the processing module. In fact, the processing module can also be other types of modules, such as content screening and falsification of the information to be processed; identification of pictures or videos in the information to be processed, etc. This embodiment does not make specific limitations on this.

[0089] Furthermore, the method provided in this embodiment further includes the following steps: 204. Send the processing result to the manual review client; 205. After the manual review client confirms the processing result, the processing result is sent to the client corresponding to at least one party of the dispute event.

[0090] The following describes the solution provided in the embodiments of this application by taking the two scenarios of smart refund and defense against chargeback as examples.

[0091] See also Figure 4 In the example shown, the knowledge base may include a product information base, a dynamic rule base, and a high-quality sample base.

[0092] The product information database stores product information for various products. This information may include, but is not limited to, product categories, product attributes, stock keeping units (SKUs), and detailed product descriptions. A SKU is an image used in e-commerce to display the details of a product unit. Detailed product descriptions include, but are not limited to, product name, price (including the original price and any discounts or promotional prices applied when a buyer places an order), product description (detailed information about the product, including its functions, features, instructions for use, and materials), product specifications (such as dimensions, weight, capacity, and ingredients), user reviews, after-sales service (including return and exchange policies and warranty coverage), purchase options (including colors, sizes, and quantities), product instructional videos (which can demonstrate product usage and precautions), compliance information (indicating that the product complies with specific safety and quality standards), and a packing list (listing all items included in the package).

[0093] The dynamic rule base may include multiple dispute intentions and dispute handling rules corresponding to each dispute intention, and the dynamic rule base may be modified according to actual scenarios.

[0094] For example, in the scenario of smart refunds, if the text information in the material information is "the product is damaged and cannot be used", the preset model will understand the user's intention through the material information and determine that the user's intention is that the buyer needs a refund. The dynamic rule library contains the rule that "if the product is damaged and cannot be used at the same time, it should be classified as a damaged product category". In this case, the dispute intention should be identified as "damaged product". In this case, the dispute handling rule is actually a refund decision, such as "if the product is severely damaged, then a 60% refund will be made."

[0095] The high-quality sample library may include multiple dispute intentions and historical dispute handling samples corresponding to each dispute intention. The historical dispute handling samples may refer to historical dispute judgment samples with high quality.

[0096] It can be understood that the preset module in the embodiment of the present application can be a multimodal large language model. Of course, the multimodal large language model can also be replaced by other multimodal models with stronger intent analysis, content understanding and API call functions.

[0097] See also Figure 4 In the example shown, the API list includes multiple APIs related to the two scenarios of smart refund and defense against chargeback, such as the API corresponding to text translation, the API corresponding to the processing module, the API corresponding to evidence material query, the API corresponding to order information query, the API corresponding to logistics information query, and so on.

[0098] The preset model can call the corresponding API for order information query to activate the order information query function to obtain product-related order information (such as the product order number) and logistics-related order information (such as the logistics tracking number and logistics company). The preset model can also call the corresponding API for logistics information query to activate the logistics information query function to obtain logistics information, including whether the product has been successfully delivered and the logistics status, including but not limited to status such as collected, in transit, being delivered, and signed for. The preset model can call the corresponding API for text translation to activate the text translation function to translate text in material information, thus supporting multilingual scenarios. For example, the model can translate text in other languages into English, or vice versa; or translate text in other languages into Chinese, or vice versa. The preset model can call the corresponding API for evidence query to activate the evidence query function to query evidence and supplement the material information provided by at least one party to the dispute. Evidential materials may include, but are not limited to, messages from the buyer and / or seller during the transaction, screenshots of the product or information, videos of the product in use (i.e., videos of the buyer using the product), and attached files (such as product webpages). The pre-set model can invoke the API corresponding to the processing module to activate the structured template submodule and the template content assembly submodule to process the information to be processed, generating data in a pre-set format (such as a rebuttal letter conforming to the specified format). The information to be processed may include knowledge information retrieved from the knowledge base, target information obtained by invoking the corresponding API function, and material information provided by at least one party to the dispute.

[0099] It is understandable that Figure 4 The API list shown includes relevant APIs available for the two scenarios of smart refund and chargeback. For other dispute scenarios, corresponding APIs can also be configured in the API list so that the preset model can be called.

[0100] Suppose the text in the material information provided by a buyer reads, "The product is damaged and cannot be used." For example, the material information also includes photos of the damaged product and the product's order number. The preset model understands the user's intent based on the material information and determines the need for the order information query function and the logistics information query function. The preset model searches the API list for APIs corresponding to the order information query function and the logistics information query function. It then calls the API corresponding to the order information query function to initiate the order query function, querying the product's order information (order time, payment success time, whether it is a gift, etc.). It also calls the API corresponding to the logistics information query function to initiate the logistics information query function, querying the product's logistics tracking number and logistics information (such as online delivery and receipt records). Based on the material information, the preset model understands the user's intent and determines the need for knowledge items related to product damage refund rules. The preset model then searches the knowledge base's dynamic rule library for refund rules corresponding to these knowledge items related to product damage refund rules.

[0101] After obtaining the order information, the product's logistics tracking number, logistics information, and the refund rules corresponding to the knowledge item, the customer's material information, the obtained order information, the product's logistics tracking number, logistics information, and the refund rules corresponding to the knowledge item are input into the preset model. The preset model is executed and outputs a processing result, such as a full refund, a partial refund (such as a 60% refund), etc. The processing result may include the partial refund ratio (such as "60%) and the basis for the refund (such as "the product is severely damaged and may not be usable by the user").

[0102] For example, in a smart refund scenario, if the text in the material information reads "The product is damaged and unusable," the dispute intent is identified as "Merchant damage." However, the preset model, combined with product information (such as positive reviews from other buyers, product usage video instructions, and product compliance information) and the buyer's evidence (such as a video of the product in use), determines that the damage was caused by improper operation by the buyer. Therefore, the preset model may output a refund denial and a reason for the denial (e.g., "The product was damaged due to improper operation by the buyer; the merchant is not responsible for the consequences").

[0103] In another possible scenario, a pre-set model analyzes intent and material information without requiring knowledge items or API calls. The pre-set model can directly output processing results based on the material information entered by the user. For example, for low-priced items (such as those priced below a pre-set price), the pre-set model can directly output a full refund.

[0104] It is understood that in embodiments of the present application, a determination may be made first as to whether a knowledge base search is necessary. If the preset model determines that a knowledge base search is necessary, the preset model may be used to retrieve supplementary information from the knowledge base, and then the preset model may be used to determine whether an API call is necessary. Furthermore, if the preset model determines that a knowledge base search is not necessary, the preset model may be used directly to determine whether an API call is necessary. In other words, a determination may be made first as to whether a knowledge base search is necessary, and then as to whether an API call is necessary.

[0105] Alternatively, embodiments of the present application may first determine whether an API call is necessary. If a preset model determines that an API call is necessary, the preset model may be used to retrieve the target API to be called, then the target API may be called to obtain supplementary information, and then a determination may be made as to whether a knowledge base search is necessary. Furthermore, if a preset model determines that an API call is not necessary, a determination may be made directly as to whether a knowledge base search is necessary. In other words, a determination as to whether an API call is necessary may be made first, followed by a determination as to whether a knowledge base search is necessary.

[0106] The embodiment of the present application does not limit the order of using the preset model to determine whether the knowledge base needs to be searched and whether the API needs to be called.

[0107] In a chargeback scenario, suppose a consumer (i.e., the buyer) files a claim for compensation due to false advertising by a merchant (i.e., the seller). The merchant, arguing that it is not responsible or needs to mitigate its liability, intends to refute the consumer's complaint or claim through a letter of rebuttal (or appeal) providing evidence and reasoning. In this case, the merchant can use the automatic rebuttal text generation function provided by the server through the client. A pre-set model generates the rebuttal text based on the merchant's provided materials. The content of this rebuttal text may include, but is not limited to, text, photos, videos, and links. The merchant's provided materials may include, but are not limited to, complete pre-shipment quality inspection records, packaging videos, original product descriptions captured from product detail pages, and third-party inspection reports. The pre-set model can interact with the merchant to determine the merchant's intention to generate the rebuttal text, such as by entering text into the corresponding input box on the client interface. Alternatively, the merchant can enter a rebuttal text generation instruction (i.e., the merchant's intention) through the client (e.g., by touching the "Generate Rebuttal Text" control). Upon receiving this instruction, the server enters the rebuttal text generation instruction as a prompt when invoking the pre-set model for execution. Alternatively, the preset model identifies the merchant's intention based on the material information provided by the merchant. The preset model can directly generate a defense text based on the material information provided by the merchant. Of course, in order to make the defense text generated by the preset model more professional, see Figure 4As shown, the embodiment of the present application is also designed with a processing module, which includes structured template understanding and template content assembly. The server side can be pre-installed with templates for various texts, such as defense letter templates, rebuttal letter templates, objection letter templates, etc. corresponding to different types of disputes. After the preset model determines the merchant's intention (i.e., the defense intention), it can search the API list to retrieve the API corresponding to the processing module; then call this API to start the processing module to obtain the defense letter template corresponding to the "merchant false advertising claim" dispute. The structured template understanding submodule parses the defense letter template to identify the fixed and variable parts of the defense letter template, the hierarchical structure of the defense letter template, the dynamic fields that need to be filled in the defense letter template, and the constraints corresponding to the dynamic fields. This facilitates the template content assembly submodule to identify and extract relevant content from the material information provided by the merchant (or also includes at least one supplementary information obtained by the preset model from the knowledge base and / or through the API corresponding to various information query functions), and then assembles the extracted content with the defense letter template to generate a defense letter. The rebuttal letter generated by the processing module can be input into a preset model, and the preset model can be used to adjust the rebuttal letter, such as sentence coherence, word accuracy, rigor, logic, etc.

[0108] Furthermore, in another feasible technical solution, the server generates a corresponding dispute ticket for the rebuttal letter generated by the processing module. The preset model retrieves the dispute ticket offline and then, based on the associated rebuttal letter and the consumer's dispute claim, handles the dispute and outputs the resolution.

[0109] For example, the corresponding processing results of the defense chargeback may include: the buyer user wins the case (that is, the seller user needs to bear the responsibility), the seller user wins the case (that is, the buyer user needs to bear the responsibility), or the responsibility is shared in proportion (such as the buyer user pays 70% and the seller user refunds 30%), etc.

[0110] The embodiments of the present application can also support multilingual scenarios, that is, the text, pictures, videos and audio included in the material information can be in multiple languages. Supplementary information retrieved through various query function APIs, such as product screenshots, messages, etc., can also be in multiple languages. When there are languages other than the first language in the material information and / or supplementary information, the preset model can start the text translation function by calling the API corresponding to the text translation to translate it into the first language. The first language can be Chinese, English, etc. For pictures, videos, etc., OCR technology can be used to recognize the text in the pictures and videos, and then translate it into the first language. For audio, language recognition technology can be used to recognize speech to generate text, and then the speech text can be translated into the first language.

[0111] It is understandable that the embodiments of the present application may involve multiple languages, any of which may be used as the first language.

[0112] The dispute scenarios of the embodiments of the present application may be scenarios such as smart refunds and defense and refusal to pay. In the defense and refusal to pay scenario, the response to the evidence and the templated defense letter can be automatically generated based on the material information input or uploaded by the user and the supplementary information obtained through the knowledge base and / or API, thereby improving the efficiency and quality of the seller user in preparing the defense materials, thereby reducing the user appeal rate and the seller's loss rate. In the scenario of smart refunds, in the process of the user initiating a refund, based on the material information input or uploaded by the user and the supplementary information obtained through the knowledge base and / or API, the decision on the dispute event is automatically made or assisted in making a decision on the dispute event, thereby improving the efficiency of dispute handling and reducing labor costs, as well as improving the accuracy of dispute handling.

[0113] Figure 5 A flowchart of a dispute resolution method provided by another embodiment of the present application is shown. The execution subject of the method provided by this embodiment may be the server in the above-mentioned service system. Of course, in the case where a local preset model is deployed on the client, the execution subject of each step in the following method may also be the client. Alternatively, the execution subject of some steps in the following method is the client, and the execution subject of other steps is the server. Specifically, the method includes: 301. Obtaining materials and information provided by at least one party to the dispute; 302. Determine the intention of at least one party to the dispute based on the material information; 303. When it is determined that supplementary information needs to be obtained based on the intention, determining at least one information acquisition channel, and obtaining at least one supplementary information through the at least one information acquisition channel; 304. Perform dispute resolution on the dispute event based on the material information and the at least one supplementary information.

[0114] At least one of the above steps 302, 303 and 304 can be implemented using a preset model. Specific details of implementing the above steps using the preset model can be found above and will not be described here in detail.

[0115] In addition to using pre-set models, steps 302, 303, and 304 can also be implemented using other tools. For example, intent recognition in step 302 can be implemented using rule-based methods, such as keyword matching, regular expressions, or decision trees. Alternatively, specific algorithms such as Bayesian and random forests can be used.

[0116] In step 303, determining whether supplementary information is required based on the intent can be determined based on pre-set rules. For example, multiple intents can be pre-configured, along with the required evidentiary information for each intent. In practice, the evidentiary information corresponding to the determined intent can be obtained. The evidentiary information can then be compared with the material information obtained in step 301, and the need for supplementary information determined based on the comparison results. If supplementary information is required, the source (e.g., a knowledge base or API search) can be determined based on the supplementary information.

[0117] In step 304, the dispute resolution rules may be pre-configured before implementation. When making a dispute decision, the dispute resolution rules are matched against the material information and at least one supplementary information, and a decision is made based on the matched rule items to obtain a dispute resolution result.

[0118] Figure 6 The following is a flow chart of a method for generating a defense document provided by an embodiment of the present application. The execution subject of the method provided by this embodiment can be the server or the client in the above-mentioned service system. Alternatively, the execution subject of some steps in the following method is the client, and the execution subject of other steps is the server. Specifically, the method includes: 401. Obtain the materials and information provided by the user regarding the dispute; 402. Determine the user's defense intention; 403. When it is determined that supplementary information is required based on the defense intention, determining at least one information acquisition channel, and obtaining at least one supplementary information through the at least one information acquisition channel; 404. Generate a defense document based on the defense intention, the material information, and at least one supplementary information.

[0119] At least one of the above steps 402, 403 and 404 can be implemented using a preset model. Specific details of implementing the above steps using the preset model can be found above and will not be described here in detail.

[0120] In addition to utilizing pre-set models, steps 402, 403, and 404 can also be implemented using other tools. For example, the identification of adversarial intent in step 402 can be implemented using rule-based methods, such as keyword matching, regular expressions, or decision trees. Alternatively, specific algorithms such as Bayesian and random forest algorithms can be used.

[0121] In step 403, determining whether supplementary information is necessary based on the defense intent can be determined based on pre-set rules. For example, multiple intents can be pre-configured, along with the required evidentiary information for each intent. In practice, the corresponding evidentiary information can be obtained based on the determined intent. The evidentiary information can then be compared with the material information obtained in step 401, and the need for supplementary information determined based on the comparison results. If supplementary information is necessary, the source (e.g., a knowledge base or API search) can be determined based on the supplementary information.

[0122] The above step 404 can be implemented by using the processing module mentioned above. The details can be found in the above description and will not be repeated here.

[0123] In a specific embodiment, the above step 402 "determining the user's defense intention" can be implemented using a preset model. Specifically, it includes: The preset model obtains the user's defense intention by interacting with the user; or The preset model identifies the user's defense intention based on the material information.

[0124] In the above step 403, "determining at least one information acquisition channel and acquiring at least one supplementary information through the at least one information acquisition channel" may include: Analyze and plan the required knowledge items and / or functional modules based on the defense intention and the material information; If the required knowledge items are analyzed and planned, one of the information acquisition channels is determined to be the target knowledge base containing the knowledge items; If the required functional modules are analyzed and planned, one of the information acquisition channels is determined to be the API for calling the functional modules; The at least one supplementary information is obtained based on the target knowledge base and / or the API.

[0125] In a specific embodiment, the above-mentioned “obtaining the at least one supplementary information based on the target knowledge base and / or the API” may include: Retrieving knowledge information corresponding to the knowledge item in the target knowledge base using the preset model; Using the preset model to start the functional module by calling the API to retrieve target information or process the information to be processed to obtain result information; The information to be processed includes at least one of the following: at least part of the material information, at least part of the retrieved knowledge information, and at least part of the retrieved target information; The at least one supplementary information includes at least one of the following: the knowledge information, the target information, and the result information.

[0126] The technical solution provided in the embodiments of this application determines the user's defense intent after obtaining the material information of the dispute incident provided by the user. When it is determined based on the defense intent that supplementary information is needed, the solution can determine at least one information acquisition channel and obtain the supplementary information through the at least one determined information acquisition channel to enrich the content of the defense document. Finally, a defense document is generated based on the defense intent, the material information, and the at least one supplementary information. Throughout the entire process, the user only needs to provide the material information, and the defense document will be automatically generated, which provides convenience for the user's defense and reduces the difficulty of the defense.

[0127] Figure 7 A flowchart of a method for generating a defense document provided in another embodiment of the present application is shown. The method provided in this embodiment is executed by a client. Specifically, the method includes: 501. Obtaining materials and information provided by users regarding dispute incidents; 502. Responding to a document generation instruction triggered by the user through the client interface, displaying a defense document generated by a preset model; 503. Displaying an interactive control so that the user can confirm the defense document and / or edit the defense document through the interactive control; In which, the defense document is generated by the preset model based on the material information, or is generated by the preset model based on the material information and at least one supplementary information; the at least one supplementary information is related to the dispute event and is obtained by the preset model through at least one information acquisition channel.

[0128] In this embodiment, the relevant content of the defense document generated by the preset model can be found in the corresponding description above and will not be repeated here.

[0129] It should be noted that the execution entity of each step of the method provided in the above embodiment can be the same device, or the dispute resolution method can be executed by different devices. For example, the execution entity of steps 201 to 203 can be device A; for another example, the execution entity of step 201 can be device A, and the execution entities of steps 202 and 203 can be device B, and so on.

[0130] In addition, in some of the processes described in the above embodiments and the accompanying drawings, multiple operations that appear in a specific order are included, but it should be clearly understood that these operations may not be executed in the order in which they appear in this article or may be executed in parallel. The serial numbers of the operations, such as 201, 202, etc., are only used to distinguish between different operations, and the serial numbers themselves do not represent any execution order. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions of "first", "second", etc. in this article are used to distinguish different messages, devices, modules, etc., and do not represent a sequential order, nor do they limit "first" and "second" to different types.

[0131] An embodiment of a device of the present application provides a dispute resolution device. The dispute resolution device includes a first acquisition module, a second acquisition module, and an execution module. The first acquisition module is configured to acquire material information provided by at least one party to a dispute. The second acquisition module is configured to utilize a preset model to determine at least one information acquisition channel based on the material information, and to acquire at least one supplementary information through the at least one information acquisition channel. The execution module is configured to utilize the preset model to output a resolution result for the dispute based on the material information and the at least one supplementary information.

[0132] Optionally, when the second acquisition module uses a preset model to determine at least one information acquisition channel based on the material information, and obtains at least one supplementary information through the at least one information acquisition channel, it is specifically used to: use the preset model to identify the intention of at least one party to the dispute event based on the material information; analyze and plan at least one information acquisition channel based on the intention of the at least one party and the material information. Furthermore, the second acquisition module is also used to: analyze and plan the required knowledge items and / or functional modules based on the intention of the at least one party and the material information; if the required knowledge items are analyzed and planned, determine one of the information acquisition channels to be the target knowledge base containing the knowledge items; if the required functional modules are analyzed and planned, determine one of the information acquisition channels to be the application program interface API for calling the functional module.

[0133] Furthermore, when the second acquisition module acquires at least one supplementary information through the at least one information acquisition channel, it is specifically configured to: use the preset model to retrieve the knowledge information corresponding to the knowledge item in the target knowledge base; and use the preset model to activate the function module by calling the API to retrieve the target information or process the information to be processed to obtain result information. The information to be processed includes at least one of the following: at least part of the material information, at least part of the retrieved knowledge information, and at least part of the retrieved target information. The at least one supplementary information includes at least one of the following: the knowledge information, the target information, and the result information.

[0134] Furthermore, when the second acquisition module uses the preset model to start the function module by calling the API to retrieve the target information or process the information to be processed to obtain result information, it performs at least one of the following: Using the preset model to call a first API to start a search function module, retrieve the target information to provide supplementary explanation and evidence for the material information; Using the preset model to call a second API to start a processing module to process the information to be processed and generate data of a preset structure; The preset model is used to call a third API to start a text translation module to perform language conversion on the information to be processed.

[0135] Furthermore, when the processing module processes the information to be processed and generates data of a preset structure, the second acquisition module is specifically configured to: calling a structured template understanding submodule, which determines a template based on the intent, identifies fixed and variable parts in the template, the hierarchical structure of the template, dynamic fields to be filled in the template, and constraints corresponding to the dynamic fields, to obtain a template recognition result; The template content assembly submodule is called, and the template content assembly submodule performs template content assembly based on the template recognition result and the information to be processed to generate data of a preset structure.

[0136] Optionally, the dispute resolution device provided in this embodiment may further include a sending module. The sending module is configured to send the processing result to a manual review client; after the manual review client confirms the processing result, the sending module sends the processing result to a client corresponding to at least one party to the dispute event.

[0137] Another device embodiment of the present application provides a dispute resolution device. The dispute resolution device includes: a first acquisition module, a determination module, a second acquisition module, and an execution module. The first acquisition module is used to obtain material information provided by at least one party to the dispute event. The determination module is used to determine the intention of at least one party to the dispute event based on the material information. The second acquisition module is used to determine at least one information acquisition channel when it is determined that supplementary information needs to be obtained based on the intention, and obtain at least one supplementary information through the at least one information acquisition channel. The execution module is used to perform dispute resolution on the dispute event based on the material information and the at least one supplementary information.

[0138] The present application also provides a device for generating a defense document in accordance with an embodiment of a device. The defense document generation device includes: a first acquisition module, a determination module, a second acquisition module, and a generation module. The first acquisition module is used to obtain material information provided by the user regarding the dispute event. The determination module is used to determine the user's defense intention. The second acquisition module is used to determine at least one information acquisition channel when it is determined that supplementary information needs to be obtained based on the defense intention, and obtain at least one supplementary information through the at least one information acquisition channel. The generation module is used to generate a defense document based on the defense intention, the material information, and the at least one supplementary information.

[0139] Optionally, when determining the user's defense intention, the determination module uses a preset model to determine the defense intention. Specifically, the preset model obtains the user's defense intention by interacting with the user; or the preset model identifies the user's defense intention based on the material information.

[0140] Optionally, when the second acquisition module determines at least one information acquisition channel and acquires at least one supplementary information through the at least one information acquisition channel, it is specifically used to: analyze and plan the required knowledge items and / or functional modules based on the defense intention and the material information; if the required knowledge items are analyzed and planned, determine one of the information acquisition channels as the target knowledge base containing the knowledge items; if the required functional modules are analyzed and planned, determine one of the information acquisition channels as the API for calling the functional module; and acquire the at least one supplementary information based on the target knowledge base and / or the API.

[0141] Furthermore, when the second acquisition module acquires the at least one supplementary information based on the target knowledge base and / or the API, it is specifically configured to: use the preset model to retrieve the knowledge information corresponding to the knowledge item in the target knowledge base; and use the preset model to activate the function module by calling the API to retrieve the target information or process the information to be processed to obtain result information. The information to be processed includes at least one of the following: at least some information items in the material information, at least some information in the retrieved knowledge information, and at least some information in the retrieved target information. The at least one supplementary information includes at least one of the following: the knowledge information, the target information, and the result information.

[0142] Another embodiment of the present application provides a device for generating a defense document. The device includes an acquisition module and a display module. The acquisition module is configured to obtain user-provided material information regarding a dispute. The display module is configured to respond to a document generation instruction triggered by the user through a client interface, display a defense document generated by a preset model, and display interactive controls so that the user can confirm and / or edit the defense document through the interactive controls.

[0143] The defense document is generated by the preset model based on the material information, or is generated by the preset model based on the material information and at least one supplementary information; the at least one supplementary information is related to the dispute event and is obtained by the preset model through at least one information acquisition channel.

[0144] It should be noted that the various devices provided in the above embodiments can implement the technical solutions described in the corresponding method embodiments, and their implementation principles and technical effects are not described in detail here. For the specific methods for executing the operations of the various modules or units in the various devices provided in the above embodiments, please refer to the relevant contents in the corresponding method embodiments, and no further details will be given here.

[0145] Figure 8 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. In one possible implementation, the structure of each of the above-mentioned devices (such as a dispute resolution device, a defense document generation device) can be implemented as an electronic device, which can be a server, and the server can be a single server, a service cluster composed of multiple servers, a cloud server, or a virtual server, etc. This embodiment of the present application does not specifically limit this. Alternatively, the electronic device can also be a client device, such as a smart phone, a desktop computer, a laptop computer, a smart wearable device, etc. Figure 8 As shown, the electronic device may include a memory 601 and a processor 602 .

[0146] Memory 601 is used to store computer programs and can be configured to store various other data to support operations on the electronic device. Examples of such data include instructions for any application or method operating on the electronic device, data structures, contact data, phone book data, messages, images, videos, etc.

[0147] The processor 602 is coupled to the memory 601 and is configured to execute the computer program stored in the memory 601 to implement the steps in the above method embodiments.

[0148] Further, if Figure 8 As shown, the electronic device also includes: a communication component 603, a power component 604, a display 605, an audio component 606 and other components. Figure 8 Only some components are shown schematically, which does not mean that the electronic device only includes Figure 8 In addition, Figure 8 The components in the dotted box are optional components, not mandatory components, and the specific components depend on the product form of the working node. The working node of this embodiment can be implemented as a terminal device such as a desktop computer, laptop computer, smart phone or IOT device, or a server device such as a conventional server, cloud server or server array. If the working node of this embodiment is implemented as a terminal device such as a desktop computer, laptop computer, smart phone, etc., it can include Figure 8 If the working node of this embodiment is implemented as a server device such as a conventional server, a cloud server or a server array, it may not include Figure 8 Components within the dotted box.

[0149] The above-mentioned memory can be implemented by any type of volatile or non-volatile memory 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 memory, flash memory, magnetic disk or optical disk.

[0150] Accordingly, embodiments of the present application further provide a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, the processor is enabled to implement each step of the above-mentioned method embodiments. The computer-readable storage medium includes volatile or non-volatile storage, or a combination thereof, and may be removable or non-removable. Examples of computer-readable storage media include, but are not limited to, phase-change random access memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, tape, disk storage or other magnetic storage devices, or any other non-transmission medium.

[0151] Accordingly, an embodiment of the present application further provides a computer program product, which includes a computer program or instructions. When the computer program or instructions are executed by a processor, the processor is enabled to implement the steps in the above-mentioned method embodiments. It should be understood that each process or a combination of multiple processes in the above-mentioned method flow can be implemented by a computer program or instruction. In addition, these computer programs or instructions can be applied to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device, so that the processor of the general-purpose computer, the special-purpose computer, the embedded processor or other programmable data processing device can be implemented as a device for implementing the corresponding functions in the above-mentioned method embodiments.

[0152] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a series of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, commodity, or apparatus that includes the element.

[0153] The above are merely embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should all be included within the scope of the claims of the present application.

Claims

1. A dispute resolution method, characterized in that: include: Obtain material information provided by at least one party to the dispute; Determining at least one information acquisition channel based on the material information using a preset model, and acquiring at least one supplementary information through the at least one information acquisition channel; Outputting a processing result for the dispute event based on the material information and the at least one supplementary information using the preset model; Wherein, the preset model is a multimodal model, and the material information and / or the supplementary information is multimodal information.

2. The method according to claim 1, characterized in that Determining at least one information acquisition channel based on the material information using a preset model, including: Identifying the intention of at least one party to the dispute event based on the material information using the preset model; At least one information acquisition channel is analyzed and planned based on the intention of the at least one party and the material information.

3. The method according to claim 2, characterized in that Analyze and plan at least one information acquisition channel based on the intention of the at least one party and the material information, including: Analyze and plan required knowledge items and / or functional modules based on the intention of the at least one party and the material information; If the required knowledge items are analyzed and planned, one of the information acquisition channels is determined to be the target knowledge base containing the knowledge items; If the required functional modules are analyzed and planned, it is determined that one of the information acquisition channels is to call the application program interface API of the functional module.

4. The method according to claim 3, characterized in that Acquiring at least one supplementary information through the at least one information acquisition channel includes: Retrieving knowledge information corresponding to the knowledge item in the target knowledge base using the preset model; Using the preset model to start the functional module by calling the API to retrieve target information or process the information to be processed to obtain result information; The information to be processed includes at least one of the following: at least part of the material information, at least part of the retrieved knowledge information, and at least part of the retrieved target information; The at least one supplementary information includes at least one of the following: the knowledge information, the target information, and the result information.

5. The method according to claim 3, characterized in that The dispute is related to a commodity trading order; The target knowledge base includes at least one of the following: a product information base, a rule base, and a sample base; The commodity information database stores commodity information of multiple commodities; The rule base stores intentions and dispute resolution rules associated with the intentions; The sample library stores intentions and historical dispute handling samples associated with the intentions.

6. The method according to claim 4, characterized in that The function module is started by calling the API using the preset model to retrieve target information or process the information to be processed to obtain result information, including at least one of the following: The preset model calls the first API to start the search function module to retrieve the target information to provide supplementary explanation and evidence for the material information; The preset model calls a second API to start a processing module to process the information to be processed and generate data of a preset structure; The preset model calls a third API to start a text translation module to perform language conversion on the information to be processed.

7. The method according to claim 6, characterized in that The processing module includes: a structured template understanding submodule and a template content assembly submodule; and The processing module processes the information to be processed and generates data of a preset structure, including: The structured template understanding submodule determines a template according to the intent, identifies fixed and variable parts in the template, the hierarchical structure of the template, dynamic fields to be filled in the template, and constraints corresponding to the dynamic fields, to obtain a template recognition result; The template content assembly submodule performs template content assembly based on the template recognition result and the information to be processed to generate data of a preset structure.

8. The method according to any one of claims 1 to 7, characterized in that Also includes: Sending the processing result to the manual review client; After the manual review client confirms the processing result, the processing result is sent to the client corresponding to at least one party of the dispute event.

9. A dispute resolution method, characterized in that: include: Obtain material information provided by at least one party to the dispute; Determining the intention of at least one party to the dispute based on the material information; When it is determined that supplementary information needs to be obtained according to the intention, determining at least one information acquisition channel, and obtaining at least one supplementary information through the at least one information acquisition channel; The dispute event is handled according to the material information and the at least one supplementary information.

10. The method according to claim 9, characterized in that At least one of the following steps is implemented by a preset model: Determining the intention of at least one party to the dispute based on the material information; When it is determined that supplementary information needs to be obtained according to the intention, determining at least one information acquisition channel, and obtaining at least one supplementary information through the at least one information acquisition channel; The dispute event is handled according to the material information and the at least one supplementary information.

11. A method for generating a defense document, characterized in that: include: Obtaining materials and information provided by users regarding dispute incidents; Determine the user's defense intent; When it is determined that supplementary information is required based on the defense intention, determining at least one information acquisition channel, and obtaining at least one supplementary information through the at least one information acquisition channel; A defense document is generated according to the defense intention, the material information and the at least one supplementary information.

12. The method according to claim 11, characterized in that Determine the user's defense intent, including: The preset model obtains the user's defense intention by interacting with the user; or The preset model identifies the user's defense intention based on the material information.

13. The method according to claim 11 or 12, characterized in that Determining at least one information acquisition channel, and acquiring at least one supplementary information through the at least one information acquisition channel, includes: Analyze and plan the required knowledge items and / or functional modules based on the defense intention and the material information; If the required knowledge items are analyzed and planned, one of the information acquisition channels is determined to be the target knowledge base containing the knowledge items; If the required functional modules are analyzed and planned, one of the information acquisition channels is determined to be the API for calling the functional modules; The at least one supplementary information is obtained based on the target knowledge base and / or the API.

14. The method according to claim 13, characterized in that Acquiring the at least one supplementary information based on the target knowledge base and / or the API includes: Retrieving knowledge information corresponding to the knowledge item in the target knowledge base; The function module is started by calling the API to retrieve target information or process the information to be processed to obtain result information; The information to be processed includes at least one of the following: at least part of the material information, at least part of the retrieved knowledge information, and at least part of the retrieved target information; The at least one supplementary information includes at least one of the following: the knowledge information, the target information, and the result information.

15. A method for generating a defense document, characterized in that: include: Obtaining materials and information provided by users regarding dispute incidents; In response to the document generation instruction triggered by the user through the client interface, the defense document generated by the preset model is displayed; Displaying interactive controls so that the user can confirm the defense document and / or edit the defense document through the interactive controls; The defense document is generated by the preset model based on the material information, or is generated by the preset model based on the material information and at least one supplementary information; The at least one supplementary information is related to the dispute event and is obtained by the preset model through at least one information acquisition channel.

16. A service system, characterized in that: include: The client is used to respond to user operations, obtain material information corresponding to the dispute event, and send the material information to the server; A server, configured to implement the steps of the dispute resolution method according to any one of claims 1 to 8, or the steps of the dispute resolution method according to claim 9 or 10; The client is also used to display the processing results returned by the server.

17. A service system, characterized in that: include: The client is used to obtain the material information provided by the user in response to the dispute event, respond to the document generation instruction triggered by the user through the client interface, and send a request containing the material information to the server; A server, configured to implement the steps of the method for generating a defense document according to any one of claims 11 to 14, and send the defense document to the client; The client is further used to display the defense document and interactive controls; The user can confirm the defense document and / or edit the defense document through the interactive control.

18. An electronic device, characterized in that: include: A memory and a processor, wherein the memory is used to store a program; The processor is coupled to the memory and is used to execute the program stored in the memory to implement the steps of the dispute resolution method as described in any one of claims 1 to 10, or to implement the steps of the defense document generation method as described in any one of claims 11 to 15.

19. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program or instructions. When the computer program or instructions are executed, the steps of the dispute resolution method as described in any one of claims 1 to 10 are implemented, or the steps of the defense document generation method as described in any one of claims 11 to 15 are implemented.

20. A computer program product, characterized in that It comprises a computer program which, when run, enables a computer to execute the steps of the dispute resolution method according to any one of claims 1 to 10, or implement the steps of the defense document generation method according to any one of claims 11 to 15.

Images