Commodity anomaly detection method and device and electronic equipment

Abstract

The embodiment of the invention provides a commodity anomaly detection method and device and electronic equipment. The commodity anomaly detection method comprises the steps of obtaining a to-be-detected commodity, wherein the to-be-detected commodity is in an abnormal state in a current detection period; determining the number of detection periods in which the to-be-detected commodity is continuously in an abnormal state and a dynamic threshold value corresponding to the to-be-detected commodity in the current detection period; and performing anomaly detection on the to-be-detected commodity based on the detection period number and the dynamic threshold. According to the embodiment of the invention, the to-be-detected commodities in the abnormal state in a short time can be effectively screened out, and the number of the concerned commodities is effectively reduced through the above operation because the commodities concerned by the operator are the commodities in the abnormal state in a long time. Therefore, repeated invalid communication between the platform operation and the merchant based on the concerned commodity is avoided, the operation efficiency of the platform is improved, and the operation management cost is reduced.

Description

Technical Field

[0001] The present application relates to the field of e-commerce technology, and in particular to a method, device, and electronic device for detecting anomalies in commodities. Background Art

[0002] In the e-commerce world, product anomalies (such as inventory anomalies and price anomalies) have a direct impact on product conversion rates. As platforms expand, the variety and quantity of products rapidly increase. To ensure real-time product anomaly detection, existing product anomaly detection systems typically use a high detection frequency, which often results in a large number of abnormal product results. When platform operators receive product anomaly alerts corresponding to abnormal product results, they often immediately notify the corresponding merchants to address the abnormal products.

[0003] However, if every time an abnormal product is detected, the platform operation is triggered to notify the merchant, when a large number of abnormal products need to be processed at the same time, by the time the platform operation notifies the merchant, the abnormal status of the product may have disappeared, that is, it has returned to normal. This will result in multiple ineffective communications between the platform operation and the merchant, thereby increasing the maintenance and management costs of the platform operation. Summary of the Invention

[0004] The embodiments of the present application provide a method, device, and electronic device for detecting anomalies in goods, which can, to a certain extent, avoid multiple ineffective communications between platform operators and merchants, thereby reducing the maintenance and management costs of platform operators.

[0005] An embodiment of the present invention provides a method for detecting anomalies in a commodity, including:

[0006] Acquire a commodity to be inspected, wherein the commodity to be inspected is in an abnormal state in a current inspection cycle;

[0007] Determine the number of detection cycles in which the product to be detected is in an abnormal state continuously and the dynamic threshold corresponding to the product to be detected in the current detection cycle;

[0008] Anomaly detection is performed on the commodity to be detected based on the number of detection cycles and the dynamic threshold.

[0009] An embodiment of the present invention provides a device for detecting anomalies in a commodity, including:

[0010] A first acquisition module is used to acquire a commodity to be detected, wherein the commodity to be detected is in an abnormal state in a current detection cycle;

[0011] A first determination module is configured to determine the number of detection cycles in which the commodity to be detected is continuously in an abnormal state and a dynamic threshold corresponding to the commodity to be detected in a current detection cycle;

[0012] The first processing module is configured to perform abnormality detection on the commodity to be detected based on the number of detection cycles and the dynamic threshold.

[0013] An embodiment of the present invention provides an electronic device, comprising: a memory and a processor; wherein the memory is used to store one or more computer instructions, wherein the one or more computer instructions, when executed by the processor, implement the method described in the first aspect above.

[0014] An embodiment of the present invention provides a computer storage medium for storing a computer program, wherein the computer program enables a computer to implement the method described in the first aspect when executed.

[0015] An embodiment of the present invention provides a computer program product, comprising: a computer-readable storage medium storing computer instructions, which, when executed by one or more processors, causes the one or more processors to execute the steps of the method described in the first aspect above.

[0016] The method, device and electronic device for detecting abnormalities of goods provided in this embodiment obtain the goods to be detected, determine the number of detection cycles in which the goods to be detected are continuously in an abnormal state and the dynamic threshold corresponding to the goods to be detected in the current detection cycle, wherein the goods to be detected are in an abnormal state in the current detection cycle, and then the abnormality detection operation can be performed again on the goods to be detected in the abnormal state based on the number of detection cycles in which the goods to be detected are continuously in an abnormal state and the dynamic threshold corresponding to the current detection cycle, so as to effectively identify the goods to be detected that are in an abnormal state for a short period of time and the goods to be detected that are in an abnormal state for a long period of time. Since the goods that the platform operation needs to pay attention to are often the goods to be detected that are in an abnormal state for a long period of time, a large number of goods to be detected that are in an abnormal state for a short period of time can be screened out through the above-mentioned identification operation, which not only effectively reduces the number of goods that the operation needs to pay attention to, and to a certain extent effectively avoids multiple invalid communications between the platform operation and merchants based on the short-term abnormal goods to be detected, but also reduces the platform's operation, maintenance and management costs, ensures the operation quality and effect, and further improves the practicality of the method. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0018] Figure 1 A schematic diagram of a scenario of a method for detecting anomalies in a commodity provided by an exemplary embodiment of the present application;

[0019] Figure 2 A flowchart of a method for detecting anomalies in a commodity provided by an exemplary embodiment of the present application;

[0020] Figure 3 A schematic diagram of a flow chart for determining a dynamic threshold corresponding to the commodity to be inspected in the current inspection cycle provided by an exemplary embodiment of the present application;

[0021] Figure 4 A schematic diagram of a process for obtaining a commodity to be inspected provided in an exemplary embodiment of the present application;

[0022] Figure 5 A schematic diagram of the principle of performing anomaly detection on candidate products in a candidate product set using detection rules and product information of the candidate products provided in an exemplary embodiment of the present application;

[0023] Figure 6 A flowchart of another method for detecting anomalies in a commodity provided by an exemplary embodiment of the present application;

[0024] Figure 7 A signaling interaction diagram of a method for detecting anomalies in a commodity provided in an exemplary application embodiment of the present application;

[0025] Figure 8 A schematic diagram of the structure of a commodity anomaly detection system provided in an exemplary application embodiment of the present application;

[0026] Figure 9 A schematic structural diagram of a device for detecting anomalies in a commodity provided by an exemplary embodiment of the present application;

[0027] Figure 10 A schematic structural diagram of an electronic device provided as an exemplary embodiment of the present application. DETAILED DESCRIPTION

[0028] To make the purpose, technical solutions, and advantages of this application more clear, the technical solutions of this application will be clearly and completely described below in conjunction with the specific embodiments of this application and the corresponding drawings. Obviously, the embodiments described are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0029] It should be noted that, in the case of user information involved in the embodiments of the present application, the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in the embodiments of the present application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data need to comply with the relevant laws, regulations and standards of the relevant countries and regions, and provide corresponding operation entrances for users to choose to authorize or refuse. In addition, the various models involved in this application (including but not limited to language models or large models) are in compliance with relevant laws and standards.

[0030] In addition, it should be noted that when the embodiments of the present application involve user interaction operations or triggering operations, the user interaction operations or triggering operations involved in the embodiments of the present application include but are not limited to: touch operations, gesture operations, voice operations, head movement operations, eye movement operations and other interactive operations in various ways; among which, touch operations include but are not limited to: click operations, double-click operations, long press operations, sliding operations, pinch operations or mouse hover operations, etc. Sliding operations include but are not limited to: straight sliding, curved sliding, etc.

[0031] Definition of terms:

[0032] LLM: Large Language Model, a large-scale natural language processing model based on deep learning.

[0033] Product anomaly detection: This refers to identifying and handling abnormal status of products on e-commerce platforms, such as out-of-stock, price fluctuations, expired promotions, etc.

[0034] Conversion efficiency: The efficiency of the entire process from product listing to final sales during the operation of an e-commerce platform.

[0035] Poisson Distribution: Poisson Distribution is a statistical probability distribution that describes the number of times an event occurs within a fixed time period. Poisson Distribution is often used to model the occurrence of sparse events, such as the number of events in anomaly detection.

[0036] To facilitate understanding of the product anomaly detection method, device, and electronic device provided in the embodiments of the present application, the following briefly describes the relevant technologies:

[0037] In e-commerce applications, product anomalies have a direct impact on user experience and sales conversion efficiency. As the platform scales up and the variety and quantity of products increase rapidly, systems that detect product anomalies face significant challenges:

[0038] (1) The contradiction between real-time performance and massive data

[0039] To ensure real-time performance, existing product anomaly detection systems usually adopt a higher detection frequency, which often results in a large number of abnormal product results.

[0040] However, because merchants typically perform immediate fixes for product anomalies, many anomalies are quickly resolved, rendering the anomaly status obsolete. This makes it difficult for operations teams, faced with processing large amounts of anomaly data, to accurately identify the truly abnormal products requiring attention. This not only increases workload, reduces the efficiency and accuracy of anomaly detection operations, but also fails to provide effective support to the operations team.

[0041] (2) Challenges of large-scale commodity and multi-dimensional testing

[0042] When there are many dimensions to detect at the commodity level, it is difficult for the operations team to have an accurate perception of the overall commodity situation, which will affect the comprehensiveness and accuracy of operational decisions.

[0043] In order to solve the above technical problems, the embodiments of the present application provide a method, device and electronic device for detecting abnormalities of commodities. Figure 1 As shown, the execution subject of the product anomaly detection method can be the product anomaly detection device 200, and the product anomaly detection device 200 can be implemented as a local server or a cloud server. In particular, when the product anomaly detection device 200 is implemented as a cloud server, the product anomaly detection method can be executed in the cloud, and several computing nodes (cloud servers) can be deployed in the cloud, each computing node having computing, storage and other processing resources. In the cloud, multiple computing nodes can be organized to provide a certain service. Of course, a computing node can also provide one or more services. The way to provide the service in the cloud can be to provide a service interface to the outside world, and the user calls the service interface to use the corresponding service. The service interface includes a software development kit (SDK), an application programming interface (API), and the like.

[0044] The product anomaly detection device 200 is in communication with a client 100. The client 100 is used by users to trigger product anomaly detection. The client 100 can be any computing device with certain information exchange capabilities. In specific implementations, the client 100 can be a mobile phone, a personal computer (PC), a tablet computer, a configuration application, and the like. Furthermore, the basic structure of the client 100 may include at least one processor. The number of processors depends on the configuration and type of the client. The client 100 may also include memory. This memory may be volatile, such as random access memory (RAM), non-volatile, such as read-only memory (ROM), flash memory, or both. The memory typically stores an operating system (OS), one or more application programs, and may also store program data. In addition to the processing unit and memory, the client 100 also includes some basic configurations, such as a network card chip, an I / O bus, a display component, and some peripheral devices. Optionally, some peripheral devices may include, for example, a keyboard, a mouse, an input pen, a printer, etc. Other peripheral devices are well known in the art and will not be described in detail here.

[0045] The product anomaly detection device 200 refers to a device that can perform product anomaly detection operations in a network virtual environment, typically referring to a device that utilizes a network to perform information planning and product anomaly detection operations. The product anomaly detection device 200 can be implemented as a product anomaly detection model for performing product anomaly detection operations. Physically, the product anomaly detection device 200 can be any device capable of providing computing services and performing corresponding product anomaly detection operations, such as a processor, server, etc. The product anomaly detection device 200 primarily comprises a processor, hard disk, memory, system bus, and other components, similar to a general-purpose computer architecture.

[0046] In the above embodiment, a network connection is established between the abnormality detection device 200 of the product and the client 100, and the network connection can be a wireless or wired network connection. If the abnormality detection device 200 of the product and the client 100 are connected by communication, the network standard of the mobile network can be any one of 2G (Global System for Mobile Communications GSM), 2.5G (General Packet Radio Service GPRS), 3G (Wideband Code Division Multiple Access (WCDMA), Time Division Synchronous Code Division Multiple Access (TD-SCDMA), 4G (Long Term Evolution LTE), 4G+ (Enhanced Long Term Evolution LTE+), Worldwide Interoperability for Microwave Access (WiMax), 5G, 6G, etc.

[0047] In an embodiment of the present application, the client 100 is used for users to generate or obtain an anomaly detection request for a commodity to be detected, wherein the number of commodities to be detected corresponding to the anomaly detection request may be one or more. When the number of commodities to be detected is multiple, multiple commodities to be detected may constitute a set of commodities to be detected. In some instances, the anomaly detection request may be implemented through human-computer interaction. Specifically, an interactive interface for interactive operation with the commodity anomaly detection device 200 may be displayed on the client 100. The user may input an execution operation in the interactive interface. The execution operation may be a human-computer interaction operation or a natural language voice interaction operation, thereby generating and obtaining an anomaly detection request for the commodity to be detected. In order to perform an anomaly detection operation on the commodity to be detected, the client 100 may send an anomaly detection request to the commodity anomaly detection device 200.

[0048] The abnormality detection device 200 of the commodity is used to obtain the abnormality detection request sent by the client 100, and then determine the commodity to be detected corresponding to the abnormality detection request, wherein the commodity to be detected is in an abnormal state in the current detection cycle, that is, it indicates that an abnormality detection operation has been performed on the above-mentioned commodity to be detected in the current detection cycle, and the result of the detection operation is that the commodity to be detected is in an abnormal state. Since there are a large number of commodities in an abnormal state (i.e., commodities that are in an abnormal state for a short period of time) in a single detection cycle, in order to avoid multiple invalid communications between the platform operation and the merchant, it is necessary to perform an abnormality detection operation again on the commodity to be detected that is in an abnormal state in the current detection cycle. At this time, the number of detection cycles in which the commodity to be detected is in an abnormal state continuously and the dynamic threshold corresponding to the commodity to be detected in the current detection cycle can be determined first, wherein the dynamic threshold corresponds to the detection cycle, that is, the dynamic threshold can change with the change of the detection cycle, and the dynamic threshold corresponding to different detection cycles can be different. In this way, when the abnormality detection of the commodity to be detected is performed based on the number of detection cycles and the dynamic threshold, the accuracy and reliability of the abnormality detection operation for the commodity to be detected can be effectively improved.

[0049] In this embodiment, the abnormality detection operation can be performed again on the products to be detected that are in an abnormal state based on the number of detection cycles in which the products to be detected are in an abnormal state continuously and the dynamic threshold corresponding to the current detection cycle, thereby effectively identifying and screening out the products to be detected that are in an abnormal state for a short period of time and the products to be detected that are in an abnormal state for a long period of time. Since the products that are in an abnormal state for a long period of time are products that need attention, the operation management and maintenance operations of the products that need attention can be performed based on the abnormality detection results. This not only effectively reduces the multiple ineffective communications between the platform operation and merchants based on the products to be detected that are in an abnormal state for a short period of time, but also reduces the operation, maintenance and management costs of the platform, and further improves the practicality of the method.

[0050] The following describes in detail the technical solutions provided by various embodiments of the present application in conjunction with the accompanying drawings.

[0051] Figure 2 A flowchart of a method for detecting anomalies of a commodity provided by an exemplary embodiment of the present application; Figure 2 As shown, this embodiment provides a method for detecting anomalies in a product. The execution subject of this method is a device for detecting anomalies in a product. The device for detecting anomalies in a product can be implemented as software or a combination of software and hardware. When the device for detecting anomalies in a product is implemented as hardware, it can be various electronic devices that can perform anomaly detection operations on the product, including but not limited to personal computers, servers, etc. When the device for detecting anomalies in a product is implemented as software, it can be installed in the electronic devices listed above. Specifically, the method for detecting anomalies in a product provided in this embodiment can include:

[0052] Step S201: Acquire a commodity to be inspected, wherein the commodity to be inspected is in an abnormal state in a current inspection cycle.

[0053] Step S202: Determine the number of detection cycles in which the product to be detected is continuously in an abnormal state and the dynamic threshold corresponding to the product to be detected in the current detection cycle.

[0054] Step S203: performing anomaly detection on the commodity to be detected based on the number of detection cycles and the dynamic threshold.

[0055] The following is a detailed description of the specific implementation methods and principles of each of the above steps:

[0056] Step S201: Acquire a commodity to be inspected, wherein the commodity to be inspected is in an abnormal state in a current inspection cycle.

[0057] Among them, the goods to be inspected may refer to goods that require abnormal detection operations. The above-mentioned abnormal operations can be implemented as periodic detection operations according to a preset detection cycle. It should be noted that the goods to be inspected are not ordinary goods that require abnormal detection operations, but goods that are in an abnormal state in the current detection cycle. The abnormal state may include at least one of the following: the goods to be inspected are out of stock in the current detection cycle, the price of the goods to be inspected in the current detection cycle is in a fluctuating state, the goods to be inspected have an activity expiration state in the current detection cycle, etc.

[0058] For example, for the same product, when the number of detection cycles includes 3, including detection cycle T1, detection cycle T2, and detection cycle T3, when detection cycle T3 is the current detection cycle, the historical detection cycle includes detection cycle T1 and detection cycle T2. When the detection status of the product in detection cycle T2 is abnormal, and the detection status in detection cycle T3 is normal, then the above-mentioned product will not become the product to be detected in the embodiment of the present application. Correspondingly, when the detection status of the product in detection cycle T2 is normal, and the detection status in detection cycle T3 is abnormal, then the above-mentioned product can become the product to be detected in the embodiment of the present application.

[0059] In some instances, the goods to be inspected can be obtained through a client. In this case, obtaining the goods to be inspected may include: determining a client that is in communication with the abnormality detection device of the goods, where the client includes the goods to be inspected that are in an abnormal state in the current detection cycle; based on the client actively or passively obtaining the goods to be inspected, this ensures to a certain extent the accuracy and reliability of obtaining the goods to be inspected.

[0060] In other instances, the products to be inspected can not only be obtained through the client, but can also be determined by analyzing and processing the product status of the products in the current inspection cycle. At this time, obtaining the products to be inspected may include: obtaining multiple alternative products; determining the preliminary inspection status corresponding to multiple alternative products in the current inspection cycle; when the preliminary inspection status of the alternative products is an abnormal state, the alternative products are determined as the products to be inspected; when the preliminary inspection status of the alternative products is a normal state, the alternative products are determined as not being the products to be inspected. This also ensures the accuracy and reliability of obtaining the products to be inspected.

[0061] Step S202: Determine the number of detection cycles in which the product to be detected is continuously in an abnormal state and the dynamic threshold corresponding to the product to be detected in the current detection cycle.

[0062] Since the number of products in abnormal status in a single detection cycle is often large when performing preliminary anomaly detection operations on the massive number of products on the e-commerce platform, in order to avoid multiple invalid communications between the platform operation and merchants based on products in abnormal status, after obtaining the products to be detected, the products to be detected can be subjected to another anomaly detection operation to identify whether the products to be detected are products that the operation personnel or the operation platform really need to pay attention to.

[0063] In order to be able to perform another abnormality detection operation on the product to be detected, after obtaining the product to be detected, the number of detection cycles in which the product to be detected is continuously in an abnormal state can be determined first, wherein the number of detection cycles can be determined by the historical detection results of the product to be detected in the historical detection cycles and the current detection results in the current detection cycle. At this time, determining the number of detection cycles in which the product to be detected is continuously in an abnormal state can include: obtaining the historical detection results corresponding to the product to be detected in the historical detection cycles and the current detection results corresponding to the current detection cycle; based on the historical detection results and the current detection results, determining the number of detection cycles in which the product to be detected is continuously in an abnormal state.

[0064] Specifically, for the product to be inspected, current relevant data for the current inspection cycle and historical relevant data for previous inspection cycles can be obtained. Based on the current relevant data and historical relevant data, anomaly detection operations can then be performed on the product to be inspected in the previous inspection cycle and the product to be inspected in the current inspection cycle, respectively. In some instances, the anomaly detection operation can be implemented using a state detection model used to implement anomaly detection operations for the product, thereby obtaining historical and current detection results.

[0065] Then, the historical detection results and the current detection results can be analyzed and processed to obtain the number of detection cycles in which the product to be detected is continuously in an abnormal state. Specifically, when the current detection result is an abnormal detection result, it can be identified whether the historical detection result adjacent to the current abnormal detection result is an abnormal detection result. If so, it is possible to continue to count whether the historical detection results adjacent to the historical abnormal detection result are abnormal detection results. If so, the above operation is performed until it is counted that the historical detection result is not an abnormal detection result. Then, the number of detection cycles in which the product to be detected is continuously in an abnormal state can be determined based on the current abnormal detection result and the number of continuous historical abnormal detection results.

[0066] Among them, the state detection model involved in the embodiment of the present application can be a language model (Language Mode, abbreviated as LM) or a multimodal model (Multimodal Model, abbreviated as MM) based on artificial intelligence, etc. The embodiment of the present application does not limit the number of model parameters supported by the model, with the goal of meeting actual needs.

[0067] For example, if the current detection result corresponding to the current detection cycle is abnormal, the historical detection result corresponding to the historical detection cycle can be queried from the current detection result corresponding to the current detection cycle to see if it is abnormal. If the historical detection result is abnormal, the number of historical detection cycles that are continuously in abnormal state can be counted. When the number of continuous historical detection cycles in abnormal state is 2, since the current detection result is abnormal, it can be determined that the number of detection cycles in which the product to be detected is continuously in abnormal state is 3. When the historical detection result corresponding to the historical detection cycle adjacent to the current detection cycle is normal, it can be determined that the number of detection cycles in which the product to be detected is continuously in abnormal state is 1. This effectively ensures the accuracy and reliability of determining the number of detection cycles in which the product to be detected is continuously in abnormal state.

[0068] Since the re-abnormal detection operation for the product to be detected requires not only determining the number of detection cycles in which the product to be detected is in an abnormal state continuously, but also determining the dynamic threshold corresponding to the product to be detected in the current detection cycle, in some instances, the dynamic threshold can be determined by a pre-configured mapping relationship corresponding to the detection cycle. At this time, determining the dynamic threshold corresponding to the product to be detected in the current detection cycle may include: obtaining the mapping relationship between each pre-configured detection cycle and each dynamic threshold; determining the dynamic threshold corresponding to the current detection cycle based on the mapping relationship and the identity identifier of the current detection cycle, which improves the accuracy and reliability of determining the dynamic threshold to a certain extent.

[0069] Step S203: performing anomaly detection on the commodity to be detected based on the number of detection cycles and the dynamic threshold.

[0070] After obtaining the number of detection cycles in which the product to be detected is in an abnormal state continuously and the dynamic threshold corresponding to the product to be detected in the current detection cycle, an abnormality detection operation can be performed on the product to be detected based on the number of detection cycles and the dynamic threshold. This can effectively identify whether the product to be detected is a product to be detected that is in an abnormal state for a short period of time.

[0071] Furthermore, when there are multiple products to be detected, an abnormality detection operation can be performed on each product to be detected to identify whether any product to be detected is a product to be detected that is in an abnormal state for a short period of time. In this way, for multiple products to be detected, the products to be detected that are in an abnormal state for a short period of time and the products to be detected that are in an abnormal state for a long period of time can be effectively screened out from the multiple products to be detected.

[0072] In some instances, anomaly detection can be performed using a pre-trained anomaly detection model. In this case, performing anomaly detection on a product to be detected based on the number of detection cycles and a dynamic threshold may include: obtaining the pre-trained anomaly detection model; inputting the number of detection cycles, the dynamic threshold, and the product data corresponding to the product to be detected into the anomaly detection model, performing an anomaly detection operation, and obtaining an anomaly detection result. The anomaly detection result may include a first result identifying the product to be detected as abnormal, or a second result identifying the product to be detected as normal.

[0073] In other instances, the anomaly detection operation can be achieved not only by analyzing and processing the number of detection cycles, dynamic thresholds, and product data corresponding to the product to be detected through an anomaly detection model, but also by comparing the number of detection cycles with the dynamic threshold. At this time, anomaly detection of the product to be detected based on the number of detection cycles and the dynamic threshold may include: when the number of detection cycles is greater than or equal to the dynamic threshold, determining that the product to be detected is an abnormal product; or, when the number of detection cycles is less than the dynamic threshold, determining that the product to be detected is a non-abnormal product.

[0074] Among them, since the dynamic threshold can be used to identify the lower limit value of the detection cycle for the product to be detected as being in an abnormal state, after obtaining the number of detection cycles and the dynamic threshold, the number of detection cycles can be analyzed and compared with the dynamic threshold. When the number of detection cycles is greater than or equal to the dynamic threshold, it means that the current product to be detected has met the detection conditions for being in an abnormal state, and the product to be detected can be determined as a product in an abnormal state; when the number of detection cycles is less than the dynamic threshold, it means that the current product to be detected does not meet the detection conditions for being in an abnormal state, and the product to be detected can be determined as a product in a non-abnormal state. In this way, the abnormal detection operation of the product to be detected is also accurately realized.

[0075] In addition, after determining that the product to be detected is an abnormal product, abnormal prompt information can be generated based on the abnormal detection results obtained by the above operations. At this time, the method in this embodiment can also include: generating a detection result and abnormal prompt information for identifying the product to be detected as an abnormal product; outputting the detection result and abnormal prompt information.

[0076] Specifically, after determining that the product to be detected is an abnormal product, a detection result can be generated to identify the product to be detected as an abnormal product, and abnormal prompt information corresponding to the detection result can be generated, wherein the above-mentioned detection result and abnormal prompt information can be generated by a pre-trained detection and analysis model.

[0077] After obtaining the test results and abnormal prompt information, the test results and abnormal prompt information can be associated and output. Specifically, the test results and abnormal prompt information can be output and displayed through a display device, or the test results and abnormal prompt information can be sent to the operator in the form of communication messages (for example, email, instant messaging software). In this way, the operator can quickly understand the abnormal status of the product to be tested through the output test results and abnormal prompt information, and can flexibly adjust and configure the operation operations of the product to be tested, thereby improving the quality and effect of the operation operations for the product to be tested to a certain extent, and at the same time improving the product conversion efficiency of the product to be tested.

[0078] Furthermore, after performing abnormality detection on the goods to be detected based on the number of detection cycles and dynamic thresholds, operational suggestion information corresponding to the goods to be detected can also be generated to provide operational personnel with adjustment directions for operational operations, which is conducive to improving the operational quality and effect of the goods to be detected. At this time, the method in this embodiment may include: generating operational suggestion information corresponding to the goods to be detected based on the abnormality detection results obtained through the abnormality detection operation; and associating and outputting the abnormality detection results and the operational suggestion information.

[0079] Specifically, after performing anomaly detection on the product to be detected based on the number of detection cycles and the dynamic threshold, the anomaly detection results obtained through the anomaly detection operation can be analyzed and processed to generate operational recommendation information for the product to be detected, wherein the above-mentioned operational recommendation information can be determined by a pre-trained product analysis model, that is, the anomaly detection results and the product to be detected are input into the product analysis model for analysis and processing, and the operational recommendation information corresponding to the product to be detected output by the product analysis model is obtained. In some instances, the operational recommendation information may include at least one of the following: content display suggestions corresponding to the product to be detected, advertising placement suggestions corresponding to the product to be detected, operational channel suggestions corresponding to the product to be detected, etc.

[0080] After obtaining the operation suggestion information, the abnormal detection results and the operation suggestion information can be correlated and outputted. Specifically, the abnormal detection results and the operation suggestion information can be output and displayed through a display device, or the abnormal detection results and the operation suggestion information can be sent to the operation personnel in the form of a communication message. In this way, the operation personnel can quickly understand the status of the goods to be inspected through the output abnormal detection results and operation suggestion information, and can flexibly adjust and configure the operation operations of the goods to be inspected, thereby improving the quality and effect of the operation operations for the goods to be inspected to a certain extent.

[0081] The method for detecting abnormalities of commodities provided in this embodiment obtains commodities to be detected, determines the number of detection cycles in which the commodities to be detected are continuously in an abnormal state and the dynamic threshold corresponding to the current detection cycle of the commodities to be detected, wherein the commodities to be detected are in an abnormal state in the current detection cycle, and then the abnormality detection operation can be performed again on the commodities to be detected in the abnormal state based on the number of detection cycles in which the commodities to be detected are continuously in an abnormal state and the dynamic threshold corresponding to the current detection cycle, thereby being able to effectively identify commodities to be detected that are in an abnormal state for a short period of time and commodities to be detected that are in an abnormal state for a long period of time. Since the commodities that the platform operation needs to pay attention to are often commodities to be detected that are in an abnormal state for a long period of time, a large number of commodities to be detected that are in an abnormal state for a short period of time can be screened out through the above-mentioned identification operation, which not only effectively reduces the number of commodities that the operation needs to pay attention to, and to a certain extent effectively avoids multiple invalid communications between the platform operation and merchants based on short-term abnormal commodities to be detected, but also reduces the operation, maintenance and management costs of the platform, ensures the operation quality and effect, and further improves the practicality of the method.

[0082] Figure 3 A flow chart of determining the dynamic threshold value corresponding to the commodity to be detected in the current detection cycle provided by an exemplary embodiment of the present application; based on the above embodiment, refer to the attached Figure 3 As shown, the dynamic threshold value can be determined not only by a pre-configured mapping relationship corresponding to the detection cycle, but also by analyzing and processing the detection product set to which the detection product belongs. In this case, determining the dynamic threshold value corresponding to the detection product in the current detection cycle may include:

[0083] Step S301: Obtain the detection commodity set to which the commodity to be detected belongs.

[0084] For the product to be detected, it can be implemented not only as a single product, but also as any one in a set of detected products. In this way, when performing abnormality detection operations on the product to be detected, it is possible not only to perform abnormality detection operations on a certain product to be detected, but also to perform abnormality detection operations on multiple products to be detected in a certain set of detected products.

[0085] When the product to be detected is implemented as any one of a set of detected products, since the dynamic threshold is related to the detection cycle and the set of detected products, in order to accurately determine the dynamic threshold corresponding to the product to be detected in the current detection cycle, the set of detected products to which the product to be detected belongs can be obtained first. In some instances, the set of detected products can be determined by a preset mapping relationship corresponding to the product to be detected. In this case, obtaining the set of detected products to which the product to be detected belongs can include: obtaining the product identifier of the product to be detected; and using the mapping relationship between the preset product identifier and the set identifier of the set of detected products and the product identifier to determine the set of detected products to which the product to be detected belongs. This effectively ensures the accuracy and reliability of obtaining the set of detected products.

[0086] Step S302: Determine the average abnormality occurrence rate corresponding to the detected product set within a preset time period, wherein the preset time period is greater than or equal to at least two detection cycles.

[0087] Since the dynamic threshold is used to perform anomaly detection operations on the products to be detected, it is often related to the average anomaly incidence rate corresponding to the set of detected products. Therefore, in order to accurately determine the dynamic threshold corresponding to the products to be detected in the current detection cycle, after obtaining the set of detected products to which the products to be detected belong, the average anomaly incidence rate corresponding to the set of detected products in the preset time period can be determined, where the preset time period includes multiple detection cycles.

[0088] In some instances, the average abnormality incidence rate can be determined by a pre-trained abnormality detection model. At this time, determining the average abnormality incidence rate corresponding to the detected product set within a preset time period may include: obtaining a pre-trained abnormality detection model; determining the detected product data corresponding to each to-be-detected product in the detected product set within a preset time period, wherein the detected product data may include at least one of the following: product identification of the to-be-detected product, product operation information of the to-be-detected product, product name of the to-be-detected product, brand information of the to-be-detected product, description information of the to-be-detected product, inventory and pricing information of the to-be-detected product, etc.; and then the detected product data corresponding to each to-be-detected product may be input into the abnormality detection model for processing to obtain the average abnormality incidence rate output by the abnormality detection model, which ensures the accuracy and reliability of the determination of the average abnormality incidence rate to a certain extent.

[0089] In other instances, the average anomaly incidence rate can be determined not only by a pre-trained anomaly detection model, but also based on the total number of anomalies corresponding to the detected products in the detected product set and a preset time period. At this time, determining the average anomaly incidence rate corresponding to the detected product set within the preset time period may include: obtaining the total number of anomalies corresponding to the detected products in the detected product set; determining the number of detection cycles included in the preset time period; and determining the average anomaly incidence rate based on the total number of anomalies and the number of detection cycles.

[0090] Among them, since a detection product set often includes multiple detection products, different detection products may correspond to different status information within a preset time period, and the average abnormality occurrence rate is related to the abnormal status corresponding to the detection products in the detection product set. Therefore, in order to accurately determine the average abnormality occurrence rate corresponding to the detection product set within the preset time period, the total number of abnormalities corresponding to the detection products in the detection product set can be obtained first. The total number of abnormalities can be obtained by summarizing the number of abnormalities corresponding to each detection product in the detection product set.

[0091] Since the average abnormality occurrence rate is not only related to the total number of abnormalities corresponding to the tested products in the tested product set, but also related to the preset time period, the preset time period can be analyzed and processed to determine the number of detection cycles included in the preset time period. In some examples, the number of detection cycles can be determined by the preset time period and the detection cycle. In this case, determining the number of detection cycles included in the preset time period may include: obtaining the cycle length corresponding to the detection cycle; dividing the preset time period based on the cycle length to obtain the number of detection cycles included in the preset time period. This ensures, to a certain extent, the accuracy and reliability of determining the number of detection cycles.

[0092] After obtaining the total number of anomalies and the number of detection cycles, the total number of anomalies and the number of detection cycles can be analyzed and processed to determine the average anomaly incidence rate. In some examples, based on the total number of anomalies and the number of detection cycles, determining the average anomaly incidence rate may include: determining the ratio of the total number of anomalies to the number of detection cycles as the average anomaly incidence rate. For example, when the total number of anomalies is Total_Anomalies and the number of detection cycles is Total_Periods, the average anomaly incidence rate λ can be determined to be This effectively ensures the accuracy and reliability of determining the average abnormality occurrence rate λ.

[0093] In other instances, the average abnormality incidence rate can be determined not only by the ratio of the total number of abnormalities to the number of detection cycles, but also by a pre-trained incidence calculation model. At this time, based on the total number of abnormalities and the number of detection cycles, determining the average abnormality incidence rate may include: obtaining an incidence determination model, which may be a pre-trained network model for calculating the average abnormality incidence rate, or a large language model that can calculate the average abnormality incidence rate; and then the total number of abnormalities and the number of detection cycles can be processed using the incidence determination model to determine the average abnormality incidence rate. Specifically, the total number of abnormalities and the number of detection cycles are input into the incidence determination model for analysis and processing, so as to obtain the average abnormality incidence rate output by the incidence determination model, which also ensures the accuracy and reliability of the determination of the average abnormality incidence rate.

[0094] Step S303: Determine a dynamic threshold based on the average abnormality occurrence rate and a preset confidence level.

[0095] After obtaining the average abnormality incidence rate, the average abnormality incidence rate and the preset reliability can be analyzed and processed to determine the dynamic threshold, wherein the preset reliability is a pre-configured parameter used to describe the reliability degree of the dynamic threshold, which ensures the accuracy and reliability of the determination of the dynamic threshold to a certain extent.

[0096] In some instances, the dynamic threshold can be determined by a pre-trained threshold calculation model. At this time, based on the average abnormality incidence rate and the preset confidence level, determining the dynamic threshold may include: obtaining a pre-trained threshold calculation model; inputting the average abnormality incidence rate and the preset execution degree into the threshold calculation model for analysis and processing, and obtaining the dynamic threshold output by the threshold calculation model. This ensures the accuracy and reliability of the determination of the dynamic threshold to a certain extent.

[0097] In other instances, the dynamic threshold can be determined not only by a pre-trained threshold calculation model, but also by the cumulative distribution function of the Poisson distribution. At this time, based on the average abnormality incidence rate and the preset confidence level, determining the dynamic threshold can include: obtaining the cumulative distribution function of the Poisson distribution; using the cumulative distribution function to process the average abnormality incidence rate and the preset confidence level to obtain the dynamic threshold output by the cumulative distribution function.

[0098] Among them, for the goods to be tested, without loss of generality, the following assumption can be made: the abnormal state of the goods to be tested at a certain moment is a random process, which is independent of the state at the previous moment. Therefore, its state at the next moment satisfies the Poisson distribution. At this time, the Poisson distribution model can be used to determine the dynamic threshold. Specifically, the cumulative distribution function of the Poisson distribution can be obtained first. The cumulative distribution function (CDF) gives the probability that the random variable (X) is less than or equal to a specific value (k). The cumulative distribution function can then be used to process the average abnormality incidence rate and the preset confidence level, so that the dynamic threshold output by the cumulative distribution function can be stably obtained.

[0099] Furthermore, the average abnormality occurrence rate and the preset confidence level are processed using the cumulative distribution function to obtain the dynamic threshold output by the cumulative distribution function, which may include: obtaining the probability mass function (PMF) of the Poisson distribution, which probability mass function (PMF) gives the probability that an event occurs exactly (k) times within a certain interval; and then using the probability mass function and the cumulative distribution function to process the average abnormality occurrence rate and the preset confidence level, so that the dynamic threshold can be stably obtained, which also ensures the accuracy and reliability of determining the dynamic threshold.

[0100] In this embodiment, by obtaining the set of detected products to which the product to be detected belongs, the average abnormality incidence rate corresponding to the set of detected products within a preset time period is determined, and then the dynamic threshold is determined based on the average abnormality incidence rate and the preset confidence level, thereby effectively realizing the flexible and stable determination operation of the dynamic threshold, thereby improving the flexibility and reliability of the abnormality detection operation on the product.

[0101] Figure 4 This is a flow chart of obtaining a commodity to be inspected provided by an exemplary embodiment of the present application; based on any of the above embodiments, refer to the attached Figure 4 As shown, the products to be tested can be determined not only by the client, but also by performing a preliminary test operation on a set of candidate products. In this case, obtaining the products to be tested may include:

[0102] Step S401: obtaining a set of candidate products, wherein the set of candidate products includes multiple candidate products, and the candidate products correspond to product information.

[0103] The product to be inspected can be any one of a plurality of candidate products. In order to accurately obtain the product to be inspected that requires re-anomaly detection, a set of candidate products can be first obtained, which includes multiple candidate products. In order to further improve the accuracy and reliability of obtaining the product to be inspected, the candidate products in the set of candidate products can correspond to product information, and the product information can include at least one of the following: product attributes (including at least one of the following: product name, product number, product brand, product category), product description information, product inventory and pricing, product image information, product logistics information, product transaction records, etc.

[0104] In some examples, the candidate product set may be obtained through a client, wherein the candidate product set may be stored in the client. In this case, the candidate product set may be obtained actively or passively through the client.

[0105] In other instances, the set of alternative products can be obtained not only through the client, but also based on preset product filtering information. In this case, obtaining the set of alternative products may include: obtaining the product filtering information; identifying the identity identification information corresponding to at least one alternative product that meets the product filtering information in a preset product database; performing a product information completion operation based on the identity identification information corresponding to at least one alternative product to obtain the set of alternative products.

[0106] Specifically, product screening information may refer to information or conditions used to perform preliminary screening operations on candidate products in a candidate product set. In some instances, the product screening information may be obtained through human-computer interaction or a client. The product screening information may include at least one of the following: product screening rules and product screening keywords. The aforementioned product screening rules may include at least one of the following: whether the product attributes are compatible with the distribution channel, and whether the penalty status is normal. The aforementioned penalty status may include at least one of the following: the penalty status of the product, the penalty status of the store where the product is located, etc.

[0107] After obtaining the product screening information, the identity identification information corresponding to at least one alternative product that meets the product screening information can be identified in a preset product database, wherein the preset product database can refer to a database including product data corresponding to a large number of products. In order to accurately obtain the set of alternative products, the obtained product screening information can be used to filter the products in the product database, so that the identity identification information corresponding to at least one alternative product can be obtained.

[0108] Since each alternative product in the set of alternative products has corresponding product information, in order to accurately obtain the set of alternative products, the product information can be supplemented based on the identity identification information corresponding to at least one alternative product. Specifically, information search can be performed in a preset database based on the identity identification corresponding to at least one alternative product to obtain product information, and then the product information supplement operation is completed. In this way, at least one alternative product corresponding to the product information can be obtained, and thus an alternative product set consisting of at least one alternative product and the product information corresponding to each alternative product can be obtained. This ensures the accuracy and reliability of obtaining the alternative product set to a certain extent.

[0109] Step S402: determining a detection rule for performing a preliminary detection operation on the candidate product set.

[0110] Since the candidate products in the candidate product set may be in a normal state or an abnormal state, and the products to be detected are products that are in an abnormal state during the current detection cycle, in order to accurately obtain the products to be detected, a detection rule for performing a preliminary detection operation on the candidate product set can be determined. In some examples, the detection rule may include at least one of the following: whether the product price is compatible with the distribution channel; whether the product corresponds to preset rights and interests information (for example, preset product promotion activities, preset product discount activities, etc.); whether there is inventory in a preset area; whether the inventory value of the product is greater than a preset threshold.

[0111] Specifically, the method for determining the detection rules in this embodiment is similar to the method for obtaining the product to be detected in the above embodiment. For details, please refer to the above statements and will not be repeated here.

[0112] Step S403: performing abnormality detection on the candidate products in the candidate product set using the detection rules and the product information of the candidate products to obtain detection results of the candidate products.

[0113] After obtaining the detection rules and the product information of the candidate products, an abnormality detection operation can be performed on the candidate products in the candidate product set using the detection rules and the product information of the candidate products, thereby obtaining a detection result for the candidate products. The detection result can be a first result that identifies the candidate products as normal; or the detection result can be a second result that identifies the candidate products as abnormal.

[0114] Among them, the detection rule may include multiple detection sub-rules. When using the detection rule and the product information of the candidate products to perform anomaly detection operations on the candidate products in the candidate product set, the abnormality detection operations can be performed on the candidate products in sequence according to each detection sub-rule and the data of the candidate products, so as to obtain the detection results of the candidate products. For example, refer to the attached Figure 5As shown, a detection sub-rule within a detection rule can be first retrieved, and data for a candidate product can be obtained. The engine can then be used to perform an anomaly detection operation based on the candidate product data and the detection sub-rule, obtaining a sub-detection result corresponding to the detection sub-rule. The sub-detection result is then identified to determine whether the candidate product has passed the anomaly detection operation. If it has not passed the anomaly detection operation, the candidate product can be determined to be in an abnormal state, and a corresponding anomaly record can be generated based on the candidate product. A corresponding anomaly detection result can then be generated based on the anomaly record.

[0115] Correspondingly, when determining that an alternative product has passed the anomaly detection operation corresponding to a detection sub-rule, it is possible to identify whether all detection rules have been tested. If the anomaly detection operation has not been performed on the alternative product using all detection rules, the next detection sub-rule can be pulled, and the next detection sub-rule and the data of the alternative product can be used to perform an anomaly detection operation on the alternative product to obtain the detection sub-result corresponding to the next detection sub-rule. When the anomaly detection operation has been performed on the alternative product using all detection rules, an anomaly detection result corresponding to the alternative product can be generated based on the detection sub-results corresponding to all detection sub-rules, thereby effectively ensuring the accuracy and reliability of the determination of the anomaly detection result.

[0116] It should be noted that for detection rules that include multiple detection sub-rules, not only can one engine be used to perform anomaly detection operations on alternative products based on one detection sub-rule in sequence for multiple detection sub-rules; or multiple engines can be used to synchronously use different multiple detection sub-rules to perform anomaly detection operations on alternative products. This ensures the quality and efficiency of the anomaly detection operation to a certain extent.

[0117] Step S404: when the detection result is used to identify the candidate commodity as being in an abnormal state, the candidate commodity corresponding to the detection result is determined as the commodity to be detected.

[0118] When the test result is used to identify the alternative product as being in a normal state, it means that the alternative product at this time is in a normal state in the current test cycle, and the alternative product corresponding to the test result can be determined as not being the product to be tested; when the test result is used to identify the alternative product as being in an abnormal state, it means that the alternative product at this time is in an abnormal state in the current test cycle, and the alternative product corresponding to the test result can be determined as the product to be tested.

[0119] In this embodiment, by obtaining a set of alternative products, a detection rule for performing a preliminary detection operation on the set of alternative products is determined, and then the detection rule and the product information of the alternative products are used to perform abnormality detection on the alternative products in the set of alternative products to obtain the detection results of the alternative products. In addition, when the detection result is used to identify the alternative product as being in an abnormal state, the alternative product corresponding to the detection result can be determined as the product to be detected, which effectively ensures the flexibility and reliability of obtaining the products to be detected.

[0120] Figure 6 A flowchart of another method for detecting anomalies of a commodity provided by an exemplary embodiment of the present application; based on any of the above embodiments, refer to the attached Figure 6 As shown, after performing anomaly detection on the detected products based on the number of detection cycles and the dynamic threshold, an abnormal product trend corresponding to the detected product set may be generated. In this case, the method in this embodiment may further include:

[0121] Step S601: Obtain a set of test products corresponding to the product to be tested.

[0122] In order to provide comprehensive perspective information and related status trends about the inspected products and help operators make more informed decisions in complex environments, after performing anomaly detection on the products to be inspected based on the number of detection cycles and dynamic thresholds, in order to accurately determine the abnormal product trends of the inspection product set corresponding to the products to be inspected, you can first obtain the inspection product set corresponding to the products to be inspected.

[0123] The method for obtaining the detected commodity set in this embodiment is similar to the method for obtaining the detected commodity set in the above embodiment. For details, please refer to the description in the above embodiment, which will not be repeated here.

[0124] Step S602: determining the current detection results of the detected commodity set in the current detection cycle and the historical detection results in the historical detection cycles.

[0125] After obtaining the set of detected products, the set of detected products can be analyzed and processed to determine the current detection results of the set of detected products in the current detection cycle and the historical detection results in the historical detection cycles. The current detection results can be obtained by performing an anomaly detection operation on the data of each detected product in the set of detected products in the current detection cycle. In this case, determining the current detection results of the set of detected products in the current detection cycle may include: obtaining a pre-trained anomaly detection model; using the anomaly detection model to perform an anomaly detection operation on each detected product in the set of detected products in the current detection cycle to obtain the product detection results corresponding to each detected product; and then statistically analyzing the product detection results corresponding to all detected products, thereby obtaining the current detection results of the set of detected products in the current detection cycle. This ensures, to a certain extent, the accuracy and reliability of determining the current detection results.

[0126] In addition, the specific method for determining the historical test results of the test product set in the historical test cycles is similar to the specific method for determining the current test results of the test product set in the current test cycle. For details, please refer to the above description and will not be repeated here. It should be noted that each of the above historical test cycles corresponds to a historical test result. When there are multiple historical test cycles, the number of historical test results obtained is also multiple.

[0127] Step S603: Based on the current detection result and the historical detection results, an abnormal commodity trend corresponding to the detected commodity set is generated.

[0128] After obtaining the current detection results and historical detection results, the current detection results and historical detection results can be analyzed and processed. Specifically, by analyzing and processing the current detection results and historical detection results, abnormal product trends corresponding to the detected product set can be generated. There can be multiple abnormal product trends, and different abnormal product trends can correspond to different parameter dimensions. The different parameter dimensions can include at least one of the following: product inventory, product abnormality rate, product price, etc.

[0129] In some instances, abnormal product trends can be generated by a pre-trained trend generation model. At this time, based on the current detection results and historical detection results, generating abnormal product trends corresponding to the detected product set may include: obtaining a pre-trained trend generation model; inputting the current detection results and historical detection results into the trend generation model to perform trend generation operations, and obtaining the abnormal product trends corresponding to the detected product set output by the trend generation model, thereby effectively ensuring the stable and reliable generation of abnormal product trends.

[0130] Furthermore, after generating the abnormal product trend corresponding to the detected product set, operation suggestion information corresponding to the detected product set can also be generated. At this time, the method in this embodiment can also include: generating operation suggestion information corresponding to the detected product set based on the abnormal product trend; and associating and outputting the abnormal product trend and the operation suggestion information.

[0131] Specifically, since different abnormal product trends can represent different states of the detected product set, in order to enable users to promptly and quickly obtain the state of the detected product set, after generating the abnormal product trends corresponding to the detected product set, the abnormal product trends can be analyzed and processed, thereby generating operational suggestion information corresponding to the detected product set. The operational suggestion information is often suggestion information related to product operational operations. For example, the operational suggestion information may include at least one of the following: suggestion information related to product inventory, suggestion information related to the number of products offline, suggestion information related to the number of products online, suggestion information related to product promotion activities, etc.

[0132] In some instances, operational recommendation information can be determined by a pre-trained trend analysis model. At this time, based on the abnormal product trend, generating operational recommendation information corresponding to the detected product set may include: obtaining a pre-trained trend analysis model; using the trend analysis model to analyze and process the abnormal product trend, and obtaining the operational recommendation information corresponding to the detected product set output by the trend analysis model. This ensures the accuracy and reliability of the generation of operational recommendation information to a certain extent.

[0133] In this embodiment, by obtaining the detection product set corresponding to the product to be detected, the current detection results of the detection product set in the current detection cycle and the historical detection results in the historical detection cycles are determined, and then the abnormal product trend corresponding to the detection product set is generated based on the current detection results and the historical detection results. This effectively ensures the accuracy and reliability of the generation of abnormal product trends.

[0134] For specific applications, refer to the attached Figure 7-Figure 8 As shown, this application embodiment provides a method for detecting anomalies in commodities. The execution subject of the anomaly detection method is an anomaly detection system. The anomaly detection system can use an abnormal commodity extraction model generated by a mathematical model to filter short-term and invalid abnormal commodities, and can combine a large language model to generate a predicted trend. Then, commodity operation operations can be performed based on the predicted trend. This can provide the operation team with more accurate management support operations, thereby improving the quality and efficiency of commodity operation operations to a certain extent. Specifically, the anomaly detection method implemented based on the anomaly detection system can include the following steps:

[0135] Step 1: Get the product to be tested.

[0136] Among them, the anomaly detection main link can obtain the goods to be detected through the inspected goods acquisition service. Specifically, the goods to be detected can be obtained according to the preset timed detection tasks. In addition, the inspected goods acquisition service can obtain the ID information of the goods to be detected from the goods library according to the preset goods screening rules and goods screening keywords. The above-mentioned timed detection task can trigger the acquisition operation of the goods to be detected once every hour, so that the anomaly detection operation of the goods can be performed periodically.

[0137] In some instances, product screening rules can be based on product attributes, specifically including at least one of the following: whether the image / title matches the distribution channel, and whether the penalty status is normal. For example, a screening rule could be whether the image / title matches the spring promotion channel. This screening rule can be used to select products related to the spring promotion channel.

[0138] Additionally, product screening keywords can include product feature information such as search tags, search keywords, or search activities entered by the user. This allows users to identify products to be tested that meet their needs. Furthermore, scheduled testing tasks can be implemented as offline tasks, filtering out products that meet the product screening criteria from the full data set of the e-commerce platform.

[0139] Step 2: Complete the product to be tested.

[0140] Since the product to be tested only has ID information, in order to accurately detect anomalies on the product, the product completion service can be used to complete the information in the external inventory / price system based on the ID information. Specifically, the necessary fields to be checked can be pulled from the ID information to generate a completion result corresponding to the product to be tested. The necessary fields to be checked can include at least one of the following: product description information, product main image, product inventory, product price, product discount price, etc.

[0141] Step 3: Perform abnormal product detection on the products to be detected after the completion operation to obtain preliminary abnormal product detection results.

[0142] Among them, the metadata storage service can store detection rules specified by the operator, and the detection rules may include at least one of the following: detection rules for the product attribute dimension, detection rules for the product price dimension, and detection rules for the product inventory dimension; the above-mentioned detection rules for the product attribute dimension may include at least one of the following: whether the picture / title matches the delivery channel, and whether the penalty status is normal; the detection rules for the product price dimension may include at least one of the following: whether the price matches the delivery channel, and whether specific rights can be used; the detection rules for the product inventory dimension may include at least one of the following: whether there is inventory in a specific area, and whether the inventory value is greater than a threshold.

[0143] In order to realize the detection operation of abnormal products, the rule detection service can be used to perform preliminary detection operations on the products to be detected after the completion operation. Specifically, the rule detection service can pull the detection rules from the metadata storage service, and then the rule detection service can use the avatar expression engine to perform preliminary detection operations on the products to be detected based on the pulled detection rules, thereby obtaining preliminary detection results of abnormal products.

[0144] Specifically, when using the expression engine to perform preliminary detection operations on the products to be detected, a product rule expression page can be provided first, so that the operator can configure the detection rules based on the product rule expression page. The above-mentioned detection rules can be rules for identifying whether the product to be detected is in a preset state of concern (for example: abnormal listing state, abnormal removal state, abnormal state of promotion activities, etc.).

[0145] Step 4: The anomaly detection main link can extract abnormal products that need to be processed based on the preliminary detection results of abnormal products.

[0146] Among them, the rule detection service can be used to extract abnormal products that need to be processed based on the preliminary detection results of abnormal products. Specifically, when the preliminary detection results are used to identify that the product to be detected is in an abnormal state in the current detection cycle, the product to be detected can be determined as an abnormal product that needs to be processed; when the preliminary detection results are used to identify that the product to be detected is in a normal state in the current detection cycle, the product to be detected can be determined as not an abnormal product that needs to be processed. This effectively achieves the accurate and reliable extraction of abnormal products that need to be processed.

[0147] Step 5: Perform abnormal reasoning on the abnormal products to be processed to obtain the target products and abnormal reasoning results.

[0148] Among them, the abnormal reasoning operation can be implemented by a pre-trained abnormal product extraction model. Specifically, the abnormal product extraction model can automatically extract the target products that operators really need to deal with. The target products can be products to be processed that have been in an abnormal state for a long time.

[0149] In some instances, the abnormal product extraction model can be a model implemented based on the Poisson distribution function. In order to improve the real-time performance of abnormal reasoning operations and help operators extract target products that are in a persistent abnormal state, the abnormal product extraction model can be used to perform abnormal reasoning operations on abnormal products that need to be processed. Specifically, without loss of generality, the following assumption can be made: the abnormal state of a product at a certain moment is a random process and is independent of the state at the previous moment. Therefore, its state at the next moment satisfies the Poisson distribution. Performing abnormal reasoning operations on abnormal products that need to be processed using the abnormal product extraction model can include the following steps:

[0150] Step 51: Calculate the total number of anomalies TotalAnomalies corresponding to the product set to which the abnormal product to be processed belongs and the total number of detection cycles TotalPeriods corresponding to the product data, and then use the total number of anomalies TotalAnomalies and the total number of detection cycles TotalPeriods to calculate the average anomaly incidence rate. Specifically, the average anomaly incidence rate = total number of anomalies / total number of detection cycles.

[0151] Step 52: Use the Poisson PPF (Poisson Percentile Function) to analyze and process the average abnormality rate and the preset execution level to calculate the dynamic threshold. The Poisson PPF (Poisson Percentile Function) may include the Poisson cumulative distribution function (CDF) and the Poisson probability mass function (PMF).

[0152] Step 53: For each abnormal product to be processed, determine whether the number of consecutive abnormalities in its abnormal state sequence reaches or exceeds the above-mentioned dynamic threshold. If so, mark it as a continuous abnormality, and then determine the above-mentioned abnormal product to be processed as a target product of attention.

[0153] Step 6: Export the abnormal reasoning results.

[0154] The result rendering service can be used to export the abnormal reasoning results to obtain the abnormal reasoning results corresponding to the target product, and the detection result storage service can be used to store the abnormal reasoning results. Specifically, the detection result notification service can be used to output the abnormal reasoning results and notify the operation personnel of the abnormal reasoning results. For example, the abnormal reasoning results can be sent to the operation personnel or merchant users via instant messaging programs, so that the operation personnel or merchant users can quickly locate the specific target product based on the abnormal reasoning results and further perform operations on the target product, thereby ensuring the operation quality and effectiveness of the target product.

[0155] Furthermore, in addition to obtaining the abnormal reasoning results corresponding to the target products, the abnormal reasoning results can also be rendered using the result rendering service. Specifically, a detection result report can be generated based on the large language model (LLM). The detection result report can include at least one of the following: a curve table, a bar chart, a pie chart, etc., and the detection result report can be output so that operators can view the detection result report. In addition, the indicator calculation service can be used to perform indicator calculation operations on the target products, and the product indicator data corresponding to each indicator can be obtained. The product indicator data can then be displayed to obtain the product indicator data corresponding to each indicator corresponding to the target products.

[0156] Furthermore, not only can the abnormal reasoning results corresponding to the target products be obtained, but also the product trends corresponding to the product set corresponding to the target products can be generated through LLM. Specifically, LLM can obtain prompt words for generating product trends based on the LLM prompt storage service through the data completion service, and then perform reasoning operations on the product data corresponding to the product set based on the prompt words, wherein the product data may include historical product data corresponding to the product set in the historical detection cycle and current product data corresponding to the product set in the current detection cycle, so that the product trend corresponding to the product set output by LLM can be obtained.

[0157] In addition, after obtaining the product trends corresponding to the product set, LLM can be used to analyze and process the product trends corresponding to the product set. Specifically, the current detection results corresponding to the product set in the current detection cycle, the statistical data of each dimension corresponding to the product set, the current detection reasons, the historical detection results of the historical detection cycles corresponding to the product set, and the historical detection reasons are obtained; then the current detection results corresponding to the product set in the current detection cycle, the statistical data of each dimension corresponding to the product set, the current detection reasons, the historical detection results of the historical detection cycles corresponding to the product set, and the historical detection reasons can be input into the LLM model for analysis and processing, and the operational operation suggestions corresponding to the product set output by LLM can be obtained. The operational operation suggestions can then be output and displayed so that operators and merchants can adjust the product operational operations based on the operational operation suggestions, which is conducive to improving the product operation effect in the product set.

[0158] The technical solution provided by this application embodiment implements a more advanced and intelligent anomaly detection system that can dynamically adapt to market changes and provide more accurate anomaly identification and management support. Specifically, it can achieve the following effects:

[0159] (1) Reduced false alarm rate. By adopting an anomaly detection model implemented by a mathematical model to filter short-term and invalid abnormal results, the determined dynamic threshold can be dynamically changed with the changes in product data and detection cycle, realizing a dynamic filtering mechanism, thereby reducing the false alarm rate of product detection, separating the anomalies that really need attention from irrelevant information, reducing the workload of the operation team, and improving work efficiency and decision-making accuracy. In addition, an intelligent abnormality feedback and processing operation is realized, specifically establishing a fully automated process from detection to feedback, and notifying the operation personnel in time to handle it, ensuring the efficient operation of the entire chain, significantly improving the product operation efficiency of the entire chain, and bringing higher competitiveness to the e-commerce platform, providing a more efficient and intelligent task management method, improving the operational efficiency and user experience of the entire chain, and providing more accurate management support for the operation team.

[0160] (2) Breaking through the limitations of dimension and scale, the system realizes a more efficient and intelligent product anomaly detection operation through advanced algorithms and LLM processing capabilities, and the detection mechanism combined with rules. The system can process large-scale products and multi-dimensional data at the same time, thereby providing more comprehensive anomaly detection results and improving the accuracy and flexibility of product anomaly detection operations. Based on the comprehensive anomaly detection results, the system can enhance the perception of the overall product value changes, ensure the comprehensiveness and accuracy of the detection results, help the operation team make more informed decisions, and further improve the practicality of the solution.

[0161] Figure 9This is a schematic diagram of a device for detecting abnormalities in a commodity provided by an exemplary embodiment of the present application; Figure 9 As shown, this embodiment provides a commodity abnormality detection device, which is used to perform the above Figure 2 The anomaly detection method shown, specifically, the anomaly detection device may include:

[0162] The first acquisition module 11 is used to acquire a commodity to be detected, wherein the commodity to be detected is in an abnormal state in a current detection cycle;

[0163] The first determination module 12 is used to determine the number of detection cycles in which the product to be detected is in an abnormal state continuously and the dynamic threshold corresponding to the product to be detected in the current detection cycle;

[0164] The first processing module 13 is configured to perform anomaly detection on the commodity to be detected based on the number of detection cycles and a dynamic threshold.

[0165] The abnormality detection device in this embodiment can also perform the above Figures 1-8 For the description of the embodiment shown, please refer to the detailed description of the above embodiment, and will not be elaborated here.

[0166] In addition, some of the processes described in the above embodiments and the accompanying drawings include multiple operations that appear in a specific order, but it should be clearly understood that these operations may not be executed in the order in which they appear in this article or may be executed in parallel. The sequence numbers of the operations, such as 11, 12, etc., are only used to distinguish between different operations, and the sequence numbers themselves do not represent any order of execution. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions of "first", "second", etc. in this article are used to distinguish different messages, devices, modules, etc., and do not represent a sequential order, nor do they limit "first" and "second" to being different types.

[0167] Figure 10 A schematic diagram of the structure of an electronic device provided by an exemplary embodiment of the present application; Figure 10 As shown, this embodiment provides an electronic device for performing the above Figure 2 The method for detecting abnormalities of commodities shown in the figure, wherein the electronic device may include: a memory 24 and a processor 25.

[0168] The memory 24 is used to store computer programs and can be configured to store various other data to support operations on the electronic device. Examples of such data include instructions for any application or method operating on the electronic device, data structures, contact data, phone book data, messages, pictures, videos, etc.

[0169] The processor 25 is coupled to the memory 24 and is used to execute the computer program in the memory 24 to: obtain the product to be detected, wherein the product to be detected is in an abnormal state in the current detection cycle; determine the number of detection cycles in which the product to be detected is in an abnormal state and the dynamic threshold corresponding to the product to be detected in the current detection cycle; and perform abnormality detection on the product to be detected based on the number of detection cycles and the dynamic threshold.

[0170] In some instances, when the processor 25 determines the dynamic threshold corresponding to the product to be inspected in the current inspection cycle, the processor 25 is used to execute: obtaining the inspection product set to which the product to be inspected belongs; determining the average abnormality incidence rate corresponding to the inspection product set within a preset time period, wherein the preset time period is greater than or equal to at least two inspection cycles; and determining the dynamic threshold based on the average abnormality incidence rate and a preset confidence level.

[0171] In some instances, when the processor 25 determines the average abnormality occurrence rate corresponding to the detected product set within a preset time period, the processor 25 is used to execute: obtaining the total number of abnormalities corresponding to the detected products in the detected product set; determining the number of detection cycles included in the preset time period; and determining the average abnormality occurrence rate based on the total number of abnormalities and the number of detection cycles.

[0172] In some examples, when the processor 25 determines the average abnormality occurrence rate based on the total number of abnormalities and the number of detection cycles, the processor 25 is configured to perform: determining the ratio of the total number of abnormalities to the number of detection cycles as the average abnormality occurrence rate.

[0173] In some instances, when the processor 25 determines the average abnormality occurrence rate based on the total number of abnormalities and the number of detection cycles, the processor 25 is used to execute: obtaining an occurrence rate determination model; using the occurrence rate determination model to process the total number of abnormalities and the number of detection cycles to determine the average abnormality occurrence rate.

[0174] In some instances, when the processor 25 determines the dynamic threshold based on the average abnormality incidence rate and the preset confidence level, the processor 25 is used to execute: obtaining the cumulative distribution function of the Poisson distribution; using the cumulative distribution function to process the average abnormality incidence rate and the preset confidence level to obtain the dynamic threshold value output by the cumulative distribution function.

[0175] In some instances, when the processor 25 performs abnormality detection on the product to be detected based on the number of detection cycles and the dynamic threshold, the processor 25 is used to execute: when the number of detection cycles is greater than or equal to the dynamic threshold, determining that the product to be detected is an abnormal product; when the number of detection cycles is less than the dynamic threshold, determining that the product to be detected is a non-abnormal product.

[0176] In some instances, after determining that the commodity to be detected is an abnormal commodity, the processor 25 in this embodiment is used to execute: generating a detection result and abnormal prompt information for identifying the commodity to be detected as an abnormal commodity; and outputting the detection result and abnormal prompt information.

[0177] In some instances, when the processor 25 obtains the product to be inspected, the processor 25 is used to execute: obtaining a set of alternative products, wherein the set of alternative products includes multiple alternative products, and the alternative products correspond to product information; determining a detection rule for performing a preliminary detection operation on the set of alternative products; performing anomaly detection on the alternative products in the set of alternative products using the detection rule and the product information of the alternative products to obtain a detection result of the alternative products; and when the detection result is used to identify the alternative product as being in an abnormal state, determining the alternative product corresponding to the detection result as the product to be inspected.

[0178] In some instances, when the processor 25 obtains a set of alternative products, the processor 25 is used to execute: obtaining product screening information; identifying the identity identification information corresponding to at least one alternative product that meets the product screening information in a preset product database; and performing a product information completion operation based on the identity identification information corresponding to at least one alternative product to obtain a set of alternative products.

[0179] In some instances, after performing anomaly detection on the product to be detected based on the number of detection cycles and the dynamic threshold, the processor 25 in this embodiment is used to execute: generating operation recommendation information corresponding to the product to be detected based on the anomaly detection results obtained through the anomaly detection operation; and associating and outputting the anomaly detection results and the operation recommendation information.

[0180] In some instances, after performing abnormality detection on the products to be detected based on the number of detection cycles and the dynamic threshold, the processor 25 in this embodiment is used to execute: obtaining a set of detected products corresponding to the products to be detected; determining the current detection results of the detected product set in the current detection cycle and the historical detection results in the historical detection cycles; and generating an abnormal product trend corresponding to the detected product set based on the current detection results and the historical detection results.

[0181] In some instances, after generating abnormal commodity trends corresponding to the detected commodity set, the processor 25 in this embodiment is used to execute: based on the abnormal commodity trends, generate operation suggestion information corresponding to the detected commodity set; and associate and output the abnormal commodity trends with the operation suggestion information.

[0182] Further, if Figure 10 As shown, the electronic device also includes: a communication component 26, a display 27, a power component 28, an audio component 29 and other components. Figure 10Only some components are shown schematically, which does not mean that the electronic device only includes Figure 10 In addition, Figure 10 The components in the center line frame are optional components, not mandatory components, and the specific configuration depends on the product form of the working node. The working node of this embodiment can be implemented as a terminal device such as a desktop computer, a laptop computer, a smart phone or an IOT device, or as a server-side device such as a conventional server, a cloud server or a server array. If the working node of this embodiment is implemented as a terminal device such as a desktop computer, a laptop computer, a smart phone, etc., it can include Figure 10 If the working node of this embodiment is implemented as a server device such as a conventional server, cloud server or server array, it may not include Figure 10 Components within the center wireframe.

[0183] The above-mentioned memory can be implemented by any type of volatile or non-volatile memory device or a combination thereof, such as static random-access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0184] The communication component is configured to facilitate wired or wireless communication between the device in which the communication component resides and other devices. The device in which the communication component resides can access a wireless network based on a communication standard, such as a 2G, 3G, 4G / LTE, 5G, or other mobile communication network, or a combination thereof. In an exemplary embodiment, the communication component receives broadcast signals or broadcast-related information from an external broadcast management system via a broadcast channel.

[0185] The above-mentioned display includes a screen, which may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen may be implemented as a touch screen to receive input signals from the user. The touch panel includes one or more touch sensors to sense touch, slide, and gestures on the touch panel. The touch sensor can not only sense the boundary of the touch or slide action, but also detect the duration and pressure associated with the touch or slide operation.

[0186] The power supply assembly provides power to various components of the device in which the power supply assembly is located. The power supply assembly may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power to the device in which the power supply assembly is located.

[0187] The above-mentioned audio component can be configured to output and / or input audio signals. For example, the audio component includes a microphone (MIC), and when the device where the audio component is located is in an operating mode, such as call mode, recording mode, and voice recognition mode, the microphone is configured to receive external audio signals. The received audio signal can be further stored in a memory or sent via a communication component. In some embodiments, the audio component also includes a speaker for outputting audio signals.

[0188] Accordingly, an embodiment of the present application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, enables the processor to implement the steps in the above method embodiment. The computer-readable storage medium includes volatile or non-volatile or a combination thereof, and may be removable or non-removable. Examples of computer-readable storage media include, but are not limited to, phase-change random access memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), flash memory or other memory technology, CD-ROM, digital versatile disc (DVD) or other optical storage, magnetic cassette, tape disk storage or other magnetic storage device or any other non-transmission medium.

[0189] Accordingly, the present application embodiment also provides a computer program product, which includes a computer program or instructions. When the computer program or instructions are executed by a processor, the processor is enabled to implement the steps in the above-mentioned method embodiment. It should be understood that each process or a combination of multiple processes in the above-mentioned method flow can be implemented by a computer program or instruction. In addition, these computer programs or instructions can be applied to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device, so that the processor of the general-purpose computer, the special-purpose computer, the embedded processor or other programmable data processing device can be implemented as a device for implementing the corresponding functions in the above-mentioned method embodiment.

[0190] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a series of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, commodity, or apparatus that includes the element.

[0191] The above are merely embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should all be included within the scope of the claims of the present application.

Claims

1. A method for detecting anomalies of commodities, characterized in that: include: Acquire a commodity to be inspected, wherein the commodity to be inspected is in an abnormal state in a current inspection cycle; Determine the number of detection cycles in which the product to be detected is in an abnormal state continuously and the dynamic threshold corresponding to the product to be detected in the current detection cycle; Anomaly detection is performed on the commodity to be detected based on the number of detection cycles and the dynamic threshold.

2. The method according to claim 1, characterized in that Determining the dynamic threshold corresponding to the product to be detected in the current detection cycle includes: Obtaining the test product set to which the product to be tested belongs; Determining an average abnormality occurrence rate corresponding to the set of detected products within a preset time period, wherein the preset time period is greater than or equal to at least two detection cycles; The dynamic threshold is determined based on the average abnormality occurrence rate and a preset confidence level.

3. The method according to claim 2, characterized in that Determining an average abnormality occurrence rate corresponding to the set of detected products within a preset time period includes: Obtaining the total number of abnormalities corresponding to the test products in the test product set; Determining the number of detection cycles included in the preset time period; The average abnormality occurrence rate is determined based on the total number of abnormalities and the number of detection cycles.

4. The method according to claim 3, characterized in that Determining the average abnormality occurrence rate based on the total number of abnormalities and the number of detection cycles includes: The ratio of the total number of abnormalities to the number of detection cycles is determined as the average abnormality occurrence rate.

5. The method according to claim 3, characterized in that Determining the average abnormality occurrence rate based on the total number of abnormalities and the number of detection cycles includes: Obtaining incidence determination models; The total number of abnormalities and the number of detection cycles are processed using the occurrence rate determination model to determine the average abnormality occurrence rate.

6. The method according to claim 2, characterized in that Determining the dynamic threshold based on the average abnormality occurrence rate and a preset confidence level includes: Get the cumulative distribution function of the Poisson distribution; The average abnormality occurrence rate and the preset confidence level are processed using the cumulative distribution function to obtain a dynamic threshold output by the cumulative distribution function.

7. The method according to any one of claims 1 to 6, characterized in that Performing abnormality detection on the commodity to be detected based on the number of detection cycles and the dynamic threshold includes: When the number of detection cycles is greater than or equal to the dynamic threshold, determining that the commodity to be detected is an abnormal commodity; When the number of detection cycles is less than the dynamic threshold, it is determined that the commodity to be detected is a non-abnormal commodity.

8. The method according to claim 7, characterized in that After determining that the commodity to be detected is an abnormal commodity, the method further includes: Generate a detection result and abnormal prompt information for identifying the commodity to be detected as an abnormal commodity; Output the detection results and abnormal prompt information.

9. The method according to any one of claims 1 to 6, characterized in that Get the products to be tested, including: Acquire a candidate product set, wherein the candidate product set includes multiple candidate products, and the candidate products correspond to product information; Determining a detection rule for performing a preliminary detection operation on the candidate product set; Performing anomaly detection on the candidate products in the candidate product set using the detection rule and the product information of the candidate products to obtain a detection result of the candidate products; In a case where the detection result is used to identify that the candidate commodity is in an abnormal state, the candidate commodity corresponding to the detection result is determined as the commodity to be detected.

10. The method according to claim 9, characterized in that Get a set of candidate products, including: Get product screening information; Identifying identity information corresponding to at least one candidate product that meets the product screening information in a preset product database; The product information is completed based on the identity identification information corresponding to each of the at least one candidate product to obtain the candidate product set.

11. The method according to any one of claims 1 to 6, characterized in that After performing abnormality detection on the commodity to be detected based on the number of detection cycles and the dynamic threshold, the method further includes: Based on the anomaly detection results obtained through the anomaly detection operation, generating operation suggestion information corresponding to the product to be detected; The anomaly detection result and the operation suggestion information are correlated and output.

12. The method according to any one of claims 1 to 6, characterized in that After performing abnormality detection on the commodity to be detected based on the number of detection cycles and the dynamic threshold, the method further includes: Obtaining a set of test products corresponding to the product to be tested; Determine the current test results of the test product set in the current test cycle and the historical test results in the historical test cycles; Based on the current detection result and the historical detection results, an abnormal commodity trend corresponding to the detected commodity set is generated.

13. The method according to claim 12, characterized in that After generating the abnormal commodity trend corresponding to the detected commodity set, the method further includes: Based on the abnormal commodity trend, generating operation suggestion information corresponding to the detected commodity set; The abnormal product trend is associated with the operation suggestion information and outputted.

14. A device for detecting abnormality of a commodity, characterized in that: include: A first acquisition module is used to acquire a commodity to be detected, wherein the commodity to be detected is in an abnormal state in a current detection cycle; A first determination module is configured to determine the number of detection cycles in which the commodity to be detected is continuously in an abnormal state and a dynamic threshold corresponding to the commodity to be detected in a current detection cycle; The first processing module is configured to perform abnormality detection on the commodity to be detected based on the number of detection cycles and the dynamic threshold.

15. An electronic device, characterized in that: include: A memory, a processor; wherein the memory is used to store one or more computer instructions, wherein the one or more computer instructions, when executed by the processor, implement the method of any one of claims 1 to 13.

16. A computer storage medium, characterized in that Used to store a computer program, which enables a computer to implement the method according to any one of claims 1 to 13 when executed.

17. A computer program product, characterized in that include: A computer program, when executed by a processor of an electronic device, causes the processor to perform the steps of the method according to any one of claims 1 to 13.

Images