Qualitative analysis method, device and equipment for transaction risk

Preliminary and qualitative analysis of transaction risks through a large language model, combined with the evidence provided by users and the complained party, the problem of merchants being highly passive in complaint handling is solved, and more efficient and accurate complaint handling and cost reduction are achieved. .

CN118886908APending Publication Date: 2024-11-01ALIPAY COM CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410917825.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-09
Publication Date
2024-11-01

AI Technical Summary

Technical Problem

In the existing technology, merchants are passive in handling complaints and lack opportunities for proactive defense, resulting in high labor costs and low complaint handling efficiency.

Method used

Using a qualitative analysis method of transaction risk based on a large language model, a preliminary and qualitative analysis is conducted by receiving complaint requests submitted by users, providing evidence and feedback mechanisms, allowing the complained party to appeal and provide second evidence information, based on the information of both parties. Reanalyze.

Benefits of technology

It reduces labor costs, improves the efficiency and accuracy of complaint handling, allows merchants to actively defend, and reduces platform operation costs and user participation incentive costs.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118886908A_ABST
    Figure CN118886908A_ABST
Patent Text Reader

Abstract

The embodiment of the invention discloses a qualitative analysis method, device and equipment for transaction risks, and the method comprises the steps: receiving a transaction complaint request which is submitted by a user and aims at a target transaction, and the transaction complaint request comprises a complaint risk type of the target transaction and first proof information aiming at the complaint risk type; inputting the complaint risk type and the first proof information into a pre-trained large model to obtain a preliminary qualitative analysis result aiming at whether the target transaction has a risk corresponding to the complaint risk type; sending the preliminary qualitative analysis result to a complained party in the target transaction, and receiving feedback information of the complained party for the preliminary qualitative analysis result; and if the feedback information comprises a complaint request of the complaint party, obtaining second proof information provided by the complaint party for the complaint request, and performing secondary qualitative analysis on whether the target transaction has a risk corresponding to the complaint risk type through a large model based on the first proof information and the second proof information.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This document relates to the field of computer technology, and in particular to a qualitative analysis method, device and equipment for transaction risks. Background Art

[0002] Financial applications provide many financial service functions (such as payment and transfer functions) for merchants and buyers of various businesses, adapting to various business scenarios such as online, offline, chain, and pre-authorization. The buyer scans the code or triggers the specified financial service function in the application provided by the merchant. The terminal device calls up the corresponding interface of the financial application to complete the processing of the financial service to obtain physical or virtual goods or services.

[0003] In the above process, the buyer may be dissatisfied with or disapproval of the goods or services obtained, and people are paying more and more attention to their own privacy data. Financial applications will open up channels such as "reporting centers" or complaint portals to allow users to provide feedback on merchants, transaction channels, and detailed events, report or complain about transactions that they do not approve of, and submit proof related to the transaction, and wait for the results of the complaint. However, from the merchant's perspective, the current process of handling complaints can include proactive refunds, negotiations with users and proactive contact with users to withdraw complaints, business rectification, etc. The above methods are all passive. Therefore, it is necessary to provide a better qualitative analysis solution for transaction risks, so as to provide merchants with an opportunity to actively defend themselves, so as to re-qualitatively analyze the buyer's reports and complaints, and reduce labor costs at the same time. Summary of the invention

[0004] The purpose of the embodiments of this specification is to provide a better qualitative analysis solution for transaction risks, thereby providing merchants with an opportunity to actively defend themselves, to re-qualitatively analyze the buyer's reported complaints, and to reduce labor costs.

[0005] In order to implement the above technical solution, the embodiments of this specification are implemented as follows:

[0006] The embodiment of this specification provides a qualitative analysis method of transaction risk, the method comprising: receiving a transaction complaint request submitted by a user for a target transaction, the transaction complaint request including the complaint risk type of the target transaction and first evidence information for the complaint risk type. The complaint risk type and the first evidence information are input into a pre-trained large model to obtain a preliminary qualitative analysis result of whether the target transaction has a risk corresponding to the complaint risk type, and the large language model is used to perform a qualitative analysis of whether the transaction has a preset risk. The preliminary qualitative analysis result is sent to the complained party in the target transaction, and feedback information from the complained party on the preliminary qualitative analysis result is received. If the feedback information includes a complaint request from the complained party, the second evidence information provided by the complained party for the complaint request is obtained, and based on the first evidence information and the second evidence information, a secondary qualitative analysis is performed on whether the target transaction has a risk corresponding to the complaint risk type through the large model to obtain a secondary qualitative analysis result.

[0007] The embodiment of this specification provides a qualitative analysis device for transaction risk, the device comprising: a request module, receiving a transaction complaint request submitted by a user for a target transaction, the transaction complaint request including the complaint risk type of the target transaction and the first evidence information for the complaint risk type. A preliminary qualitative analysis module, inputting the complaint risk type and the first evidence information into a pre-trained large model, obtaining a preliminary qualitative analysis result of whether the target transaction has a risk corresponding to the complaint risk type, and the large language model is used to perform a qualitative analysis of whether the transaction has a preset risk. A feedback receiving module, sending the preliminary qualitative analysis result to the complained party in the target transaction, and receiving feedback information from the complained party for the preliminary qualitative analysis result. A secondary qualitative analysis module, if the feedback information includes the complaint request of the complained party, obtaining the second evidence information provided by the complained party for the complaint request, and based on the first evidence information and the second evidence information, performing a secondary qualitative analysis of whether the target transaction has a risk corresponding to the complaint risk type through the large model to obtain a secondary qualitative analysis result.

[0008] The embodiment of the present specification provides a qualitative analysis device for transaction risk, the qualitative analysis device for transaction risk includes: a processor; and a memory arranged to store computer executable instructions, the executable instructions when being executed, enable the processor to: receive a transaction complaint request for a target transaction submitted by a user, the transaction complaint request includes the complaint risk type of the target transaction and the first evidence information for the complaint risk type. The complaint risk type and the first evidence information are input into a pre-trained large model to obtain a preliminary qualitative analysis result of whether the target transaction has a risk corresponding to the complaint risk type, and the large language model is used to perform a qualitative analysis of whether the transaction has a preset risk. The preliminary qualitative analysis result is sent to the complained party in the target transaction, and feedback information from the complained party on the preliminary qualitative analysis result is received. If the feedback information includes a complaint request from the complained party, the second evidence information provided by the complained party for the complaint request is obtained, and based on the first evidence information and the second evidence information, a secondary qualitative analysis is performed on whether the target transaction has a risk corresponding to the complaint risk type through the large model to obtain a secondary qualitative analysis result.

[0009] The embodiment of the specification also provides a storage medium, the storage medium is used to store computer executable instructions, and the executable instructions implement the following process when executed by the processor: receiving a transaction complaint request submitted by a user for a target transaction, the transaction complaint request includes the complaint risk type of the target transaction and the first evidence information for the complaint risk type. The complaint risk type and the first evidence information are input into the pre-trained large model to obtain a preliminary qualitative analysis result of whether the target transaction has a risk corresponding to the complaint risk type, and the large language model is used to perform a qualitative analysis on whether the transaction has a preset risk. The preliminary qualitative analysis result is sent to the complained party in the target transaction, and feedback information from the complained party on the preliminary qualitative analysis result is received. If the feedback information includes the complaint request of the complained party, the second evidence information provided by the complained party for the complaint request is obtained, and based on the first evidence information and the second evidence information, a secondary qualitative analysis is performed on whether the target transaction has a risk corresponding to the complaint risk type through the large model to obtain a secondary qualitative analysis result.

[0010] The embodiment of this specification also provides a computer program product, including a computer program, which implements the following process when executed by a processor: receiving a transaction complaint request submitted by a user for a target transaction, wherein the transaction complaint request includes the complaint risk type of the target transaction and the first evidence information for the complaint risk type. The complaint risk type and the first evidence information are input into a pre-trained large model to obtain a preliminary qualitative analysis result of whether the target transaction has a risk corresponding to the complaint risk type, and the large language model is used to perform a qualitative analysis of whether the transaction has a preset risk. The preliminary qualitative analysis result is sent to the complained party in the target transaction, and feedback information from the complained party on the preliminary qualitative analysis result is received. If the feedback information includes a complaint request from the complained party, the second evidence information provided by the complained party for the complaint request is obtained, and based on the first evidence information and the second evidence information, a secondary qualitative analysis is performed on whether the target transaction has a risk corresponding to the complaint risk type through the large model to obtain a secondary qualitative analysis result. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the technical solutions in the embodiments of this specification or the prior art, the drawings required for use in the embodiments or the prior art description will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this specification. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative labor.

[0012] Figure 1 This is a schematic diagram of the structure of a qualitative analysis system for transaction risks in this manual;

[0013] Figure 2 This is an example of a qualitative analysis method for transaction risk in this specification;

[0014] Figure 3 This is another example of a qualitative analysis method for transaction risk in this specification;

[0015] Figure 4 This is another embodiment of a qualitative analysis method of transaction risk in this specification;

[0016] Figure 5 This is another embodiment of a qualitative analysis method of transaction risk in this specification;

[0017] Figure 6 This is an embodiment of a qualitative analysis device for transaction risk in this specification;

[0018] Figure 7 This is an embodiment of a qualitative analysis device for transaction risks in this specification. DETAILED DESCRIPTION

[0019] The embodiments of this specification provide a method, device and equipment for qualitative analysis of transaction risks.

[0020] In order to enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the drawings in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Based on the embodiments in this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this specification.

[0021] The embodiments of this specification provide a qualitative analysis mechanism based on a large model in a complaint or reporting scenario. The financial application provides many financial service functions (such as payment function, transfer function, etc.) for merchants and buyers of various businesses, and is suitable for various business scenarios such as online, offline, chain, and pre-authorization. Figure 1It is a system architecture diagram, which includes the server, the terminal device of the buyer and the terminal device of the merchant. The server can be the background server of the financial application. The buyer scans the code or triggers the specified financial service function (such as payment or transfer, etc.) in the application provided by the merchant. The terminal device calls up the corresponding interface of the financial application to complete the processing of the financial service to obtain physical or virtual goods or services. In the above process, the buyer may be dissatisfied with or disapproval of the goods or services obtained. Therefore, the financial application will open up channels such as the "reporting center" or complaint entrance to allow the buyer to provide feedback to the merchant, transaction channels, detailed events, report or complain about unapproved transactions and submit proof related to the transaction, and wait for the result of the complaint. Merchants can also enter the merchant center to query and handle relevant complaints. However, the current process of merchants handling complaints is relatively passive from the merchant's perspective. Merchants can handle complaints in three ways, namely, proactive refund; negotiate with the buyer and proactively contact the buyer to withdraw the complaint; do not handle the complaint separately, but rectify the business or model that is prone to complaints (i.e. business rectification). From the feedback of merchants, merchants are not willing to deal with some disputes passively; at the same time, merchants also tend to have a channel for complaints, provide evidence to feedback the merchant's understanding of the complaint, and apply for requalification of the complaint. Therefore, in the complaint reporting scenario, merchants hope to have an opportunity to actively defend themselves (another way of handling). Since the complaint scenario on the merchant side is relatively special, the handling solution of active defense will also exist in other products or businesses, for example, the consumer court: the complaining parties provide evidence to each other, and the platform and other buyers on the platform act as third-party evaluators to vote based on the evidence of both parties to determine whether the complaint is established and whether the merchant should return the resources transferred by the buyer or other corresponding treatment. The above method relies on two key points, namely the reputation (i.e. credibility) established by the platform's proper handling; other buyers on the platform have the motivation to participate in the consumer court. However, the continuous operation of the consumer court will incur high costs (the platform's labor costs; the need to operate a specific review team for manual qualitative judgment, so there are operating costs; and the plan for other buyers to participate needs to provide incentives to the buyer in the form of points, coupons, and rights, which also has operating costs). With the continuous development of big models, the current conversational capabilities of big models can already perform functions such as interpreting text and images, and context series comparison. Therefore, the embodiments of this specification provide a feasible method, that is, the text and images provided by the purchaser when making a complaint can allow the big model to interpret events and determine risks; when the merchant requires evidence to be provided for the complaint, it can also recommend information that the merchant should provide as evidence (including text descriptions, purchaser contact information, and auxiliary evidence images, etc.), and evaluate whether the submitted evidence information is consistent with the feedback from the purchaser above, and again conduct a qualitative analysis of the reported complaint to evaluate whether the complaint is established and whether the merchant needs to handle it.The merchant defense qualitative solution based on the above large model provides a channel for merchants to provide evidence, and no longer passively handles complaints; at the same time, compared with other consumer courts, it also reduces the human cost of platform manual review and the operating cost of incentivizing platform buyers in the form of points, coupons, and rights. For specific processing, please refer to the specific content in the following embodiment.

[0022] like Figure 2 As shown, the embodiments of this specification provide a qualitative analysis method for transaction risks. The execution subject of the method may be a terminal device or a server, etc. The terminal device may be a mobile terminal device such as a mobile phone, a tablet computer, or a computer device such as a laptop or a desktop computer, or an IoT device (specifically a smart watch, a car device, etc.), etc. The server may be an independent server, or a server cluster composed of multiple servers, etc. The server may be a background server for a financial service or an online shopping service, or a background server for an application, etc. In this embodiment, the execution subject is taken as an example for detailed description. For the case where the execution subject is a terminal device, please refer to the following server situation processing, which will not be repeated here. The method may specifically include the following steps:

[0023] In step S202, a transaction complaint request for a target transaction submitted by a user is received, where the transaction complaint request includes a complaint risk type of the target transaction and first evidence information for the complaint risk type.

[0024] Among them, the target transaction can be any transaction, for example, payment, transfer, etc. in financial services, or it can also be insurance transactions or other forms of equivalent exchange, etc., which can be set according to actual conditions, and the embodiments of this specification do not limit this. The user can be any user who conducts the target transaction. The target transaction can usually include both parties to the transaction, one of which can be the buyer and the other can be the merchant. In the transaction complaint, the two parties to the transaction can include the complaining party and the complained party. For the convenience of expression, the complaining party can be the user, and the complaining party can be the above-mentioned buyer or the above-mentioned merchant. The complaint risk type can be the type of risk that may exist in the transaction being complained. The complaint risk type can include multiple types, for example, the complaint risk type can include fraud risk, illegal financial activity risk, privacy leakage risk, misappropriation risk, etc., which can be set according to actual conditions, and the embodiments of this specification do not limit this. The first evidence information can be the voucher information used to support the risk corresponding to the complaint risk type in the target transaction. The first evidence information can include text information, images, etc., which can be set according to actual conditions.

[0025] In practice, the buyer can select the goods to be purchased in the merchant's store (such as physical goods, virtual goods or services, etc.). After selecting the goods, the buyer scans the code or triggers the specified financial service (such as payment or transfer, etc.) in the application provided by the merchant. The buyer's terminal device calls up the corresponding interface of the financial application and completes the processing of the financial service to obtain the goods. The merchant can distribute the goods to the buyer based on the goods purchased by the buyer (the goods can be distributed through designated online channels or by mail, etc.). During the above-mentioned transaction (i.e., the target transaction), if the purchaser or the merchant finds that there may be a certain risk, a complaint may be filed on the transaction platform (i.e., the server) of the above-mentioned transaction. A complaint mechanism may be set up in the transaction platform. Specifically, a "complaint center" (or "reporting center") or a complaint entrance (such as a complaint link or a complaint button, etc.) may be set up in the transaction platform. The complainant may select the transaction to be complained in the "complaint center" (i.e., the target transaction, such as transaction A with order number 123XXXX or transaction B with waybill number K822XXXX, etc.), and at the same time, the content to be complained and the complaint risk type of the target transaction may be entered, and text descriptions, images and other information provided by the user (i.e., the complainant) may be uploaded, and the text descriptions, images and other information may be used as the first evidence information for the complaint risk type. The text descriptions may include multiple types, for example, "After purchasing the goods, the goods have not been shipped for a long time after waiting, suspected fraud", etc. The images may be screenshots of the interactive information of the instant messaging between the two parties of the target transaction, screenshots of relevant pages in the target transaction process (such as the transfer page, the product delivery status page, etc.), etc., which may be set according to actual conditions. After completing the input of the above information, you can click the "Complaint" button. At this time, the terminal device can obtain the above information selected or entered by the user, and can generate a transaction complaint request for the target transaction based on the obtained information, and can submit the transaction complaint request to the server, so that the server can obtain the transaction complaint request submitted by the user for the target transaction.

[0026] In step S204, the above-mentioned complaint risk type and the first evidence information are input into the pre-trained large model to obtain a preliminary qualitative analysis result on whether the target transaction has the risk corresponding to the complaint risk type. The large language model is used to perform a qualitative analysis on whether the transaction has the preset risk.

[0027] Among them, the large model can be composed of hundreds of millions of parameters, based on deep learning algorithms, and pre-trained with large-scale corpus to perform semantic understanding, generation and conversion of data. The large model can include discriminative large models and generative large models. The large model can specifically include large language models (LLM), large models built with BERT, or large models built based on the Transformer architecture, etc. The large language models can include GPT-3, ERNIE Bot, etc., which can be set according to actual conditions. The preset risk can be a pre-specified risk, such as the risk corresponding to the complaint risk type submitted by the user, or it can be a pre-set risk, such as fraud risk, illegal financial activity risk, etc.

[0028] In implementation, a big model that can be used to perform a qualitative analysis on whether a transaction has a preset risk can be directly obtained from a specified database, or a specified algorithm can be obtained, and the model architecture of the big model can be constructed by the specified algorithm. Then, a training sample can be obtained, and the training sample and a preset loss function can be used to train the big model until the loss function converges to obtain a trained big model. Alternatively, a pre-trained big model can be obtained, and then sample data for fine-tuning the pre-trained big model can be obtained, and the sample data can be used to fine-tune the pre-trained big model to obtain a fine-tuned big model, etc. The specific setting can be based on actual conditions, and the embodiments of this specification are not limited to this.

[0029] After obtaining the big model in the above manner, corresponding prompt information (i.e., Prompt, used to illustrate and explain the data input into the big model to prompt the big model to output the correct result) can be constructed based on the above complaint risk type and the first evidence information. The prompt information, the complaint risk type and the first evidence information can be input into the big model. Through the big model, the first evidence information submitted by the user can be parsed. Specifically, the cause of the event and the process of the event contained in the text information can be extracted, and the specified content can be further summarized. For images, the text information (such as text interaction records, transfer records, etc.), the semantic information of the image, the type of the image can be extracted, and it can be analyzed whether the image is a tampered image, etc. Finally, a preliminary qualitative analysis result can be generated as to whether the target transaction has the risk corresponding to the complaint risk type, and the preliminary qualitative analysis result can be output.

[0030] In step S206, the preliminary qualitative analysis result is sent to the party complained against in the target transaction, and feedback information from the party complained against regarding the preliminary qualitative analysis result is received.

[0031] In practice, in order to enable the party against whom the complaint is made to understand the situation of the complaint in a timely manner and to handle the user's complaint, the server may send the above-obtained preliminary qualitative analysis results to the party against whom the complaint is made. In addition, in addition to sending the preliminary qualitative analysis results to the party against whom the complaint is made, the server may also send the relevant information of the target transaction (such as the target transaction's identification, transaction details), the complaint risk type, and one or more of the first evidence information to the party against whom the complaint is made. The terminal device of the party against whom the complaint is made may display the received information, and the party against whom the complaint is made may understand the situation of the complaint through the displayed information.

[0032] If the preliminary qualitative analysis results indicate that the target transaction has a risk corresponding to the complaint risk type, and the party complained against has an objection to the above-mentioned user complaint, an appeal can be filed based on the preliminary qualitative analysis results; if the party complained against has no objection to the above-mentioned user complaint, it can be handled in a variety of ways. For example, the merchant can return the transferred resources to the purchaser (such as a refund, etc.), or the party complained against can actively negotiate with the user and request the user to withdraw the complaint after reaching a consensus through negotiation. Alternatively, the party complained against can also make rectifications to the designated business, etc. The specific settings can be based on actual conditions, and the embodiments of this specification do not limit this. Through the above method, the party complained against can provide corresponding feedback on the preliminary qualitative analysis results. Specifically, a complaint handling page can be sent to the party complained against, and the complaint handling page can include a user complaint handling portal (which can be a website or hyperlink, etc.), and the corresponding page can be obtained through the user complaint handling portal, and the corresponding feedback can be provided through the processing method provided in the page. For example, if the party complained against has objections to the above user complaint, the party complained against can appeal against the preliminary qualitative analysis results, and the appeal of the party complained against can be the above feedback. For another example, if the party complained against has no objection to the above user complaint and needs a refund, the feedback of the party complained against can be an active refund, etc., which can be set according to the actual situation. The terminal device of the party complained against can generate corresponding feedback information according to the feedback of the party complained against, and can send the feedback information to the server, and the server can receive the feedback information of the party complained against the preliminary qualitative analysis results.

[0033] If the preliminary qualitative analysis results indicate that the target transaction does not have the risk corresponding to the complaint risk type, the corresponding feedback information may include confirmation or no objection, etc., which can be set according to the actual situation.

[0034] In step S208, if the above feedback information includes a complaint request from the party complained against, the second evidence information provided by the party complained against for the complaint request is obtained, and based on the first evidence information and the second evidence information, a secondary qualitative analysis is performed on whether the target transaction has a risk corresponding to the complaint risk type through the big model to obtain a secondary qualitative analysis result.

[0035] Among them, the appeal request can be the act of the appellant stating the reasons to the corresponding institution, equipment or procedure and requesting reprocessing when he believes that the handling result of a certain event (or problem) is incorrect. The second evidence information can be the voucher information used to support the incorrectness of the preliminary qualitative analysis result. The second evidence information can include text information, images, etc., which can be set according to the actual situation.

[0036] In implementation, the above feedback information can be parsed. If it is determined based on the parsing result that the feedback information includes the complaint request of the party being complained against, it indicates that the preliminary qualitative analysis result obtained above is likely to be that the target transaction has a risk corresponding to the complaint risk type. At this time, a credential submission entry can be provided for the party being complained against. For example, the server can send a URL for submitting credentials to the party being complained against, or the server can directly send the data of the credential submission page to the party being complained against. The party being complained against can use the above URL or the data of the credential submission page through its terminal device to obtain and construct a corresponding credential submission page. The party being complained against can enter the corresponding credential data (which may include text information, images, etc.) in the credential submission page for submission. The terminal device of the party being complained against can generate second evidence information based on the credential data submitted by the party being complained against, and the server can obtain the second evidence information provided by the party being complained against for the complaint request.

[0037] Corresponding prompt information can be constructed based on the first evidence information and the second evidence information, as well as other relevant data that may be required. The prompt information, the first evidence information and the second evidence information can be input into the big model. Through the big model, the second evidence information submitted by the party being complained can be parsed. Specifically, the cause of the event and the process of the event contained in the text information can be extracted, and the specified content can be further summarized. For images, the text information (such as text interaction records, transfer records, etc.), the semantic information of the image, the type of the image can be extracted, and it can be analyzed whether the image is a tampered image, etc. In addition, the conflict between the text or an element in the first evidence information and the second evidence information can be analyzed, and it can be evaluated whether the second evidence information submitted by the party being complained can prove its argument, and whether there is a conflict between the second evidence information submitted by the party being complained and the first evidence information submitted by the user, etc., so as to re-qualitatively analyze whether the target transaction has the risk corresponding to the complaint risk type, and finally generate a secondary qualitative analysis result on whether the target transaction has the risk corresponding to the complaint risk type, and the secondary qualitative analysis result can be output.

[0038] The embodiments of this specification provide a qualitative analysis method for transaction risk, which receives a transaction complaint request submitted by a user for a target transaction, wherein the transaction complaint request includes the complaint risk type of the target transaction and first evidence information for the complaint risk type. Then, the complaint risk type and the first evidence information can be input into a pre-trained large model to obtain a preliminary qualitative analysis result on whether the target transaction has a risk corresponding to the complaint risk type. The preliminary qualitative analysis result is sent to the complained party in the target transaction, and feedback information from the complained party on the preliminary qualitative analysis result is received. If the feedback information includes a complaint request from the complained party, second evidence information provided by the complained party for the complaint request is obtained. Based on the first evidence information and the second evidence information, a secondary qualitative analysis is performed on whether the target transaction has a risk corresponding to the complaint risk type through the large model. In this way, due to the entire qualitative analysis The entire process relies on the big model, which avoids the time-consuming and subjective qualitative results of manual review and other processing. In addition, since the evidence information submitted by the reporting parties can be analyzed in detail, the information errors and omissions caused by manual review and other processing can be reduced. Moreover, the big model is responsible for the qualitative analysis of complaints and the analysis of the defense process of the complaining parties. The training and operating costs for manual work will also decrease accordingly, including reducing the training and organizational costs of auditors, and reducing the incentive costs and operating costs for users on the platform to participate in complaint courts and other methods. In addition, since the qualitative analysis of complaints is no longer limited to the perspective of the buyer, the big model can interpret text and pictures, and compare context in series. Therefore, the two parties to the transaction are allowed to provide evidence to each other regarding the complaint, making the qualitative analysis of the complaint more comprehensive and accurate. From the perspective of the merchants, the platform provides an entry so that the party being complained is no longer just passively accepting complaints, reducing operating costs and improving experience.

[0039] In practical applications, in addition to sending the preliminary qualitative analysis results to the party being complained, the preliminary qualitative analysis results can also be displayed to the user, and an information supplement portal can be opened so that the user can submit more data for more accurate event analysis and qualitative analysis, such as Figure 3 As shown, the process may specifically include the following steps S210.

[0040] In step S210, the preliminary qualitative analysis result is sent to the user, and the first supplementary information for the target transaction submitted by the user is received.

[0041] During implementation, the preliminary qualitative analysis results can be sent to the user. At the same time, relevant information of the information supplement entrance (such as the URL or hyperlink of the information supplement entrance, etc.) can be provided to the user to open the information supplement entrance to the user. The user can supplement more data (i.e., the first supplementary information) through the information supplement entrance. The server can receive the first supplementary information submitted by the user for the target transaction, and the server can perform more accurate event analysis and qualitative analysis on the target transaction through the first supplementary information.

[0042] Based on the processing of step S210, Figure 3 As shown, in the above step S208, based on the first evidence information and the second evidence information, a secondary qualitative analysis is performed on whether the target transaction has a risk corresponding to the complaint risk type through the big model, and a specific processing method for obtaining the secondary qualitative analysis result may include: inputting the first supplementary information, the first evidence information and the second evidence information into the big model to obtain a secondary qualitative analysis result on whether the target transaction has a risk corresponding to the complaint risk type.

[0043] In implementation, corresponding prompt information can be constructed based on the first supplementary information, the first evidence information, the second evidence information, and other relevant data that may be needed. The prompt information, the first supplementary information, the first evidence information, and the second evidence information can be input into the big model. Through the big model, the first supplementary information, the first evidence information, and the second evidence information can be parsed respectively. Specifically, the cause of the event and the process of the event contained in the text information can be extracted, and the specified content can be further summarized. For images, the text information (such as text interaction records, transfer records, etc.), the semantic information of the image, and the image information can be extracted. The type to which the image belongs can be determined, and whether the image is a tampered image, etc. In addition, the conflict between the first supplementary information and the text or an element in the first evidence information and the second evidence information can be analyzed to evaluate whether the second evidence information submitted by the party complained against can prove its own argument, and whether there is a conflict between the second evidence information submitted by the party complained against and the first evidence information and the first supplementary information submitted by the user, etc., so as to re-qualitatively analyze whether the target transaction has the risk corresponding to the complaint risk type, and finally generate a secondary qualitative analysis result on whether the target transaction has the risk corresponding to the complaint risk type, and the secondary qualitative analysis result can be output.

[0044] In practical applications, the secondary qualitative analysis results can be sent to the user and the party being complained. At the same time, the information supplement entrance can be opened to the user and the party being complained, so that the user and the party being complained can submit more data for more accurate event analysis and qualitative analysis, such as Figure 4 As shown, please refer to the processing of the following steps S212 to S218 for details.

[0045] In step S212, the secondary qualitative analysis result is sent to the user and the party against whom the complaint is made.

[0046] During implementation, the secondary qualitative analysis results can be sent to the user and the party against whom the complaint is made. At the same time, the relevant information of the information supplement entrance (such as the URL or hyperlink of the information supplement entrance, etc.) can be provided to the user and the party against whom the complaint is made, respectively, so as to open the information supplement entrance to the user and the party against whom the complaint is made. The user and / or the party against whom the complaint is made can supplement more data (i.e., the second supplementary information) through the information supplement entrance, so as to conduct a more accurate event analysis and qualitative characterization of the target transaction through the second supplementary information.

[0047] In step S214, second supplementary information for the target transaction is received, where the second supplementary information includes supplementary information submitted by the user and / or supplementary information submitted by the party against whom the complaint is made.

[0048] In step S216, based on the second supplementary information, the first evidence information and the second evidence information, a qualitative analysis is performed again on whether the target transaction has a risk corresponding to the complaint risk type through the big model to obtain another qualitative analysis result.

[0049] In implementation, corresponding prompt information can be constructed based on the second supplementary information, the first evidence information and the second evidence information, as well as other relevant data that may be required. The prompt information, the second supplementary information, the first evidence information and the second evidence information can be input into the big model. Through the big model, the second supplementary information, the first evidence information and the second evidence information can be parsed respectively. Specifically, the cause of the event and the process of the event contained in the text information can be extracted, and the specified content can be further summarized. For images, the text information (such as text interaction records, transfer records, etc.), image information and other related information can be extracted. The semantic information of the image, the type of the image, and whether the image is a tampered image, etc., can also be analyzed. In addition, the second supplementary information and the conflict between the text or an element in the first and second evidence information can be analyzed, as well as whether there is a conflict between the information submitted by the complained party and the information submitted by the user, etc., so as to conduct another qualitative analysis (i.e., the third qualitative analysis) on whether the target transaction has the risk corresponding to the complaint risk type. Finally, another qualitative analysis result (i.e., the third qualitative analysis result) on whether the target transaction has the risk corresponding to the complaint risk type can be generated, and the another qualitative analysis result can be output.

[0050] In step S218, another qualitative analysis result is sent to the user and the party complained against.

[0051] It should be noted that the above is to send the secondary qualitative analysis results to the user and the party complained against. In actual applications, the sending target of the secondary qualitative analysis results can also be determined according to the content of the secondary qualitative analysis results. For example, if the secondary qualitative analysis results indicate that the target transaction has a risk corresponding to the complaint risk type, the secondary qualitative analysis results can be sent to the party complained against, and the second supplementary information submitted by the party complained against for the target transaction can be received; based on the second supplementary information, the first evidence information and the second evidence information, the large model is used to perform another qualitative analysis on whether the target transaction has a risk corresponding to the complaint risk type, and another qualitative analysis result is obtained; if the secondary qualitative analysis results indicate that the target transaction does not have a risk corresponding to the complaint risk type, the secondary qualitative analysis results can be sent to the user, and the second supplementary information submitted by the user for the target transaction can be received; based on the second supplementary information, the first evidence information and the second evidence information, the large model is used to perform another qualitative analysis on whether the target transaction has a risk corresponding to the complaint risk type, and another qualitative analysis result is obtained. The subsequent re-qualitative analysis results can also be selectively sent according to the above method, which will not be repeated here.

[0052] In practical applications, the feedback information of the complained party may be a resource return request in addition to an appeal request, such as a refund request, etc. For details, please refer to the processing of the following steps A2 and A4.

[0053] In step A2, if the feedback information includes a resource return request, a first amount of resources transferred from the user to the complained party in the target transaction is obtained, and the resource return request is used to request the return of the resources transferred from the user to the complained party in the target transaction.

[0054] In implementation, if it is necessary to return the transferred resources to the purchaser (such as a refund, etc.), the first quantity of resources transferred by the user to the complained party in the target transaction can be obtained, and a resource return request can be generated based on this. The resource return request can be submitted to the server as feedback information. After the server analyzes the feedback information, if the above feedback information includes a resource return request, the first quantity of resources transferred by the user to the complained party in the target transaction can be extracted from the resource return request.

[0055] In step A4, the acquired first amount of resources is transferred to the user.

[0056] In actual applications, the feedback information from the party complained against may be a request for voluntary withdrawal of the complaint in addition to a request for appeal. For details, please refer to the processing of the following steps B2 and B4.

[0057] In step B2, if the feedback information includes an active withdrawal request, a withdrawal request for the transaction complaint request submitted by the user is received within a first preset time period.

[0058] The first preset time period may be set according to actual conditions. For example, the first preset time period may be 7 days or one month from the date of receipt of the request for active withdrawal of the lawsuit.

[0059] In implementation, if it is necessary to negotiate with the purchaser to resolve the corresponding problem, an active withdrawal request can be generated, and the active withdrawal request can be submitted to the server as feedback information. Then, the party complained against can actively contact the user to negotiate the solution to the corresponding problem. After reaching a consensus through negotiation, the party complained against can request the user to withdraw the complaint. After analyzing the feedback information, if the above feedback information includes an active withdrawal request, the server monitors whether a withdrawal request for the transaction complaint request submitted by the user is received within the first preset time period.

[0060] In step B4, if a cancellation request submitted by the user is received within the first preset time period, the response to the transaction complaint request is cancelled.

[0061] In implementation, if a cancellation request submitted by a user is received within the first preset time period, it indicates that the user's problem or question has been resolved. At this time, the transaction complaint request previously submitted by the user is invalid, and the server can cancel the response to the transaction complaint request, that is, the server can delete the transaction complaint request, and there is no need to perform other processing on the transaction complaint request.

[0062] In actual applications, the feedback information from the party complained against may be a request for business rectification in addition to a request for appeal. For details, please refer to the processing of the following steps C2 and C4.

[0063] In step C2, if the feedback information includes a business rectification request, the transaction authority of the party complained against is restricted, and rectification result data submitted by the party complained against is received within a second preset time period.

[0064] The second preset time period may be set according to actual conditions, for example, the second preset time period may be one month or three months after receiving the business rectification request. The rectification result data may include business service information content before and after the rectification, business service effects before and after the rectification, etc., which may be set according to actual conditions, and the embodiments of this specification do not limit this.

[0065] In implementation, if the corresponding business service needs to be rectified, a business rectification request can be generated, and the business rectification request can be submitted to the server as feedback information. After the server analyzes the feedback information, if the feedback information includes the business rectification request, it monitors whether the rectification result data submitted by the complained party is received within the second preset time period.

[0066] In step C4, if the rectification result data meets the preset business rectification rules, the restriction on the transaction authority of the party complained against is cancelled.

[0067] Among them, the business rectification rules may include multiple types. For example, the business rectification rules may be such as using test data to execute a specified transaction corresponding to the rectification result data, and then, a qualitative analysis may be performed through a large model. If the qualitative analysis result indicates that the specified transaction does not have a preset risk, then it is determined that the rectification result data satisfies the preset business rectification rules. Otherwise, it is determined that the rectification result data does not satisfy the preset business rectification rules. Alternatively, the business rectification rules may be such as the content of the rectification result data satisfies a preset completeness threshold (such as 95% or 98%, etc.), then it is determined that the rectification result data satisfies the preset business rectification rules. Otherwise, it is determined that the rectification result data does not satisfy the preset business rectification rules.

[0068] In implementation, if the rectification result data submitted by the party complained against is received within the second preset time period and the rectification result data meets the preset business rectification rules, the restrictions on the transaction rights of the party complained against can be lifted so that the party complained against can resume the corresponding business services.

[0069] In practical applications, such as Figure 5 As shown, in the above step S208, if the above feedback information includes the appeal request of the party complained against, the following method may be further used for processing before obtaining the second evidence information provided by the party complained against for the appeal request.

[0070] In step S220, the first evidence information is input into a pre-trained large model to determine analysis information corresponding to the first evidence information, wherein the analysis information includes event summary information of an event corresponding to a target transaction and one or more of the evidence element information in the first evidence information.

[0071] Among them, the evidence element information in the first evidence information may include multiple ones. For example, the evidence element information in the first evidence information may include the delivery address, order number, sender information, etc., which can be set according to actual conditions, and the embodiments of this specification do not limit this.

[0072] Based on the processing of step S220, if the feedback information in step S208 includes the appeal request of the party being complained, there are many specific processing methods for obtaining the second evidence information provided by the party being complained for the appeal request. An optional processing method is provided below, such as Figure 5 As shown, the processing may specifically include the following steps S2082 and S2084.

[0073] In step S2082, if the feedback information includes a complaint request from the party being complained against, the feedback information and the analysis information are input into the big model to obtain the credential information that the party being complained against needs to submit to appeal against the preliminary qualitative analysis results.

[0074] In implementation, for example, the information of a user complaint includes: the user purchased goods at the store of the party complained against and waited for a long time without the goods being shipped, which is suspected to be a fraud. The complaint information provided by the user can be parsed and processed through the big model to obtain corresponding analysis information, and the feedback information and the analysis information are input into the big model to obtain the credential information that the party complained against needs to submit to appeal the preliminary qualitative analysis results. Based on the above example, if the party complained against needs to appeal, the credential information that the party complained against needs to submit to appeal the preliminary qualitative analysis results may include: the party complained against needs to describe the actual delivery situation and submit transaction communication records, user delivery address communication situation, express delivery number, sender's mobile phone number and other credential information.

[0075] In step S2084, second evidence information obtained based on the above-mentioned credential information and provided by the party complained against is received.

[0076] In actual applications, the type of information contained in the first evidence information includes one or more of text, image, audio, and video; the type of information contained in the second evidence information includes one or more of text, image, audio, and video.

[0077] The embodiments of this specification provide a qualitative analysis method for transaction risk, which receives a transaction complaint request submitted by a user for a target transaction, wherein the transaction complaint request includes the complaint risk type of the target transaction and first evidence information for the complaint risk type. Then, the complaint risk type and the first evidence information can be input into a pre-trained large model to obtain a preliminary qualitative analysis result on whether the target transaction has a risk corresponding to the complaint risk type. The preliminary qualitative analysis result is sent to the complained party in the target transaction, and feedback information from the complained party on the preliminary qualitative analysis result is received. If the feedback information includes a complaint request from the complained party, second evidence information provided by the complained party for the complaint request is obtained. Based on the first evidence information and the second evidence information, a secondary qualitative analysis is performed on whether the target transaction has a risk corresponding to the complaint risk type through the large model. In this way, due to the entire qualitative analysis The entire process relies on the big model, which avoids the time-consuming and subjective qualitative results of manual review and other processing. In addition, since the evidence information submitted by the reporting parties can be analyzed in detail, the information errors and omissions caused by manual review and other processing can be reduced. Moreover, the big model is responsible for the qualitative analysis of complaints and the analysis of the defense process of the complaining parties. The training and operating costs for manual work will also decrease accordingly, including reducing the training and organizational costs of auditors, and reducing the incentive costs and operating costs for users on the platform to participate in complaint courts and other methods. In addition, since the qualitative analysis of complaints is no longer limited to the perspective of the buyer, the big model can interpret text and pictures, and compare context in series. Therefore, the two parties to the transaction are allowed to provide evidence to each other regarding the complaint, making the qualitative analysis of the complaint more comprehensive and accurate. From the perspective of the merchants, the platform provides an entry so that the party being complained is no longer just passively accepting complaints, reducing operating costs and improving experience.

[0078] The above is a qualitative analysis method for transaction risk provided in the embodiment of this specification. Based on the same idea, the embodiment of this specification also provides a qualitative analysis device for transaction risk, such as Figure 6 shown.

[0079] The qualitative analysis device for transaction risk includes: a request module 601, a preliminary qualitative analysis module 602, a feedback receiving module 603 and a secondary qualitative analysis module 604, wherein:

[0080] The request module 601 receives a transaction complaint request submitted by a user for a target transaction, wherein the transaction complaint request includes a complaint risk type of the target transaction and first evidence information for the complaint risk type;

[0081] A preliminary qualitative analysis module 602 inputs the complaint risk type and the first evidence information into a pre-trained large model to obtain a preliminary qualitative analysis result of whether the target transaction has a risk corresponding to the complaint risk type. The large language model is used to perform a qualitative analysis on whether the transaction has a preset risk.

[0082] A feedback receiving module 603 sends the preliminary qualitative analysis result to the complained party in the target transaction, and receives feedback information from the complained party regarding the preliminary qualitative analysis result;

[0083] The secondary qualitative analysis module 604 obtains second evidence information provided by the party complained against for the appeal request if the feedback information includes the appeal request of the party complained against, and based on the first evidence information and the second evidence information, performs a secondary qualitative analysis on whether the target transaction has the risk corresponding to the complaint risk type through the big model to obtain a secondary qualitative analysis result.

[0084] In the embodiment of this specification, the device further includes:

[0085] A supplementary information acquisition module, sending the preliminary qualitative analysis result to the user, and receiving first supplementary information for the target transaction submitted by the user;

[0086] The secondary qualitative analysis module 604 inputs the first supplementary information, the first evidence information and the second evidence information into the large model to obtain a secondary qualitative analysis result on whether the target transaction has a risk corresponding to the complaint risk type.

[0087] In the embodiment of this specification, the device further includes:

[0088] A first result sending module, sending the secondary qualitative analysis result to the user and the party complained against;

[0089] A supplementary module receives second supplementary information for the target transaction, wherein the second supplementary information includes supplementary information submitted by the user and / or supplementary information submitted by the party against whom the complaint is filed;

[0090] A qualitative module, based on the second supplementary information, the first evidence information and the second evidence information, performs another qualitative analysis on whether the target transaction has a risk corresponding to the complaint risk type through the big model to obtain another qualitative analysis result;

[0091] The second result sending module sends the qualitative analysis result again to the user and the party complained against.

[0092] In the embodiment of this specification, the device further includes:

[0093] a refund request module, which obtains a first amount of resources transferred by the user to the complained party in the target transaction if the feedback information includes a resource refund request, wherein the resource refund request is used to request the refund of the resources transferred by the user to the complained party in the target transaction;

[0094] The resource transfer module transfers the acquired first amount of resources to the user.

[0095] In the embodiment of this specification, the device further includes:

[0096] A withdrawal request module, which receives a withdrawal request for the transaction complaint request submitted by the user within a first preset time period if the feedback information includes an active withdrawal request;

[0097] The first cancellation module cancels the response to the transaction complaint request if the cancellation request submitted by the user is received within a first preset time period.

[0098] In the embodiment of this specification, the device further includes:

[0099] a rectification request module, which, if the feedback information includes a business rectification request, restricts the transaction authority of the party complained against, and receives rectification result data submitted by the party complained against within a second preset time period;

[0100] The second cancellation module cancels the restriction on the transaction authority of the complained party if the rectification result data meets the preset business rectification rules.

[0101] In the embodiment of this specification, the device further includes:

[0102] a parsing module, inputting the first evidence information into a pre-trained large model, and determining parsing information corresponding to the first evidence information, wherein the parsing information includes one or more of event summary information of the event corresponding to the target transaction and evidence element information in the first evidence information;

[0103] The secondary qualitative analysis module 804, if the feedback information includes a request for appeal from the party complained against, inputs the feedback information and the analysis information into the large model to obtain the credential information required for the party complained against to appeal against the preliminary qualitative analysis result; and receives the second evidence information provided by the party complained against based on the credential information.

[0104] In an embodiment of the present specification, the type of information contained in the first evidence information includes one or more of text, image, audio, and video; the type of information contained in the second evidence information includes one or more of text, image, audio, and video.

[0105] The embodiment of the present specification provides a qualitative analysis device for transaction risk, which receives a transaction complaint request submitted by a user for a target transaction, wherein the transaction complaint request includes the complaint risk type of the target transaction and first evidence information for the complaint risk type. Then, the complaint risk type and the first evidence information can be input into a pre-trained large model to obtain a preliminary qualitative analysis result on whether the target transaction has a risk corresponding to the complaint risk type. The preliminary qualitative analysis result is sent to the complained party in the target transaction, and feedback information from the complained party on the preliminary qualitative analysis result is received. If the feedback information includes a complaint request from the complained party, second evidence information provided by the complained party for the complaint request is obtained. Based on the first evidence information and the second evidence information, a secondary qualitative analysis is performed on whether the target transaction has a risk corresponding to the complaint risk type through the large model. In this way, due to the entire qualitative analysis The entire process relies on the big model, which avoids the time-consuming and subjective qualitative results of manual review and other processing. In addition, since the evidence information submitted by the reporting parties can be analyzed in detail, the information errors and omissions caused by manual review and other processing can be reduced. Moreover, the big model is responsible for the qualitative analysis of complaints and the analysis of the defense process of the complaining parties. The training and operating costs for manual work will also decrease accordingly, including reducing the training and organizational costs of auditors, and reducing the incentive costs and operating costs for users on the platform to participate in complaint courts and other methods. In addition, since the qualitative analysis of complaints is no longer limited to the perspective of the buyer, the big model can interpret text and pictures, and compare context in series. Therefore, the two parties to the transaction are allowed to provide evidence to each other regarding the complaint, making the qualitative analysis of the complaint more comprehensive and accurate. From the perspective of the merchants, the platform provides an entry so that the party being complained is no longer just passively accepting complaints, reducing operating costs and improving experience.

[0106] The above is a qualitative analysis device for transaction risk provided in the embodiment of this specification. Based on the same idea, the embodiment of this specification also provides a qualitative analysis device for transaction risk, such as Figure 7 shown.

[0107] The qualitative analysis device for transaction risk may provide a terminal device or a server, etc. for the above-mentioned embodiment.

[0108] The qualitative analysis device of transaction risk may have relatively large differences due to different configurations or performances, and may include one or more processors 701 and memory 702, and the memory 702 may store one or more storage applications or data. Among them, the memory 702 may be a temporary storage or a permanent storage. The application stored in the memory 702 may include one or more modules (not shown in the figure), and each module may include a series of computer executable instructions in the qualitative analysis device of transaction risk. Furthermore, the processor 701 may be configured to communicate with the memory 702, and execute a series of computer executable instructions in the memory 702 on the qualitative analysis device of transaction risk. The qualitative analysis device of transaction risk may also include one or more power supplies 703, one or more wired or wireless network interfaces 704, one or more input and output interfaces 705, and one or more keyboards 706.

[0109] Specifically, in this embodiment, the qualitative analysis device for transaction risk includes a memory and one or more programs, wherein the one or more programs are stored in the memory, and the one or more programs may include one or more modules, and each module may include a series of computer executable instructions in the qualitative analysis device for transaction risk, and the one or more programs are configured to be executed by one or more processors, including computer executable instructions for performing the following:

[0110] receiving a transaction complaint request submitted by a user for a target transaction, wherein the transaction complaint request includes a complaint risk type of the target transaction and first evidence information for the complaint risk type;

[0111] Inputting the complaint risk type and the first evidence information into a pre-trained large model to obtain a preliminary qualitative analysis result on whether the target transaction has a risk corresponding to the complaint risk type, and the large language model is used to perform a qualitative analysis on whether the transaction has a preset risk;

[0112] Sending the preliminary qualitative analysis result to the party complained against in the target transaction, and receiving feedback information from the party complained against regarding the preliminary qualitative analysis result;

[0113] If the feedback information includes a complaint request from the party complained against, then obtain second evidence information provided by the party complained against for the complaint request; based on the first evidence information and the second evidence information, perform a secondary qualitative analysis on whether the target transaction has a risk corresponding to the complaint risk type through the big model to obtain a secondary qualitative analysis result.

[0114] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the qualitative analysis device embodiment of transaction risk, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0115] The embodiment of the present specification provides a qualitative analysis device for transaction risk, which receives a transaction complaint request submitted by a user for a target transaction, wherein the transaction complaint request includes the complaint risk type of the target transaction and first evidence information for the complaint risk type. Then, the complaint risk type and the first evidence information can be input into a pre-trained large model to obtain a preliminary qualitative analysis result on whether the target transaction has a risk corresponding to the complaint risk type. The preliminary qualitative analysis result is sent to the complained party in the target transaction, and feedback information from the complained party on the preliminary qualitative analysis result is received. If the feedback information includes a complaint request from the complained party, second evidence information provided by the complained party for the complaint request is obtained. Based on the first evidence information and the second evidence information, a secondary qualitative analysis is performed on whether the target transaction has a risk corresponding to the complaint risk type through the large model. In this way, due to the entire qualitative analysis The entire process relies on the big model, which avoids the time-consuming and subjective qualitative results of manual review and other processing. In addition, since the evidence information submitted by the reporting parties can be analyzed in detail, the information errors and omissions caused by manual review and other processing can be reduced. Moreover, the big model is responsible for the qualitative analysis of complaints and the analysis of the defense process of the complaining parties. The training and operating costs for manual work will also decrease accordingly, including reducing the training and organizational costs of auditors, and reducing the incentive costs and operating costs for users on the platform to participate in complaint courts and other methods. In addition, since the qualitative analysis of complaints is no longer limited to the perspective of the buyer, the big model can interpret text and pictures, and compare context in series. Therefore, the two parties to the transaction are allowed to provide evidence to each other regarding the complaint, making the qualitative analysis of the complaint more comprehensive and accurate. From the perspective of the merchants, the platform provides an entry so that the party being complained is no longer just passively accepting complaints, reducing operating costs and improving experience.

[0116] Furthermore, based on the above Figures 2 to 5 In one embodiment, the present specification further provides a storage medium for storing computer executable instruction information. In a specific embodiment, the storage medium may be a USB flash drive, an optical disk, a hard disk, etc. When the computer executable instruction information stored in the storage medium is executed by the processor, the following process can be implemented:

[0117] receiving a transaction complaint request submitted by a user for a target transaction, wherein the transaction complaint request includes a complaint risk type of the target transaction and first evidence information for the complaint risk type;

[0118] Inputting the complaint risk type and the first evidence information into a pre-trained large model to obtain a preliminary qualitative analysis result on whether the target transaction has a risk corresponding to the complaint risk type, and the large language model is used to perform a qualitative analysis on whether the transaction has a preset risk;

[0119] Sending the preliminary qualitative analysis result to the party complained against in the target transaction, and receiving feedback information from the party complained against regarding the preliminary qualitative analysis result;

[0120] If the feedback information includes a complaint request from the party complained against, then obtain second evidence information provided by the party complained against for the complaint request; based on the first evidence information and the second evidence information, perform a secondary qualitative analysis on whether the target transaction has a risk corresponding to the complaint risk type through the big model to obtain a secondary qualitative analysis result.

[0121] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the above-mentioned storage medium embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0122] The embodiment of this specification provides a storage medium, which receives a transaction complaint request submitted by a user for a target transaction, wherein the transaction complaint request includes a complaint risk type of the target transaction and first evidence information for the complaint risk type. Then, the complaint risk type and the first evidence information can be input into a pre-trained large model to obtain a preliminary qualitative analysis result on whether the target transaction has a risk corresponding to the complaint risk type, and the preliminary qualitative analysis result is sent to the complained party in the target transaction, and feedback information on the preliminary qualitative analysis result from the complained party is received. If the feedback information includes a complaint request from the complained party, second evidence information provided by the complained party for the complaint request is obtained, and based on the first evidence information and the second evidence information, a secondary qualitative analysis is performed on whether the target transaction has a risk corresponding to the complaint risk type through the large model. In this way, since the entire qualitative analysis process is Relying on big models avoids the time-consuming and subjective qualitative results of manual review and other processing. In addition, since the evidence information submitted by the reporting parties can be analyzed in detail, the information errors and omissions caused by manual review and other processing can be reduced. Moreover, the big model is responsible for the qualitative analysis of complaints and the analysis of the defense process of the complaining parties. The training and operating costs for manual work will also decrease, including reducing the training and organizational costs of auditors, and reducing the incentive costs and operating costs for users on the platform to participate in complaint courts. In addition, since the qualitative analysis of complaints is no longer limited to the perspective of the buyer, the big model can interpret text and pictures, and compare context in series. Therefore, the two parties to the transaction are allowed to provide evidence to each other regarding the complaint, making the qualitative analysis of the complaint more comprehensive and accurate. From the perspective of the merchants, the platform provides an entry so that the party being complained is no longer just passively accepting complaints, reducing operating costs and improving experience.

[0123] Furthermore, based on the above Figures 2 to 5 In one or more embodiments of the present specification, a computer program product is provided, including a computer program. When the computer program in the computer program product is executed by a processor, the following process can be implemented:

[0124] receiving a transaction complaint request submitted by a user for a target transaction, wherein the transaction complaint request includes a complaint risk type of the target transaction and first evidence information for the complaint risk type;

[0125] Inputting the complaint risk type and the first evidence information into a pre-trained large model to obtain a preliminary qualitative analysis result on whether the target transaction has a risk corresponding to the complaint risk type, and the large language model is used to perform a qualitative analysis on whether the transaction has a preset risk;

[0126] Sending the preliminary qualitative analysis result to the party complained against in the target transaction, and receiving feedback information from the party complained against regarding the preliminary qualitative analysis result;

[0127] If the feedback information includes a complaint request from the party complained against, then obtain second evidence information provided by the party complained against for the complaint request; based on the first evidence information and the second evidence information, perform a secondary qualitative analysis on whether the target transaction has a risk corresponding to the complaint risk type through the big model to obtain a secondary qualitative analysis result.

[0128] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the above-mentioned computer program product embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0129] The embodiment of the present specification provides a computer program product, which receives a transaction complaint request submitted by a user for a target transaction, wherein the transaction complaint request includes a complaint risk type of the target transaction and first evidence information for the complaint risk type. Then, the complaint risk type and the first evidence information can be input into a pre-trained large model to obtain a preliminary qualitative analysis result on whether the target transaction has a risk corresponding to the complaint risk type. The preliminary qualitative analysis result is sent to the complained party in the target transaction, and feedback information on the preliminary qualitative analysis result from the complained party is received. If the feedback information includes a complaint request from the complained party, second evidence information for the complaint request provided by the complained party is obtained. Based on the first evidence information and the second evidence information, a secondary qualitative analysis is performed on whether the target transaction has a risk corresponding to the complaint risk type through the large model. In this way, due to the entire qualitative analysis process, The whole process relies on big models, avoiding the time-consuming and subjective qualitative results of manual review and other processing. In addition, since it can analyze the evidence information submitted by the reporting parties in detail, it can reduce the occurrence of information errors and omissions caused by manual review and other processing. Moreover, the big model is responsible for the qualitative analysis of complaints and the analysis of the defense process of the complaining parties. The training and operating costs for manual work will also decrease accordingly, including reducing the training and organizational costs of auditors, and reducing the incentive costs and operating costs for users on the platform to participate in complaint courts. In addition, since the qualitative analysis of complaints is no longer limited to the perspective of the buyer, the big model can interpret text and pictures, and compare context in series. Therefore, the two parties to the transaction are allowed to provide evidence to each other regarding the complaint, making the qualitative analysis of the complaint more comprehensive and accurate. From the perspective of the merchants, the platform provides an entry so that the party being complained is no longer just passively accepting complaints, reducing operating costs and improving experience.

[0130] The above is a description of a specific embodiment of the specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0131] In the 1990s, improvements to a technology could be clearly distinguished as hardware improvements (for example, improvements to the circuit structure of diodes, transistors, switches, etc.) or software improvements (improvements to the method flow). However, with the development of technology, many improvements to the method flow today can be regarded as direct improvements to the hardware circuit structure. Designers almost always obtain the corresponding hardware circuit structure by programming the improved method flow into the hardware circuit. Therefore, it cannot be said that an improvement in a method flow cannot be implemented using a hardware entity module. For example, a programmable logic device (PLD) (such as a field programmable gate array (FPGA)) is such an integrated circuit whose logical function is determined by the user's programming of the device. Designers can "integrate" a digital system on a PLD by programming it themselves, without having to ask a chip manufacturer to design and produce a dedicated integrated circuit chip. Moreover, nowadays, instead of manually making integrated circuit chips, this kind of programming is mostly implemented by "logic compiler" software, which is similar to the software compiler used when developing and writing programs, and the original code before compilation must also be written in a specific programming language, which is called hardware description language (HDL). There is not only one HDL, but many kinds, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), Confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), Lava, Lola, MyHDL, PALASM, RHDL (Ruby Hardware Description Language), etc. The most commonly used ones are VHDL (Very-High-Speed ​​Integrated Circuit Hardware Description Language) and Verilog. Those skilled in the art should also know that it is only necessary to program the method flow slightly in the above-mentioned hardware description languages ​​and program it into the integrated circuit, and then it is easy to obtain the hardware circuit that implements the logic method flow.

[0132] The controller can be implemented in any appropriate manner, for example, the controller can take the form of a microprocessor or processor and a computer-readable medium storing a computer-readable program code (such as software or firmware) that can be executed by the (micro)processor, a logic gate, a switch, an application-specific integrated circuit (ASIC), a programmable logic controller, and an embedded microcontroller. Examples of controllers include, but are not limited to, the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicone Labs C8051F320. The memory controller can also be implemented as part of the control logic of the memory. Those skilled in the art also know that in addition to implementing the controller in a purely computer-readable program code manner, the controller can be implemented in the form of a logic gate, a switch, an application-specific integrated circuit, a programmable logic controller, and an embedded microcontroller by logically programming the method steps. Therefore, this controller can be considered as a hardware component, and the devices included therein for implementing various functions can also be regarded as structures within the hardware component. Or even, the devices for implementing various functions can be regarded as both software modules for implementing the method and structures within the hardware component.

[0133] The systems, devices, modules or units described in the above embodiments may be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, the computer may be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.

[0134] For the convenience of description, the above devices are described in terms of functions and are divided into various units. Of course, when implementing one or more embodiments of this specification, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0135] Those skilled in the art will appreciate that the embodiments of this specification may be provided as methods, systems, or computer program products. Therefore, one or more embodiments of this specification may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, one or more embodiments of this specification may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0136] The embodiments of this specification are described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of this specification. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable fraud case serial and parallel device to produce a machine, so that the instructions executed by the processor of the computer or other programmable fraud case serial and parallel device generate instructions for implementing the process in the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0137] These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable fraud case serial and parallel device to operate in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0138] These computer program instructions may also be loaded onto a computer or other programmable device in series and parallel to the process so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, whereby the instructions executed on the computer or other programmable device provide for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0139] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0140] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0141] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change 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), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0142] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0143] It should be understood by those skilled in the art that the embodiments of this specification may be provided as methods, systems or computer program products. Therefore, one or more embodiments of this specification may take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware. Moreover, one or more embodiments of this specification may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0144] One or more embodiments of the present specification may be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. One or more embodiments of the present specification may also be practiced in distributed computing environments where tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules may be located in local and remote computer storage media, including storage devices.

[0145] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0146] The above description is only an embodiment of this specification and is not intended to limit this document. For those skilled in the art, this specification may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of this specification should be included in the scope of the claims of this specification.

Claims

1. A qualitative analysis method for transaction risk, the method comprising: receiving a transaction complaint request submitted by a user for a target transaction, wherein the transaction complaint request includes a complaint risk type of the target transaction and first evidence information for the complaint risk type; Inputting the complaint risk type and the first evidence information into a pre-trained large model to obtain a preliminary qualitative analysis result on whether the target transaction has a risk corresponding to the complaint risk type; Sending the preliminary qualitative analysis result to the party complained against in the target transaction, and receiving feedback information from the party complained against regarding the preliminary qualitative analysis result; If the feedback information includes a complaint request from the party complained against, then obtain second evidence information provided by the party complained against for the complaint request; based on the first evidence information and the second evidence information, perform a secondary qualitative analysis on whether the target transaction has a risk corresponding to the complaint risk type through the big model to obtain a secondary qualitative analysis result.

2. The method according to claim 1, further comprising: Sending the preliminary qualitative analysis result to the user, and receiving first supplementary information for the target transaction submitted by the user; Based on the first evidence information and the second evidence information, a secondary qualitative analysis is performed on whether the target transaction has a risk corresponding to the complaint risk type through the large model to obtain a secondary qualitative analysis result, including: The first supplementary information, the first evidence information and the second evidence information are input into the large model to obtain a secondary qualitative analysis result on whether the target transaction has a risk corresponding to the complaint risk type.

3. The method according to claim 1, further comprising: Sending the secondary qualitative analysis result to the user and the party complained against; receiving second supplementary information for the target transaction, wherein the second supplementary information includes supplementary information submitted by the user and / or supplementary information submitted by the party against whom the complaint is filed; Based on the second supplementary information, the first evidence information and the second evidence information, performing another qualitative analysis on whether the target transaction has a risk corresponding to the complaint risk type through the big model, and obtaining another qualitative analysis result; The further qualitative analysis result is sent to the user and the party complained against.

4. The method according to claim 1, further comprising: If the feedback information includes a resource return request, obtaining a first amount of resources transferred by the user to the complained party in the target transaction, wherein the resource return request is used to request the return of the resources transferred by the user to the complained party in the target transaction; The acquired first amount of resources is transferred to the user.

5. The method according to claim 1, further comprising: If the feedback information includes a proactive withdrawal request, receiving a withdrawal request for the transaction complaint request submitted by the user within a first preset time period; If the cancellation request submitted by the user is received within a first preset time period, the response to the transaction complaint request is cancelled.

6. The method according to claim 1, further comprising: If the feedback information includes a business rectification request, restricting the transaction rights of the party complained against, and receiving rectification result data submitted by the party complained against within a second preset time period; If the rectification result data meets the preset business rectification rules, the restriction on the transaction authority of the party complained against is cancelled.

7. The method according to claim 1, if the feedback information includes a complaint request from the party being complained against, before obtaining second evidence information provided by the party being complained against regarding the complaint request, the method further comprises: Inputting the first evidence information into a pre-trained large model to determine parsing information corresponding to the first evidence information, wherein the parsing information includes event summary information of an event corresponding to the target transaction and one or more of the evidence element information in the first evidence information; If the feedback information includes the complaint request of the party against whom the complaint is made, obtaining second evidence information provided by the party against whom the complaint is made in response to the complaint request includes: If the feedback information includes a request for appeal from the party being complained against, the feedback information and the analysis information are input into the big model to obtain the voucher information required to be submitted by the party being complained against the preliminary qualitative analysis result; Receive second evidence information provided by the party complained against and obtained based on the credential information.

8. According to the method described in claim 7, the type of information contained in the first evidence information includes one or more of text, image, audio, and video; the type of information contained in the second evidence information includes one or more of text, image, audio, and video.

9. A qualitative analysis device for transaction risk, comprising: A request module, receiving a transaction complaint request submitted by a user for a target transaction, wherein the transaction complaint request includes a complaint risk type of the target transaction and first evidence information for the complaint risk type; A preliminary qualitative analysis module, which inputs the complaint risk type and the first evidence information into a pre-trained large model to obtain a preliminary qualitative analysis result of whether the target transaction has a risk corresponding to the complaint risk type, and the large language model is used to perform a qualitative analysis on whether the transaction has a preset risk; A feedback receiving module, which sends the preliminary qualitative analysis result to the complained party in the target transaction, and receives feedback information from the complained party regarding the preliminary qualitative analysis result; The secondary qualitative analysis module obtains second evidence information provided by the party complained against for the appeal request if the feedback information includes the appeal request; based on the first evidence information and the second evidence information, a secondary qualitative analysis is performed on whether the target transaction has the risk corresponding to the complaint risk type through the large model to obtain a secondary qualitative analysis result.

10. A qualitative analysis device for transaction risk, the qualitative analysis device for transaction risk comprising: processor; as well as a memory arranged to store computer executable instructions which, when executed, cause the processor to: receiving a transaction complaint request submitted by a user for a target transaction, wherein the transaction complaint request includes a complaint risk type of the target transaction and first evidence information for the complaint risk type; Inputting the complaint risk type and the first evidence information into a pre-trained large model to obtain a preliminary qualitative analysis result on whether the target transaction has a risk corresponding to the complaint risk type, and the large language model is used to perform a qualitative analysis on whether the transaction has a preset risk; Sending the preliminary qualitative analysis result to the party complained against in the target transaction, and receiving feedback information from the party complained against regarding the preliminary qualitative analysis result; If the feedback information includes a complaint request from the party complained against, then obtain second evidence information provided by the party complained against for the complaint request; based on the first evidence information and the second evidence information, perform a secondary qualitative analysis on whether the target transaction has a risk corresponding to the complaint risk type through the big model to obtain a secondary qualitative analysis result.