Platform risk event intelligent response system and method based on artificial intelligence

Abstract

The invention discloses a platform risk event intelligent response system and method based on artificial intelligence, and relates to the technical field of risk detection.The method comprises the steps that commodity information published on an e-commerce platform and commodity comment information in historical transactions are collected, and the difference degree of the commodity information and the commodity comment information is calculated; screening target commodities with the difference exceeding the standard, and calculating a risk early warning coefficient in combination with the information modification rate, the complaint rate and the return rate; dividing risk levels by taking the risk early warning coefficient and the difference degree as two-dimensional indexes, matching a corresponding risk response strategy according to the risk levels, constructing a multi-dimensional evaluation system to calculate a response effect coefficient, and dynamically adjusting the risk response strategy; according to the invention, intelligent identification and dynamic response of risk events can be realized, and user rights and interests and platform reputation are guaranteed.

Description

Technical Field

[0001] The present invention relates to the field of risk detection technology, and in particular to an artificial intelligence-based intelligent response system and method for platform risk events. Background Art

[0002] An e-commerce platform is an online trading platform established on the internet that provides online transaction and payment functions for buyers and sellers. As e-commerce matures and its use becomes more widespread, the issue of risk is becoming increasingly prominent. To ensure the sustainable development of e-commerce platforms, platforms often rely on manual sampling or keyword blocking to prevent false advertising. Manual sampling is not only inefficient but also inaccurate. Regarding keyword blocking, some unscrupulous merchants can circumvent the platform's keyword blocking rules through homophonic substitution, synonym substitution, and word splitting and reorganization, allowing them to continue publishing misleading product information. Therefore, an intelligent, multi-dimensional risk event response method is urgently needed to address these issues. Summary of the Invention

[0003] The purpose of the present invention is to provide an artificial intelligence-based intelligent response system and method for platform risk events to solve the problems raised in the prior art.

[0004] To achieve the above objectives, the present invention provides the following technical solution: an artificial intelligence-based intelligent response method for platform risk events, the intelligent response method comprising: Step S1: collecting product information published on the e-commerce platform and pre-processing the product information; obtaining product review information in the historical transaction records of the e-commerce platform, and calculating the degree of difference between the product information and the product review information; Step S2: obtaining a target product based on the calculated difference degree; tracking the product information modification records of the target product and calculating the information modification rate; and calculating the risk warning coefficient based on the complaint rate and return rate of the target product in historical transactions; Step S3: classifying the target products according to the calculated risk warning coefficient and the degree of difference; and executing risk response strategies for target products of different levels; Step S4: monitor the response effect of the risk response strategy in real time and calculate the response effect coefficient; dynamically adjust the risk response strategy according to the calculated response effect coefficient.

[0005] Furthermore, step S1 includes: Step S1-1: Acquire product information of all products in real time through the data interface of the e-commerce platform; the product information includes product text data, product image data, and product video data; collect product review information from user feedback in historical transaction records; the product review information includes actual product text data, actual product image data, and actual product video data; Step S1-2: Extracting text feature vectors from the product text data and image feature vectors from the product image data using the model; processing the video frame sequence of each product video in the product video data to extract dynamic feature vectors; extracting actual text feature vectors, actual image feature vectors, and actual dynamic feature vectors from the product review information; and calculating the text difference degree, image difference degree, and video difference degree, respectively; The degree of text difference is: ; Among them, D t Indicates the degree of text difference, T a represents the text feature vector, T b Represents the actual text feature vector; The image difference degree is: ; Among them, D p Represents the image difference, P a represents the image feature vector, P b Represents the actual image feature vector; The degree of video difference is: ; Among them, D m Indicates the degree of video difference, M a Represents the video feature vector, M b represents the actual video vector; Step S1-3: Calculate the degree of difference between the product information and product review information published on the e-commerce platform: ; Among them, D represents the degree of difference between the product information and product review information published on the e-commerce platform, and α, β, and γ are all weight coefficients; The above-mentioned "actual text data of the product" includes users' textual evaluations of the product, textual descriptions of the reasons for return, textual descriptions of the complaint content, etc., which are converted into structured actual text data of the product through data cleaning, preprocessing and other methods. "Actual image data of the product" refers to the actual photos of the product uploaded by the user, and "actual video data of the product" refers to the actual videos of the product taken by the user. In the above steps, for text data, natural language processing models such as BERT and WordVec are used to extract semantic vectors from product titles and detailed descriptions. For image data, convolutional neural networks such as ResNet and VGG are used to extract color, texture, shape and other features in the image. For video data, dynamic features in the video are extracted through 3D neural networks or frame sequence processing. All extracted features are finally converted into numerical vectors of uniform dimension to ensure that the differences can be calculated through cosine similarity.

[0006] Furthermore, step S2 includes: Step S2-1: Identify the product categories of all products published on the e-commerce platform, divide all products according to the identified product categories, and establish a difference degree benchmark library; calculate the mean difference degree of each product category according to the product category and standard deviation , generate the difference degree threshold corresponding to each product category: ; in, is the adjustment coefficient; when the difference degree of a certain product D≥D th , marking the product as a target product; Step S2-2: Monitor the product information of the target product, count the number of modifications to the target product information within the target time period, and extract text modification data, image modification data, and video modification data from each product information modification record; associate the target product's production batch information with the preset product traceability code to determine whether the information modification corresponds to the product review information; if the information modification is consistent with the product review information, calculate the modification rationality coefficient: ; Among them, H i Indicates the modification rationality coefficient of the i-th modification record, Indicates the degree of difference in text features before and after modification, Indicates the degree of difference in image features before and after modification, Indicates the degree of difference in video features before and after modification. 、 、 are the difference weights of text, image, and video respectively; Step S2-3: Setting the modification rationality coefficient threshold H th , when H i ≥H th When H i <H th , it is marked as an invalid modification; when the product information of the target product is invalidly modified or inconsistent with the product review information, the text feature change amount, image feature change amount, and video feature change amount are generated according to the text modification data, image modification data, and video modification data; the information modification rate of the target product is calculated: ; Where R represents the information modification rate, ∆D ti represents the text feature change in the i-th product information modification record, ∆D pi represents the change in image features in the i-th product information modification record, ∆D mi Indicates the change in video features in the i-th product information modification record. is the time weight, Indicates the target time period; Step S2-4: Collect the total number of completed orders, complaint orders, and return orders for the target product in historical transactions, and calculate the complaint rate and return rate of the target product; Stated complaint rates: ; Where C represents the complaint rate, m represents the number of complaint orders for the target product in historical transactions, and Q represents the total number of completed orders for the target product in historical transactions; Stated return rates: ; Where S represents the return rate, and n represents the number of return orders for the target product in historical transactions; Step S2-5: Calculate the risk warning coefficient of the target product based on the information modification rate, complaint rate, and return rate of the target product: ; Among them, K represents the risk warning coefficient of the target product, 、 、 All are weight coefficients; In the above steps, the "product category" identification method can be implemented through two layers of logic: the basic layer: preliminary classification based on the category labels filled in by the merchant (such as "Women's clothing - shirts"); the verification layer: using a text classification model (such as TextCNN) to perform semantic analysis on product titles and descriptions and correct incorrectly filled categories.

[0007] Furthermore, step S3 includes: Step S3-1: Obtain all product information from the risk event records of the e-commerce platform, divide the product information by product category, extract all historical product information of the same product category as the target product, calculate the risk warning coefficient and difference degree based on the historical product information, and divide the risk coefficient interval and difference degree interval according to the set risk coefficient threshold and difference degree threshold; Step S3-2: Using the risk warning coefficient and the degree of difference as two-dimensional division indicators, a risk level division matrix is ​​established; the horizontal direction of the risk level division matrix is ​​the risk warning coefficient interval, and the vertical direction is the degree of difference interval. The risk level of the target product is determined by the cross-combination of the horizontal risk warning coefficient and the vertical degree of difference; Step S3-3: For target commodities of different risk levels, a preset risk response strategy library is matched, wherein the risk response strategy library includes dynamic response rules; the dynamic response rules include: for high-risk commodities, triggering an immediate delisting verification mechanism, synchronously starting a credit rating deduction process, and pushing risk alerts to users of the e-commerce platform; for medium-risk commodities, implementing mandatory product information labeling and correction measures, restricting medium-risk commodities from participating in promotional activities of the e-commerce platform, and requiring submission of a product qualification re-inspection report within a preset time; for low-risk commodities, sending a risk warning to the backend, prompting merchants to make independent rectifications on low-risk commodities, and the warning status can be lifted only after the rectification is completed and the commodities pass the re-inspection; the risk level classification results are associated and stored with the risk response strategy execution records; In the above steps, the scope of "risk event records" not only includes information on products that have been historically judged to be risky, but also includes dispute orders caused by inconsistent product information, user reporting records, platform penalty cases, etc., to ensure that the extracted historical data can fully reflect the risk characteristics.

[0008] Furthermore, step S4 includes: Step S4-1: Construct a multi-dimensional evaluation index system for response effects, including risk mitigation rate, user satisfaction improvement rate, and timeliness compliance rate; The risk mitigation rate is calculated by counting the number of new complaint orders and new return orders for the target product within the preset monitoring period after the risk response strategy is implemented. ; Among them, E1 represents the risk mitigation rate, Q 前 represents the number of complaint orders and return orders in the same monitoring period before the risk response strategy is implemented, Q 后 Indicates the number of complaint orders and return orders within the preset monitoring period after the risk response strategy is executed; User satisfaction improvement rate: The user evaluation data of the target product is collected through the user feedback system of the e-commerce platform, and the sentiment tendency features in the evaluation text are extracted to calculate the user satisfaction improvement rate: ; Among them, E2 represents the improvement rate of user satisfaction, S 前 represents the mean user satisfaction score before the risk response strategy is implemented, S 后 represents the mean user satisfaction score after the risk response strategy is implemented; The timeliness compliance rate is calculated by recording the time from receiving the response policy instruction to completing the rectification and passing the re-inspection of the e-commerce platform; ; Among them, T 实 ≤T 标 When T 实 >T 标 When E3=0, E3 represents the timeliness compliance rate, T 标 Indicates the standard rectification time of the target product, T 实 Indicates the actual rectification time; Step S4-2: Determine the weight coefficient of each evaluation indicator through the hierarchical analysis method and calculate the response effect coefficient: ; in, 、 、 These are the weight coefficients of risk mitigation rate, user satisfaction improvement rate, and timeliness compliance rate; Step S4-3: Establish a dynamic adjustment model for the risk response strategy and optimize the risk response strategy based on the response effect coefficient.

[0009] Furthermore, in order to better implement the above method, an artificial intelligence-based platform risk event intelligent response system is also provided, which includes: an information collection and difference analysis module, a risk warning coefficient calculation module, a risk classification and response module, and an effect evaluation and strategy adjustment module; The information collection and difference analysis module is used to collect product information published on the e-commerce platform and product review information in historical transactions, pre-process this information and calculate the degree of difference; A risk warning coefficient calculation module is used to screen out target products with excessive discrepancies, track information modification records of the target products and calculate the information modification rate. This module then calculates the risk warning coefficient based on the complaint rate and return rate in historical transactions. The risk grading and response module categorizes target commodities according to risk warning coefficients and degree of difference, and implements corresponding risk response strategies for commodities of different grades; The effect evaluation and strategy adjustment module conducts a multi-dimensional evaluation of the execution effect of the risk response strategy, calculates the response effect coefficient, and dynamically adjusts the risk response strategy based on the response effect coefficient.

[0010] Furthermore, the information collection and difference analysis module includes: an information collection unit, a feature extraction unit, and a difference calculation unit; The information collection unit obtains the product information published on the e-commerce platform and the product review information in the historical transaction records in real time through the data interface of the e-commerce platform; A feature extraction unit is used to extract text feature vectors from product text data, extract image feature vectors from product image data, extract dynamic feature vectors after processing the video frame sequence of product video data; and extract corresponding actual feature vectors from product review information; The difference calculation unit calculates the text difference, image difference, and video difference respectively, and calculates the overall difference between the product information and the product review information by combining the weight coefficient; Furthermore, the risk warning coefficient calculation module includes: a target product identification unit, a modification rate calculation unit, a complaint return rate calculation unit, and a warning coefficient unit; The target product identification unit divides products into categories and establishes a difference degree benchmark library. It calculates the mean and standard deviation of the difference degree of each category of products, generates a difference degree threshold, and marks products with a difference degree exceeding the threshold as target products. The modification rate calculation unit is used to monitor the information modification records of the target product within the target time period, extract various modification data and generate feature variables, and calculate the information modification rate based on the time weight; A complaint return rate calculation unit is used to collect the total number of completed orders, complaint orders, and return orders in historical transactions of the target product, and calculate the complaint rate and return rate respectively; The early warning coefficient unit calculates the risk early warning coefficient based on the information modification rate, complaint rate, return rate, and corresponding weight coefficients; Furthermore, the risk classification and response module includes: an interval division unit, a level determination unit, and a strategy execution unit; An interval division unit extracts historical product information belonging to the same category as the target product, calculates the risk warning coefficient and difference degree of the historical product information, and divides the risk warning coefficient interval and difference degree interval according to a set critical value; The level determination unit uses the risk warning coefficient and the degree of difference as two-dimensional indicators to establish a risk level classification matrix. The risk level of the target product is determined by the cross-combination of the horizontal risk warning coefficient and the vertical degree of difference. The strategy execution unit matches and executes the dynamic response rules in the preset risk response strategy library for commodities with different risk levels, and associates and stores the risk level classification results with the response strategy execution records.

[0011] Furthermore, the effect evaluation and strategy adjustment module includes: an evaluation index unit, an effect coefficient unit, and a strategy adjustment unit; Evaluation indicator unit: building a multi-dimensional response effect evaluation indicator system including risk mitigation rate, user satisfaction improvement rate, and timeliness compliance rate; The effect coefficient unit determines the weight coefficient of each evaluation indicator through the hierarchical analysis method, and calculates the response effect coefficient based on the numerical value of each indicator; The strategy adjustment unit establishes a dynamic adjustment model for the risk response strategy and optimizes and adjusts the risk response strategy based on the response effect coefficient.

[0012] Compared with the existing technology, the beneficial effects of the present invention are: the present invention calculates the degree of difference between the product information and product review information published on the e-commerce platform through multi-dimensional feature extraction, and combines the difference degree benchmark library to accurately screen out target products whose product information published on the platform is inconsistent with the product review information; establishes a two-dimensional level matrix through risk warning coefficient and difference degree, and implements risk response strategies for different risk levels, taking into account risk prevention and control and merchant operation flexibility; evaluates response results through multiple indicators such as risk mitigation rate and user satisfaction improvement, and combines dynamic adjustment models to continuously optimize strategies, thereby improving the timeliness and effectiveness of risk handling, enhancing user trust, and ensuring the sustainable development of the platform. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 This is a flow chart of a method for an artificial intelligence-based platform risk event intelligent response system and method of the present invention; Figure 2 This is a schematic diagram of the system structure of an artificial intelligence-based platform risk event intelligent response system and method of the present invention. DETAILED DESCRIPTION

[0014] Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative work shall fall within the scope of protection of the present invention.

[0015] Example 1: Figure 1 As shown, the present invention provides a technical solution, an artificial intelligence-based intelligent response method for platform risk events, the intelligent response method comprising: Step S1: Collecting product information published by all merchants on the e-commerce platform and pre-processing the product information; obtaining product review information published by all buyers in the historical transaction records of the e-commerce platform, and calculating the degree of difference between the product information and the product review information; Wherein, step S1 includes: Step S1-1: Acquire product information of all products in real time through the data interface of the e-commerce platform; the product information includes product text data, product image data, and product video data; collect product review information from user feedback in historical transaction records; the product review information includes actual product text data, actual product image data, and actual product video data; Step S1-2: Extracting text feature vectors from the product text data and image feature vectors from the product image data using the model; processing the video frame sequence of each product video in the product video data to extract dynamic feature vectors; extracting actual text feature vectors, actual image feature vectors, and actual dynamic feature vectors from the product review information; and calculating the text difference degree, image difference degree, and video difference degree, respectively; The degree of text difference is: ; Among them, D t Indicates the degree of text difference, T a represents the text feature vector, T b Represents the actual text feature vector; The image difference degree is: ; Among them, D p Represents the image difference, P a represents the image feature vector, P b Represents the actual image feature vector; The degree of video difference is: ; Among them, D m Indicates the degree of video difference, M a Represents the video feature vector, M b represents the actual video vector; Step S1-3: Calculate the degree of difference between the product information and product review information published on the e-commerce platform: ; Among them, D represents the degree of difference between the product information and product review information published on the e-commerce platform, and α, β, and γ are all weight coefficients; Step S2: obtaining a target product based on the calculated difference degree; tracking the product information modification records of the target product and calculating the information modification rate; and calculating the risk warning coefficient based on the complaint rate and return rate of the target product in historical transactions; Wherein, step S2 includes: Step S2-1: Identify the product categories of all products published on the e-commerce platform, divide all products according to the identified product categories, and establish a difference degree benchmark library; calculate the mean difference degree of each product category according to the product category and standard deviation , generate the difference degree threshold corresponding to each product category: ; in, is the adjustment coefficient; when the difference degree of a certain product D≥D th , marking the product as a target product; Step S2-2: Monitor the product information of the target product, count the number of modifications to the target product information within the target time period, and extract text modification data, image modification data, and video modification data from each product information modification record; associate the target product's production batch information with the preset product traceability code to determine whether the information modification corresponds to the product review information; if the information modification is consistent with the product review information, calculate the modification rationality coefficient: ; Among them, H i Indicates the modification rationality coefficient of the i-th modification record, Indicates the degree of difference in text features before and after modification, Indicates the degree of difference in image features before and after modification, Indicates the degree of difference in video features before and after modification. 、 、 are the difference weights of text, image, and video respectively; Step S2-3: Setting the modification rationality coefficient threshold H th , when H i ≥H th When H i <H th , it is marked as an invalid modification; when the product information of the target product is invalidly modified or inconsistent with the product review information, the text feature change amount, image feature change amount, and video feature change amount are generated according to the text modification data, image modification data, and video modification data; the information modification rate of the target product is calculated: ; Where R represents the information modification rate, ∆D ti represents the text feature change in the i-th product information modification record, ∆D pi represents the change in image features in the i-th product information modification record, ∆D mi Indicates the change in video features in the i-th product information modification record. is the time weight, Indicates the target time period; Step S2-4: Collect the total number of completed orders, complaint orders, and return orders for the target product in historical transactions, and calculate the complaint rate and return rate of the target product; Stated complaint rates: ; Where C represents the complaint rate, m represents the number of complaint orders for the target product in historical transactions, and Q represents the total number of completed orders for the target product in historical transactions; Stated return rates: ; Where S represents the return rate, and n represents the number of return orders for the target product in historical transactions; Step S2-5: Calculate the risk warning coefficient of the target product based on the information modification rate, complaint rate, and return rate of the target product: ; Among them, K represents the risk warning coefficient of the target product, 、 、 All are weight coefficients; Step S3: classifying the target products according to the calculated risk warning coefficient and the degree of difference; and executing risk response strategies for target products of different levels; Wherein, step S3 includes: Step S3-1: Obtain all product information from the risk event records of the e-commerce platform, divide the product information by product category, extract all historical product information of the same product category as the target product, calculate the risk warning coefficient and difference degree based on the historical product information, and divide the risk coefficient interval and difference degree interval according to the set risk coefficient threshold and difference degree threshold; Step S3-2: Using the risk warning coefficient and the degree of difference as two-dimensional division indicators, a risk level division matrix is ​​established; the horizontal direction of the risk level division matrix is ​​the risk warning coefficient interval, and the vertical direction is the degree of difference interval. The risk level of the target product is determined by the cross-combination of the horizontal risk warning coefficient and the vertical degree of difference; Step S3-3: For target commodities of different risk levels, a preset risk response strategy library is matched, wherein the risk response strategy library includes dynamic response rules; the dynamic response rules include: for high-risk commodities, triggering an immediate delisting verification mechanism, synchronously starting a credit rating deduction process, and pushing risk alerts to users of the e-commerce platform; for medium-risk commodities, implementing mandatory product information labeling and correction measures, restricting medium-risk commodities from participating in promotional activities of the e-commerce platform, and requiring submission of a product qualification re-inspection report within a preset time; for low-risk commodities, sending a risk warning to the backend, prompting merchants to make independent rectifications on low-risk commodities, and the warning status can be lifted only after the rectification is completed and the commodities pass the re-inspection; the risk level classification results are associated and stored with the risk response strategy execution records; Step S4: monitoring the response effect of the risk response strategy in real time and calculating the response effect coefficient; dynamically adjusting the risk response strategy according to the calculated response effect coefficient; Wherein, step S4 includes: Step S4-1: Construct a multi-dimensional evaluation index system for response effects, including risk mitigation rate, user satisfaction improvement rate, and timeliness compliance rate; The risk mitigation rate is calculated by counting the number of new complaint orders and new return orders for the target product within the preset monitoring period after the risk response strategy is implemented. ; Among them, E1 represents the risk mitigation rate, Q 前 represents the number of complaint orders and return orders in the same monitoring period before the risk response strategy is implemented, Q 后 Indicates the number of complaint orders and return orders within the preset monitoring period after the risk response strategy is executed; User satisfaction improvement rate: The user evaluation data of the target product is collected through the user feedback system of the e-commerce platform, and the sentiment tendency features in the evaluation text are extracted to calculate the user satisfaction improvement rate: ; Among them, E2 represents the improvement rate of user satisfaction, S 前 represents the mean user satisfaction score before the risk response strategy is implemented, S 后 represents the mean user satisfaction score after the risk response strategy is implemented; The timeliness compliance rate is calculated by recording the time from receiving the response policy instruction to completing the rectification and passing the re-inspection of the e-commerce platform; ; Among them, T 实 ≤T 标When T 实 >T 标 When E3=0, E3 represents the timeliness compliance rate, T 标 Indicates the standard rectification time of the target product, T 实 Indicates the actual rectification time; Step S4-2: Determine the weight coefficient of each evaluation indicator through the hierarchical analysis method and calculate the response effect coefficient: ; in, 、 、 These are the weight coefficients of risk mitigation rate, user satisfaction improvement rate, and timeliness compliance rate; Step S4-3: Establish a dynamic adjustment model for the risk response strategy and optimize the risk response strategy based on the response effect coefficient; Example 2: Figure 2 As shown, in order to better implement the above method, an artificial intelligence-based platform risk event intelligent response system is also provided, which includes: an information collection and difference analysis module, a risk warning coefficient calculation module, a risk classification and response module, and an effect evaluation and strategy adjustment module; The information collection and difference analysis module is used to collect product information published on the e-commerce platform and product review information in historical transactions, pre-process this information and calculate the degree of difference; A risk warning coefficient calculation module is used to screen out target products with excessive discrepancies, track information modification records of the target products and calculate the information modification rate. This module then calculates the risk warning coefficient based on the complaint rate and return rate in historical transactions. The risk grading and response module categorizes target commodities according to risk warning coefficients and degree of difference, and implements corresponding risk response strategies for commodities of different grades; The effect evaluation and strategy adjustment module conducts a multi-dimensional evaluation of the execution effect of the risk response strategy, calculates the response effect coefficient, and dynamically adjusts the risk response strategy based on the response effect coefficient; Among them, the information collection and difference analysis module includes: information collection unit, feature extraction unit, and difference calculation unit; The information collection unit obtains the product information published on the e-commerce platform and the product review information in the historical transaction records in real time through the data interface of the e-commerce platform; A feature extraction unit is used to extract text feature vectors from product text data, extract image feature vectors from product image data, extract dynamic feature vectors after processing the video frame sequence of product video data; and extract corresponding actual feature vectors from product review information; The difference calculation unit calculates the text difference, image difference, and video difference respectively, and calculates the overall difference between the product information and the product review information by combining the weight coefficient; Among them, the risk warning coefficient calculation module includes: target product identification unit, modification rate calculation unit, complaint return rate calculation unit, and warning coefficient unit; The target product identification unit divides products into categories and establishes a difference degree benchmark library. It calculates the mean and standard deviation of the difference degree of each category of products, generates a difference degree threshold, and marks products with a difference degree exceeding the threshold as target products. The modification rate calculation unit is used to monitor the information modification records of the target product within the target time period, extract various modification data and generate feature variables, and calculate the information modification rate based on the time weight; A complaint return rate calculation unit is used to collect the total number of completed orders, complaint orders, and return orders in historical transactions of the target product, and calculate the complaint rate and return rate respectively; The early warning coefficient unit calculates the risk early warning coefficient based on the information modification rate, complaint rate, return rate, and corresponding weight coefficients; Among them, the risk classification and response module includes: interval division unit, level determination unit, and strategy execution unit; An interval division unit extracts historical product information belonging to the same category as the target product, calculates the risk warning coefficient and difference degree of the historical product information, and divides the risk warning coefficient interval and difference degree interval according to a set critical value; The level determination unit uses the risk warning coefficient and the degree of difference as two-dimensional indicators to establish a risk level classification matrix. The risk level of the target product is determined by the cross-combination of the horizontal risk warning coefficient and the vertical degree of difference. The strategy execution unit matches and executes dynamic response rules in the preset risk response strategy library for commodities of different risk levels, and simultaneously associates and stores the risk level classification results with the response strategy execution records; Among them, the effect evaluation and strategy adjustment module includes: evaluation index unit, effect coefficient unit, strategy adjustment unit; Evaluation indicator unit: building a multi-dimensional response effect evaluation indicator system including risk mitigation rate, user satisfaction improvement rate, and timeliness compliance rate; The effect coefficient unit determines the weight coefficient of each evaluation indicator through the hierarchical analysis method, and calculates the response effect coefficient based on the numerical value of each indicator; The strategy adjustment unit establishes a dynamic adjustment model for risk response strategies and optimizes and adjusts risk response strategies based on the response effect coefficient; In the embodiment of the present invention, taking the product “pure cotton shirt” on a certain platform as an example: The system uses the platform interface to obtain information about shirts posted by merchants (text description: "100% cotton," image of a smooth white shirt, video demonstrating the softness of the fabric). It also extracts actual feedback from buyers in historical transactions (text labeling: "Contains 30% chemical fiber," buyer-uploaded images of the shirt showing obvious pilling, and videos showing the fabric as stiff). By extracting feature vectors and calculating cosine similarity, we can get the text difference degree D. t =0.6, image difference degree D p =0.5, video difference degree D m =0.4, combined with weights α=0.4, β=0.3, γ=0.3, the total difference degree D=0.4×0.6+0.3×0.5+0.3×0.4=0.51; The average difference degree of the "Men's Tops" category described by the shirt D =0.2, standard deviation D =0.15, adjustment coefficient =2, threshold D th =0.2+2×0.15=0.5; because D=0.51>0.5, it is marked as the target product; Tracking the target product's information modification records within 30 days, a total of 2 modifications were made (the text was changed from "pure cotton" to "cotton blended", and the image was replaced once). The information modification rate R = 0.3 was calculated. In historical transactions, the total order Q = 200, the number of complaints m = 30, and the number of returns n = 40, resulting in a complaint rate C = 0.15 and a return rate S = 0.2. Combined with the weights 1=0.2, 2=0.4, 3=0.4, risk warning coefficient K=0.2×0.3+0.4×0.15+0.4×0.2=0.18; Historical data from the "Men's Tops" category shows that a risk warning coefficient greater than 0.15 and a degree of difference greater than 0.5 indicate medium risk. Therefore, this shirt was determined to be medium risk, and the response strategy was implemented: mandatory labeling of "Contains 30% chemical fiber," restrictions on participating in platform promotions, and a requirement for merchants to submit fabric testing reports within three days. 30-day monitoring after implementation: The original number of complaints and returns in the cycle was 70, but now it has dropped to 20, with a risk mitigation rate of E1 = (70-20) / 70 = 0.71; user satisfaction increased from 3.2 points to 4.5 points, with an improvement rate of E2 = (4.5-3.2) / 3.2 = 0.41; the merchant completed the rectification in 2 days, with a timeliness compliance rate of E3 = 1; the response effect coefficient is qualified, and the strategy remains unchanged.

[0016] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art will be able to modify the technical solutions described in the aforementioned embodiments or substitute equivalents for some of the technical features. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

Claims

1. An artificial intelligence-based intelligent response method for platform risk events, characterized by: The intelligent response method includes: Step S1: collecting product information published on the e-commerce platform and pre-processing the product information; obtaining product review information in the historical transaction records of the e-commerce platform, and calculating the degree of difference between the product information and the product review information; Step S2: obtaining a target product based on the calculated difference degree; tracking the product information modification records of the target product and calculating the information modification rate; and calculating the risk warning coefficient based on the complaint rate and return rate of the target product in historical transactions; Step S3: classifying the target products according to the calculated risk warning coefficient and the degree of difference; and executing risk response strategies for target products of different levels; Step S4: monitor the response effect of the risk response strategy in real time and calculate the response effect coefficient; dynamically adjust the risk response strategy according to the calculated response effect coefficient.

2. The method for intelligently responding to platform risk events based on artificial intelligence according to claim 1, characterized in that: Step S1 includes: Step S1-1: Acquire product information of all products in real time through the data interface of the e-commerce platform; the product information includes product text data, product image data, and product video data; collect product review information from user feedback in historical transaction records; the product review information includes actual product text data, actual product image data, and actual product video data; Step S1-2: Extracting text feature vectors from the product text data and image feature vectors from the product image data using the model; processing the video frame sequence of each product video in the product video data to extract dynamic feature vectors; extracting actual text feature vectors, actual image feature vectors, and actual dynamic feature vectors from the product review information; and calculating the text difference degree, image difference degree, and video difference degree, respectively; The degree of text difference is: ； Among them, D t Indicates the degree of text difference, T a represents the text feature vector, T b Represents the actual text feature vector; The degree of image difference is: ； Among them, D p Represents the image difference, P a represents the image feature vector, P b Represents the actual image feature vector; The degree of video difference is: ； Among them, D m Indicates the degree of video difference, M a Represents the video feature vector, M b Represents the actual video vector; Step S1-3: Calculate the degree of difference between the product information and product review information published on the e-commerce platform: ； Among them, D represents the degree of difference between the product information and product review information published on the e-commerce platform, and α, β, and γ are all weight coefficients.

3. The method for intelligently responding to platform risk events based on artificial intelligence according to claim 1, characterized in that: Step S2 includes: Step S2-1: Identify the product categories of all products published on the e-commerce platform, divide all products according to the identified product categories, and establish a difference degree benchmark library; calculate the mean difference degree of each product category according to the product category and standard deviation , generate the difference degree threshold corresponding to each product category: ； in, is the adjustment coefficient; when the difference degree of a certain product D≥D th , marking the product as a target product; Step S2-2: Monitor the product information of the target product, count the number of modifications to the target product information within the target time period, and extract text modification data, image modification data, and video modification data from each product information modification record; associate the target product's production batch information with the preset product traceability code to determine whether the information modification corresponds to the product review information; if the information modification is consistent with the product review information, calculate the modification rationality coefficient: ； Among them, H i Indicates the modification rationality coefficient of the i-th modification record, Indicates the degree of difference in text features before and after modification, Indicates the degree of difference in image features before and after modification, Indicates the degree of difference in video features before and after modification. 、 、 are the difference weights of text, image, and video respectively; Step S2-3: Setting the modification rationality coefficient threshold H th , when H i ≥H th When H i <H th , it is marked as an invalid modification; when the product information of the target product is invalidly modified or inconsistent with the product review information, the text feature change amount, image feature change amount, and video feature change amount are generated according to the text modification data, image modification data, and video modification data; the information modification rate of the target product is calculated: ； Where R represents the information modification rate, ∆D ti represents the text feature change in the i-th product information modification record, ∆D pi represents the change in image features in the i-th product information modification record, ∆D mi Indicates the change in video features in the i-th product information modification record. is the time weight, Indicates the target time period; Step S2-4: Collect the total number of completed orders, complaint orders, and return orders for the target product in historical transactions, and calculate the complaint rate and return rate of the target product; Stated complaint rates: ； Where C represents the complaint rate, m represents the number of complaint orders for the target product in historical transactions, and Q represents the total number of completed orders for the target product in historical transactions; Stated return rates: ； Where S represents the return rate, and n represents the number of return orders for the target product in historical transactions; Step S2-5: Calculate the risk warning coefficient of the target product based on the information modification rate, complaint rate, and return rate of the target product: ； Among them, K represents the risk warning coefficient of the target product, 、 、 are all weight coefficients.

4. The method for intelligently responding to platform risk events based on artificial intelligence according to claim 1, characterized in that: Step S3 includes: Step S3-1: Obtain all product information from the risk event records of the e-commerce platform, divide the product information by product category, extract all historical product information of the same product category as the target product, calculate the risk warning coefficient and difference degree based on the historical product information, and divide the risk coefficient interval and difference degree interval according to the set risk coefficient threshold and difference degree threshold; Step S3-2: Using the risk warning coefficient and the degree of difference as two-dimensional division indicators, a risk level division matrix is ​​established; the horizontal direction of the risk level division matrix is ​​the risk warning coefficient interval, and the vertical direction is the degree of difference interval. The risk level of the target product is determined by the cross-combination of the horizontal risk warning coefficient and the vertical degree of difference; Step S3-3: For target commodities of different risk levels, a preset risk response strategy library is matched, and the risk response strategy library includes dynamic response rules; the dynamic response rules include: for high-risk commodities, triggering an immediate delisting verification mechanism, synchronously starting a credit rating deduction process, and pushing risk warnings to users of the e-commerce platform; for medium-risk commodities, implementing mandatory product information labeling and correction measures, restricting medium-risk commodities from participating in promotional activities of the e-commerce platform, and requiring submission of a product qualification re-inspection report within a preset time; for low-risk commodities, sending a risk warning to the background, prompting the low-risk commodities to make independent rectifications, and the commodities after rectification are qualified after re-inspection can lift the warning status; the risk level classification results are associated with the risk response strategy execution records and stored.

5. The method for intelligently responding to platform risk events based on artificial intelligence according to claim 1, characterized in that: Step S4 includes: Step S4-1: Construct a multi-dimensional evaluation index system for response effects, including risk mitigation rate, user satisfaction improvement rate, and timeliness compliance rate; The risk mitigation rate is calculated by counting the number of new complaint orders and new return orders for the target product within the preset monitoring period after the risk response strategy is implemented. ； Among them, E1 represents the risk mitigation rate, Q 前 represents the number of complaint orders and return orders in the same monitoring period before the risk response strategy is implemented, Q 后 Indicates the number of complaint orders and return orders within the preset monitoring period after the risk response strategy is executed; User satisfaction improvement rate: The user evaluation data of the target product is collected through the user feedback system of the e-commerce platform, and the sentiment tendency features in the evaluation text are extracted to calculate the user satisfaction improvement rate: ； Among them, E2 represents the improvement rate of user satisfaction, S 前 represents the mean user satisfaction score before the risk response strategy is implemented, S 后 represents the mean user satisfaction score after the risk response strategy is implemented; The timeliness compliance rate is calculated by recording the time from receiving the response policy instruction to completing the rectification and passing the re-inspection of the e-commerce platform; ； Among them, T 实 ≤T 标 When T 实 >T 标 When E3=0, E3 represents the timeliness compliance rate, T 标 Indicates the standard rectification time of the target product, T 实 Indicates the actual rectification time; Step S4-2: Determine the weight coefficient of each evaluation indicator through the hierarchical analysis method and calculate the response effect coefficient: ； in, 、 、 These are the weight coefficients of risk mitigation rate, user satisfaction improvement rate, and timeliness compliance rate; Step S4-3: Establish a dynamic adjustment model for the risk response strategy and optimize the risk response strategy based on the response effect coefficient.

6. An artificial intelligence-based intelligent response system for platform risk events, configured to execute the artificial intelligence-based intelligent response method for platform risk events as described in any one of claims 1 to 5, characterized in that: The intelligent response system includes: information collection and difference analysis module, risk warning coefficient calculation module, risk classification and response module, effect evaluation and strategy adjustment module; The information collection and difference analysis module is used to collect product information published on the e-commerce platform and product review information in historical transactions, pre-process this information and calculate the degree of difference; The risk warning coefficient calculation module is used to screen out target products with excessive differences, track the information modification records of the target products and calculate the information modification rate, and calculate the risk warning coefficient based on the complaint rate and return rate in historical transactions; The risk grading and response module categorizes target commodities according to risk warning coefficients and degree of difference, and implements corresponding risk response strategies for commodities of different grades; The effect evaluation and strategy adjustment module performs a multi-dimensional evaluation on the execution effect of the risk response strategy, calculates the response effect coefficient, and dynamically adjusts the risk response strategy according to the response effect coefficient.

7. The artificial intelligence-based platform risk event intelligent response system according to claim 6, characterized in that: The information collection and difference analysis module includes: an information collection unit, a feature extraction unit, and a difference calculation unit; The information collection unit obtains product information published on the e-commerce platform and product review information in historical transaction records in real time through the data interface of the e-commerce platform; The feature extraction unit is used to extract text feature vectors from product text data, extract image feature vectors from product image data, extract dynamic feature vectors after processing the video frame sequence of product video data; and extract corresponding actual feature vectors from product review information; The difference calculation unit calculates the text difference degree, the image difference degree, and the video difference degree respectively, and calculates the overall difference degree between the product information and the product review information in combination with the weight coefficient.

8. The artificial intelligence-based platform risk event intelligent response system according to claim 6, characterized in that: The risk warning coefficient calculation module includes: a target commodity identification unit, a modification rate calculation unit, a complaint return rate calculation unit, and a warning coefficient unit; The target product identification unit divides the products into categories and establishes a difference degree benchmark library, calculates the difference degree mean and standard deviation of each category of products, generates a difference degree threshold, and marks products with a difference degree exceeding the threshold as target products; The modification rate calculation unit is used to monitor the information modification records of the target product within the target time period, extract various modification data and generate characteristic variables, and calculate the information modification rate in combination with the time weight; The complaint return rate calculation unit is used to collect the total number of completed orders, complaint orders, and return orders in historical transactions of the target product, and calculate the complaint rate and return rate respectively; The warning coefficient unit calculates the risk warning coefficient based on the information modification rate, complaint rate, return rate, and corresponding weight coefficients.

9. The artificial intelligence-based platform risk event intelligent response system according to claim 6, characterized in that: The risk grading and response module includes: an interval division unit, a level determination unit, and a strategy execution unit; The interval division unit extracts historical product information belonging to the same category as the target product, calculates the risk warning coefficient and difference degree of the historical product information, and divides the risk warning coefficient interval and the difference degree interval according to the set critical value; The level determination unit establishes a risk level classification matrix using the risk warning coefficient and the degree of difference as two-dimensional indicators, and determines the risk level of the target product by cross-combining the horizontal risk warning coefficient and the vertical degree of difference; The policy execution unit matches and executes dynamic response rules in a preset risk response policy library for commodities of different risk levels, and simultaneously associates and stores risk level classification results with response policy execution records.

10. The artificial intelligence-based platform risk event intelligent response system according to claim 6, characterized in that: The effect evaluation and strategy adjustment module includes: an evaluation index unit, an effect coefficient unit, and a strategy adjustment unit; The evaluation indicator unit constructs a multi-dimensional evaluation indicator system for response effects including risk mitigation rate, user satisfaction improvement rate, and timeliness compliance rate; The effect coefficient unit determines the weight coefficient of each evaluation index through the hierarchical analysis method, and calculates the response effect coefficient based on the numerical value of each index; The strategy adjustment unit establishes a dynamic adjustment model for the risk response strategy and optimizes and adjusts the risk response strategy according to the response effect coefficient.

Images