Method and device for detecting unqualified products

Abstract

The purpose of this application is to provide a method and device for detecting unqualified products, the method comprising: obtaining connection relationship feature information corresponding to a knowledge graph, wherein the knowledge graph is constructed based on product feature information corresponding to multiple products and object feature information corresponding to associated objects associated with the product feature information, each node in the knowledge graph corresponds to a product or an associated object, and the connection relationship feature information is used to characterize the connection relationship between each node in the knowledge graph; obtaining a product detection model through training based on the product feature information, the object feature information, the connection relationship feature information and product calibration information corresponding to the multiple products, wherein the product calibration information is used to calibrate whether each of the multiple products is an unqualified product.

Description

Technical Field

[0001] The present application relates to the field of communications, and in particular to a technology for detecting substandard goods. Background Art

[0002] With the advancement of technology and the development of society, the internet has become increasingly integrated into our lives, and online shopping has become widely accepted. However, many counterfeit, substandard, and prohibited goods are found online, making it difficult for people to identify them. Existing technologies typically rely on manual labeling of counterfeit, substandard, and prohibited goods. However, this method is difficult to detect if product descriptions change frequently, and it is also extremely labor-intensive and time-consuming. Summary of the Invention

[0003] One purpose of the present application is to provide a method and device for detecting substandard goods.

[0004] According to one aspect of the present application, a method for detecting unqualified goods is provided, the method comprising:

[0005] Obtaining connection relationship feature information corresponding to a knowledge graph, wherein the knowledge graph is constructed based on product feature information corresponding to multiple products and object feature information corresponding to associated objects associated with the product feature information, the knowledge graph includes multiple nodes, each node in the knowledge graph corresponds to a product or an associated object, and the connection relationship feature information is used to characterize the connection relationship between each node in the knowledge graph;

[0006] Obtaining a product detection model through training based on the product feature information, the object feature information, the connection relationship feature information, and product calibration information corresponding to the multiple products, wherein the product calibration information is used to calibrate whether each of the multiple products is an unqualified product;

[0007] The target commodity feature information corresponding to the target commodity is input into the commodity detection model to obtain commodity detection information corresponding to the target commodity output by the commodity detection model, wherein the commodity detection information is used to indicate whether the target commodity is an unqualified commodity.

[0008] According to one aspect of the present application, a network device for detecting unqualified goods is provided, the device comprising:

[0009] A module for obtaining connection relationship feature information corresponding to a knowledge graph, wherein the knowledge graph is constructed based on product feature information corresponding to multiple products and object feature information corresponding to associated objects associated with the product feature information. The knowledge graph includes multiple nodes, each node in the knowledge graph corresponds to a product or an associated object, and the connection relationship feature information is used to represent the connection relationship between the nodes in the knowledge graph;

[0010] Module 12 is configured to obtain a product detection model through training based on the product feature information, the object feature information, the connection relationship feature information, and product calibration information corresponding to the multiple products, wherein the product calibration information is used to calibrate whether each of the multiple products is an unqualified product;

[0011] Module 13 is used to input the target product feature information corresponding to the target product into the product detection model, and obtain the product detection information corresponding to the target product output by the product detection model, wherein the product detection information is used to indicate whether the target product is an unqualified product.

[0012] According to one aspect of the present application, a device for detecting unqualified goods is provided, wherein the device comprises:

[0013] processor; and

[0014] a memory arranged to store computer-executable instructions which, when executed, cause the processor to:

[0015] Obtaining connection relationship feature information corresponding to a knowledge graph, wherein the knowledge graph is constructed based on product feature information corresponding to multiple products and object feature information corresponding to associated objects associated with the product feature information, the knowledge graph includes multiple nodes, each node in the knowledge graph corresponds to a product or an associated object, and the connection relationship feature information is used to characterize the connection relationship between each node in the knowledge graph;

[0016] Obtaining a product detection model through training based on the product feature information, the object feature information, the connection relationship feature information, and product calibration information corresponding to the multiple products, wherein the product calibration information is used to calibrate whether each of the multiple products is an unqualified product;

[0017] The target commodity feature information corresponding to the target commodity is input into the commodity detection model to obtain commodity detection information corresponding to the target commodity output by the commodity detection model, wherein the commodity detection information is used to indicate whether the target commodity is an unqualified commodity.

[0018] According to one aspect of the present application, a computer-readable medium storing instructions is provided, wherein when the instructions are executed, the system performs the following operations:

[0019] Obtaining connection relationship feature information corresponding to a knowledge graph, wherein the knowledge graph is constructed based on product feature information corresponding to multiple products and object feature information corresponding to associated objects associated with the product feature information, the knowledge graph includes multiple nodes, each node in the knowledge graph corresponds to a product or an associated object, and the connection relationship feature information is used to represent the connection relationship between each node in the knowledge graph;

[0020] Obtaining a product detection model through training based on the product feature information, the object feature information, the connection relationship feature information, and product calibration information corresponding to the multiple products, wherein the product calibration information is used to calibrate whether each of the multiple products is an unqualified product;

[0021] The target commodity feature information corresponding to the target commodity is input into the commodity detection model to obtain commodity detection information corresponding to the target commodity output by the commodity detection model, wherein the commodity detection information is used to indicate whether the target commodity is an unqualified commodity.

[0022] Compared with the existing technology, the present application can construct a knowledge graph based on the product feature information corresponding to multiple products and the object feature information corresponding to the associated objects associated with the product feature information, and then train a product detection model based on the product feature information, object feature information, connection relationship feature information corresponding to each node in the knowledge graph and product calibration information. Through this product detection model, it is possible to quickly and accurately detect whether the target product is an unqualified product. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Other features, objects and advantages of the present application will become more apparent upon reading the detailed description of non-limiting embodiments made with reference to the following drawings:

[0024] Figure 1 A flow chart of a method for detecting substandard goods according to one embodiment of the present application is shown;

[0025] Figure 2 A flow chart of a method for constructing a knowledge graph according to one embodiment of the present application is shown;

[0026] Figure 3 A flow chart of a method for obtaining a commodity detection model through training according to one embodiment of the present application is shown;

[0027] Figure 4A diagram showing the structure of a network device for detecting unqualified goods according to one embodiment of the present application is shown;

[0028] Figure 5 An exemplary system is shown that can be used to implement the various embodiments described in this application.

[0029] The same or similar reference numerals in the drawings represent the same or similar components. DETAILED DESCRIPTION

[0030] The present application is described in further detail below with reference to the accompanying drawings.

[0031] In a typical configuration of the present application, the terminal, the device of the service network and the trusted party all include one or more processors (eg, a central processing unit (CPU)), an input / output interface, a network interface and a memory.

[0032] Memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash memory. Memory is an example of a computer-readable medium.

[0033] Computer-readable media include permanent and non-permanent, removable and non-removable media that can be used to store information by any method or technology. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PCM), programmable 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), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information that can be accessed by a computing device.

[0034] The devices referred to in this application include but are not limited to user devices, network devices, or devices formed by integrating user devices and network devices through a network. The user devices include but are not limited to any mobile electronic product that can interact with a user (for example, through a touchpad), such as a smartphone, a tablet computer, etc. The mobile electronic product can use any operating system, such as the Android operating system, the iOS operating system, etc. Among them, the network device includes an electronic device that can automatically perform numerical calculations and information processing according to pre-set or stored instructions, and its hardware includes but is not limited to a microprocessor, an application specific integrated circuit (ASIC), a programmable logic device (PLD), a field programmable gate array (FPGA), a digital signal processor (DSP), an embedded device, etc. The network device includes but is not limited to a computer, a network host, a single network server, a set of multiple network servers, or a cloud composed of multiple servers; here, the cloud is composed of a large number of computers or network servers based on cloud computing, wherein cloud computing is a type of distributed computing, a virtual supercomputer composed of a group of loosely coupled computers. The network includes but is not limited to the Internet, a wide area network, a metropolitan area network, a local area network, a VPN network, a wireless self-organizing network (Ad Hoc network), etc. Preferably, the device may also be a program running on the user device, the network device, or a device formed by integrating the user device and the network device, the network device and the touch terminal, or the network device and the touch terminal via a network.

[0035] Of course, those skilled in the art should understand that the above-mentioned devices are only examples, and other existing or future devices that are applicable to this application should also be included in the scope of protection of this application and are included here by reference.

[0036] In the description of the present application, “plurality” means two or more, unless otherwise clearly defined.

[0037] Figure 1A flow chart of a method for detecting unqualified goods according to an embodiment of the present application is shown, the method comprising steps S11, S12, and S13. In step S11, a network device obtains connection relationship feature information corresponding to a knowledge graph, wherein the knowledge graph is constructed based on product feature information corresponding to a plurality of products and object feature information corresponding to associated objects associated with the product feature information, each node in the knowledge graph corresponds to a product or an associated object, and the connection relationship feature information is used to characterize the connection relationship between the nodes in the knowledge graph; in step S12, the network device obtains a product detection model through training based on the product feature information, the object feature information, the connection relationship feature information, and product calibration information corresponding to the plurality of products, wherein the product calibration information is used to calibrate whether each of the plurality of products is an unqualified product; in step S13, the network device inputs the target product feature information corresponding to the target product into the product detection model, and obtains product detection information corresponding to the target product output by the product detection model, wherein the product detection information is used to indicate whether the target product is an unqualified product.

[0038] In step S11, the network device obtains connection relationship feature information corresponding to the knowledge graph, wherein the knowledge graph is constructed based on product feature information corresponding to multiple products and object feature information corresponding to associated objects associated with the product feature information. Each node in the knowledge graph corresponds to a product or an associated object, and the connection relationship feature information is used to characterize the connection relationship between each node in the knowledge graph.

[0039] In some embodiments, product feature information includes any information related to the characteristics of the product. Optionally, product feature information includes, but is not limited to, product title description, product category label, product price, corresponding merchant information for the product, product review information, product shipping location, product sales volume, product evaluation information, product description information, etc. In some embodiments, product feature information may have one or more associated objects. In some embodiments, the associated object associated with product feature information can be any object included in the product feature information. For example, if the product feature information of product A includes the merchant information "Merchant B" corresponding to product A, then the associated object associated with the product feature information can be Merchant B. For another example, if the product feature information of product A includes the product description information "Dogs like to use this product," then the associated object associated with the product feature information can be Dog. In some embodiments, objects associated with the semantic content of the product feature information can also be determined as the associated objects associated with the product feature information based on the semantic content of the product feature information. For example, if the semantic content of the product feature information of product A includes "High-end pet food brand," then the objects "Cat" and "Dog" associated with the semantic content can be determined as the associated objects associated with the product feature information. In some embodiments, the associated object may be any object in any form, preferably including but not limited to merchant objects, user objects, commodity objects, etc.

[0040] In some embodiments, the object feature information corresponding to the associated object includes any information related to the features of the associated object. When an associated object is a commodity, the object feature information corresponding to the associated object is the commodity feature information of the commodity; when an associated object is a user, the object feature information corresponding to the associated object includes but is not limited to the user's historical behavior information of uploading, editing, browsing, and purchasing commodities, the user's interest tag information, etc.; when an associated object is a merchant, the object feature information corresponding to the associated object includes but is not limited to other commodities sold by the merchant, merchant evaluation information, etc.

[0041] In some embodiments, a knowledge graph is constructed by using the product feature information corresponding to multiple products and the object feature information corresponding to the associated objects associated with the product feature information. The knowledge graph includes multiple nodes, each node corresponds to a product or an associated object, and the product feature information or object feature information is the attribute of the corresponding product node or associated object node. In some embodiments, the knowledge graph can reflect the relationship between each node. In the knowledge graph, a connection between two nodes (i.e., an association between products, an association between products and associated objects, and an association between associated objects) can be established through the respective attributes of the two nodes (i.e., the product feature information corresponding to the product node or the object feature information corresponding to the associated object node). The two nodes can be directly connected or indirectly connected through one or more other nodes.

[0042] For example, the node corresponding to product A is directly connected to the node of associated object B, and is indirectly connected to the node corresponding to product C through the node corresponding to associated object B; for another example, the product feature information of product A includes "once purchased by user U", and the object feature information of user B includes "once browsed product B". Therefore, through the knowledge graph, a direct association between product A and user U, a direct association between product B and user U, and an indirect association between product A and product B can be established (that is, the node corresponding to product A is directly connected to the node corresponding to user U, the node corresponding to product B is directly connected to the node corresponding to user U, and the node corresponding to product A is indirectly connected to the node corresponding to product B through the node corresponding to user U).

[0043] For another example, the product feature information of product C includes "has appeared many times in movie E", and the object feature information of movie E includes "product D has appeared many times". Therefore, through the knowledge graph, a direct association between product C and movie E, a direct association between product D and movie E, and an indirect association between product C and product D can be established (that is, the node corresponding to product C is directly connected to the node corresponding to movie E, the node corresponding to product D is directly connected to the node corresponding to movie E, and the node corresponding to product C is indirectly connected to the node corresponding to product D through the node corresponding to movie E).

[0044] For another example, the product feature information of product A includes "'dog' appears multiple times in the product description", the product feature information of product B includes "'cat' appears multiple times in the product description", the object feature information of the associated object "dog" includes "and cats are both common pets", and the object feature information of the associated object "cat" includes "and dogs are both common pets". Thus, through the knowledge graph, a direct association can be established between product A and the associated object "dog", a direct association between product B and the associated object "cat", a direct association between the associated object "dog" and the associated object "cat", and an indirect association between product A and product B (that is, the node corresponding to product A is directly connected to the node corresponding to the associated object "dog", the node corresponding to product B is directly connected to the node corresponding to the associated object "cat", and the node corresponding to product A and the node corresponding to product B are indirectly connected through the node corresponding to the associated object "dog" and the node corresponding to the associated object "cat").

[0045] In some embodiments, a connection relationship may be a direct connection relationship between two nodes, for example, node A is directly connected to node B, or a connection relationship may be an indirect connection relationship between two nodes, in which case the connection relationship includes the number of hops corresponding to the connection between the two nodes. For example, if node A is directly connected to node B, the number of hops from node A to node B is 1. For another example, if node A is indirectly connected to node C through node B, the number of hops from node A to node C is 2. In some embodiments, each node may have a connection relationship with only one node or with multiple nodes at the same time. In some embodiments, two nodes may have only one connection relationship or multiple connection relationships at the same time.

[0046] In some embodiments, the connection relationship feature information can be a collection of multiple connection relationships between various nodes in the knowledge graph. For example, the product feature information of product A includes "once purchased by user U1", and the object feature information of user U1 includes "once purchased product B", then the knowledge graph includes three nodes corresponding to product A, user U1, and product B respectively, and the connection relationship feature information corresponding to the three nodes can be obtained based on the knowledge graph. The connection relationship feature information is used to indicate that product A and user U1 have a direct connection relationship of "purchased", user U1 and product B have a direct connection relationship of "purchased", and product A and product B have an indirect connection relationship (an indirect connection relationship means that two nodes are not directly connected, but are connected through one or more other nodes, such as in this example, product A and product B are connected through user U1).

[0047] In some embodiments, the connection relationship between two nodes is directional. For example, in the above example, the connection relationship "purchased" from product A to user U1 is from the node corresponding to product A to the node corresponding to user U1, and the connection relationship "purchased" from user U1 to product B is from the node corresponding to user U1 to the node corresponding to product B.

[0048] In step S12, the network device obtains a product detection model through training based on the product feature information, the object feature information, the connection relationship feature information and the product calibration information corresponding to the multiple products, wherein the product calibration information is used to calibrate whether each of the multiple products is an unqualified product.

[0049] In some embodiments, unqualified products include, but are not limited to, counterfeit and shoddy products, illegal and prohibited products, etc. In some embodiments, product calibration information corresponding to the product is received as input. In some embodiments, at least one calibrated product connected to the product is obtained from the knowledge graph, and based on the similarity between the product feature information corresponding to the product and the product feature information corresponding to the at least one calibrated product, the product is calibrated and the corresponding product calibration information is obtained. In some embodiments, the product calibration information is manually annotated on the product by a model trainer.

[0050] In some embodiments, if at least one calibrated product that is connected to the product in the knowledge graph has similar product feature information such as similar price, similar purchase volume, similar view volume, etc., then the product can be calibrated as an unqualified product based on whether the calibrated product is an unqualified product. For example, if the calibrated product is an unqualified product, then the product can be calibrated as an unqualified product. For another example, if the calibrated product is a qualified product, then the product can be calibrated as a qualified product.

[0051] In some embodiments, multiple product samples and product calibration information corresponding to each product sample are collected to obtain object feature information corresponding to the associated object associated with each product sample and connection relationship feature information obtained through the knowledge graph. A product detection model is obtained through training based on this data. The product detection model can be generated through training based on this data, or the current product detection model can be updated through training based on this data, and the updated current product detection model is used as the product detection model. In some embodiments, the input of the product detection model is the product feature information corresponding to a certain product, and the output is product detection information used to indicate whether the product is an unqualified product.

[0052] In step S13, the network device inputs the target commodity feature information corresponding to the target commodity into the commodity detection model, and obtains commodity detection information corresponding to the target commodity output by the commodity detection model, wherein the commodity detection information is used to indicate whether the target commodity is an unqualified commodity. In some embodiments, the commodity detection information includes indication information for indicating whether the target commodity is an unqualified commodity, such as "1" indicating that the commodity is a qualified commodity and "0" indicating that the commodity is an unqualified commodity. In some embodiments, the commodity detection information includes probability information of the target commodity being an unqualified commodity, such as the commodity detection information indicating that the target commodity has a 70% probability of being an unqualified commodity, such as the commodity detection information indicating that the target commodity has a 30% probability of being a qualified commodity.

[0053] This application can construct a knowledge graph based on the product feature information corresponding to multiple products and the object feature information corresponding to the associated objects associated with the product feature information, and then train a product detection model based on the product feature information, object feature information, connection relationship feature information corresponding to each node in the knowledge graph, and product calibration information. Through this product detection model, it is possible to quickly and accurately detect whether the target product is an unqualified product.

[0054] In some embodiments, step S11 may include step S14 (not shown). In step S14, the network device constructs the knowledge graph based on the product feature information corresponding to the plurality of products and the object feature information corresponding to the associated objects associated with the product feature information.

[0055] In some embodiments, obtaining a product detection model through training includes at least one of the following:

[0056] 1) Generate the product detection model through training

[0057] In some embodiments, a product detection model is generated through training based on the product feature information corresponding to multiple products required to construct a knowledge graph, the object feature information corresponding to the associated objects associated with the product feature information, and the connection relationship feature information obtained from the knowledge graph.

[0058] 2) Update the current commodity detection model through training, and use the updated current commodity detection model as the commodity detection model

[0059] In some embodiments, in response to a new product event for the constructed current knowledge graph, at least one new product corresponding to the new product event, product feature information corresponding to the at least one new product, and object feature information corresponding to the associated object associated with the product feature information are obtained to reconstruct the current knowledge graph. Alternatively, in response to a new product feature information event for an existing product node in the constructed current knowledge graph, at least one existing product corresponding to the new product feature information event, new product feature information corresponding to the at least one existing product, and object feature information corresponding to the associated object associated with the new product feature information are obtained to reconstruct the current knowledge graph. In some embodiments, the latest connection relationship feature information is obtained from the reconstructed, latest knowledge graph, and the currently obtained product detection model is updated through training based on this data, and the updated current product detection model is used as the product detection model. In some embodiments, the new product event may be manually initiated or automatically initiated when predetermined conditions are met. In some embodiments, the currently obtained product detection model is continuously updated to continuously improve the detection accuracy and efficiency of the product detection model.

[0060] In some embodiments, the method further includes: the network device obtains, for each of the multiple commodities, commodity calibration information corresponding to the commodity. In some embodiments, the network device obtains commodity calibration information corresponding to each commodity sent by other devices. In some embodiments, the network device receives commodity calibration information input by a model trainer or other operator for each commodity. In some embodiments, at least one calibrated commodity that has a connection relationship with the commodity is obtained from the knowledge graph, and if the similarity between the commodity feature information corresponding to the commodity and the commodity feature information corresponding to the at least one calibrated commodity meets a predetermined similarity threshold, the commodity calibration information corresponding to the commodity is determined based on the commodity calibration information of the at least one calibrated commodity. In some embodiments, the commodity calibration information corresponding to the commodity may be input from other devices or directly input manually. For example, the model trainer manually determines the commodity calibration information corresponding to the commodity and inputs it to the network device, and the network device receives the commodity calibration information corresponding to the commodity input by the model trainer.

[0061] In some embodiments, obtaining the product calibration information corresponding to the product includes: obtaining at least one calibrated product that has a connection relationship with the product from the knowledge graph; if the similarity between the product feature information corresponding to the product and the product feature information corresponding to the at least one calibrated product meets a predetermined similarity threshold, determining the product calibration information corresponding to the product based on the product calibration information of the at least one calibrated product.

[0062] In some embodiments, at least one calibrated commodity that is directly connected to the commodity is obtained from the knowledge graph. In some embodiments, at least one calibrated commodity that is indirectly connected to the commodity and whose hop count to the commodity is less than a predetermined hop count threshold is obtained from the knowledge graph. In some embodiments, if there is only one calibrated commodity in the knowledge graph that is connected to the commodity, and the similarity between the commodity feature information corresponding to the commodity and the commodity feature information corresponding to the calibrated commodity meets a predetermined similarity threshold, the commodity calibration information corresponding to the calibrated commodity can be directly determined as the commodity calibration information corresponding to the commodity. In some embodiments, similarity includes but is not limited to commodity price similarity, commodity purchase volume similarity, commodity view volume similarity, etc.

[0063] In some embodiments, if there are multiple calibrated commodities that are connected to the commodity in the knowledge graph, it is determined whether there is a calibrated commodity among the multiple calibrated commodities whose similarity with the commodity meets a predetermined similarity threshold. If so, the calibrated commodity whose similarity with the commodity meets the predetermined similarity threshold is directly determined as the target calibrated commodity, or, among the calibrated commodities whose similarity with the commodity meets the predetermined similarity threshold, the calibrated commodity with higher similarity is selected as the target calibrated commodity. If there is only one target calibrated commodity, the commodity calibration information corresponding to the target calibrated commodity can be directly determined as the commodity calibration information corresponding to the commodity. If there are multiple target calibrated commodities and the commodity calibration information of the multiple target calibrated commodities is consistent, it can be directly used as the target calibrated commodity. If there are multiple target calibrated commodities and the commodity calibration information of the multiple target calibrated commodities are inconsistent, the commodity calibration information corresponding to the commodity can be determined according to the ratio between unqualified commodities and qualified commodities in the multiple target calibrated commodities (for example, when the ratio is greater than a predetermined ratio threshold, the commodity calibration information corresponding to the commodity is determined to indicate that the commodity is an unqualified commodity, otherwise the commodity calibration information corresponding to the commodity is determined to indicate that the commodity is a qualified commodity), or, if there is only one unqualified commodity among the multiple target calibrated commodities, the commodity calibration information corresponding to the commodity is directly determined to indicate that the commodity is an unqualified commodity, or, if there is only one qualified commodity among the multiple target calibrated commodities, the commodity calibration information corresponding to the commodity is directly determined to indicate that the commodity is a qualified commodity.

[0064] In some embodiments, before step S14, the method further includes: the network device obtaining, for each of the plurality of commodities, commodity feature information corresponding to the commodity, determining, based on the commodity feature information, an associated object associated with the commodity feature information, and obtaining object feature information corresponding to the associated object. In some embodiments, for each commodity, after collecting commodity-related information corresponding to the commodity locally or online, where the commodity-related information is any information related to the commodity, the commodity feature information corresponding to the commodity can be determined directly based on the commodity-related information, or the commodity feature information corresponding to the commodity can be determined after feature extraction of the commodity-related information.

[0065] In some embodiments, for each product, one or more objects are extracted from the product feature information corresponding to the product, and all or part of the one or more objects are determined as associated objects associated with the product feature information. Optionally, at least one object that has a certain connection with the product is selected from the one or more objects, wherein the object that has a certain connection with the product may be an object that has a semantic inclusion or inclusion relationship (such as "pet" and "dog"), an object that can be used as a set with the product (such as "charger" and "charging cable"), etc.

[0066] In some embodiments, based on the semantic content of the product feature information, the objects associated with the semantic content are determined as the associated objects associated with the product feature information. For example, if the semantic content of the product feature information of product A includes "high-end pet food brand", then the objects "cat" and "dog" associated with the semantic content can be determined as the associated objects associated with the product feature information. That is, the determined associated objects are not the objects directly included in the product feature information, thereby enabling a more comprehensive association.

[0067] In some embodiments, after collecting object-related information corresponding to the associated object locally or online, where the object-related information is any information related to the associated object, the object feature information corresponding to the associated object can be directly determined based on the object-related information. Alternatively, feature extraction can be performed on the object-related information before determining the object feature information corresponding to the associated object. In some embodiments, if the associated object is a product, the product feature information corresponding to the product can be directly used as its corresponding object feature information.

[0068] In some embodiments, for each of the multiple commodities, the commodity characteristic information corresponding to the commodity includes one or more objects, wherein determining the associated object associated with the commodity characteristic information based on the commodity characteristic information includes: using at least one of the one or more objects as the associated object associated with the commodity characteristic information. In some embodiments, the one or more objects can be directly used as the associated object associated with the commodity characteristic information. In some embodiments, if the commodity characteristic information corresponding to the commodity includes multiple objects, at least one object can be determined from the multiple objects as the associated object associated with the commodity characteristic information.

[0069] In some embodiments, the method of using at least one of the one or more objects as an associated object associated with the product feature information includes: determining at least one object from the one or more objects, and using the at least one object as an associated object associated with the product feature information. In some embodiments, at least one object having a degree of association with the product greater than a predetermined degree of association is selected from the one or more objects as an associated object associated with the product feature information. In some embodiments, an object with the highest access rate or click-through rate is determined from the one or more objects as an associated object associated with the product feature information. In some embodiments, at least one object is determined from the one or more objects based on the weight of each object in the product feature information, wherein the weight of each object in the at least one object in the product feature information meets a predetermined weight threshold.

[0070] In some embodiments, the determining of at least one object from the one or more objects includes: determining at least one object from the one or more objects based on the weight of each object in the product feature information, wherein the weight of each object in the at least one object in the product feature information satisfies a predetermined weight threshold. In some embodiments, the weight of an object in the product feature information is used to characterize the importance of the object in the product feature information, and the importance can, to a certain extent, reflect the degree of influence of the object on the use or sales of the product. In some embodiments, based on the weight of each object in the product feature information, at least one object having a corresponding weight greater than or equal to a predetermined weight threshold is determined from the one or more objects.

[0071] In some embodiments, the method further includes: the network device determines the proportion of each object in the product feature information based on the number of times the object appears in the product feature information. In some embodiments, the higher the number of times an object appears in the product feature information, the higher the proportion of the object in the product feature information, and vice versa. In some embodiments, the proportion of the object in the product feature information is adjusted in combination with the position of the object in the product feature information. For example, if an object appears multiple times in the product feature information, and most of them appear in the product description information of the product, the proportion of the object in the product feature information is increased. Optionally, different weighting coefficients can be set for different appearance positions to adjust the proportion of the object in the product feature information.

[0072] In some embodiments, the method further includes: the network device determining the weight of each object in the product feature information based on the semantic importance of the object in the product feature information. In some embodiments, the semantic importance can, to a certain extent, reflect the relevance between the object and the product, with the higher the semantic importance, the higher the relevance between the object and the product. In some embodiments, the higher the semantic importance of an object in the product feature information, the higher the weight of the object in the product feature information, and vice versa.

[0073] In some embodiments, determining the associated objects associated with the product feature information based on the product feature information includes: obtaining the semantic content of the product feature information; and determining the objects associated with the semantic content as the associated objects associated with the product feature information based on the semantic content. In some embodiments, the semantic content of the product feature information is obtained by performing semantic analysis on the product feature information. In some embodiments, one or more keywords in the semantic content are obtained, and the objects associated with the one or more keywords are determined as the associated objects associated with the product feature information; for example, if the semantic content of the product feature information of product A includes "high-end pet food brand", the keyword "pet" in the semantic content can be obtained, and the objects "cat" and "dog" associated with the keyword "pet" can be determined as the associated objects associated with the product feature information.

[0074] In some embodiments, the method further includes: for each associated object, the network device determines the secondary associated object associated with the associated object based on the object feature information corresponding to the associated object, and obtains the object feature information corresponding to the secondary associated object; wherein, step S11 includes: constructing a knowledge graph based on the product feature information corresponding to multiple products, the object feature information corresponding to the associated object associated with the product feature information, and the object feature information corresponding to the secondary associated object associated with the associated object. In some embodiments, the secondary associated object is obtained by performing one or more association operations on the associated object associated with the product feature information, and an associated object may be associated with one or more secondary associated objects. In some embodiments, the secondary associated object associated with the associated object can be determined based on a pre-set association relationship. The implementation method of obtaining the object feature information corresponding to the secondary associated object is the same or similar to the implementation method of obtaining the object feature information corresponding to the associated object described above, and will not be repeated here. In some embodiments, a knowledge graph is constructed based on the product feature information corresponding to multiple products, the object feature information corresponding to the associated objects associated with the product feature information, and the object feature information corresponding to the secondary associated objects associated with the associated objects. The constructed knowledge graph includes multiple nodes, each of which corresponds to a product, an associated object, or a secondary associated object. Thus, a node corresponding to a product may be indirectly connected to a secondary associated object corresponding to the associated object through an associated object corresponding to the product, or may be indirectly connected to other secondary associated objects corresponding to the associated object through an associated object corresponding to the product and at least one secondary associated object corresponding to the associated object. In some embodiments, the secondary associated object corresponding to the associated object can be any object included in the object feature information corresponding to the associated object. In some embodiments, the object associated with the semantic content of the object feature information corresponding to the associated object can be determined as the associated object associated with the associated object based on the semantic content of the object feature information corresponding to the associated object. In some embodiments, the secondary associated object can be any object of any form, preferably including but not limited to merchant objects, user objects, product objects, etc. It should be noted that any explanatory description of the associated object in the aforementioned embodiment can also be applied to the secondary associated object.

[0075] In some embodiments, determining the secondary associated objects corresponding to the associated object based on the object feature information corresponding to the associated object includes steps S15 (not shown), S16 (not shown), and S17 (not shown). In step S15, the network device performs an association operation on the associated object based on the object feature information corresponding to the associated object, obtaining one or more secondary associated objects; in step S16, the network device performs an association operation again based on the object feature information corresponding to each secondary associated object obtained in this association operation, obtaining one or more secondary associated objects; in step S17, the network device repeats step S16 until the stop association condition is met. In some embodiments, an association operation refers to an operation used to determine the secondary associated objects of an object (which may be an associated object corresponding to product feature information or an already obtained secondary associated object), wherein performing one association operation may obtain one or more secondary associated objects. In some embodiments, after performing an association operation on the object feature information corresponding to the associated object and obtaining one or more secondary associated objects, at least one association operation may be performed again to obtain more secondary associated objects. For example, if the associated object corresponding to the product feature information of product A is "dog", then an association operation is performed on the object feature information of the associated object to obtain the secondary associated object "cat" corresponding to "dog". After that, an association operation is performed on the object feature information corresponding to "cat" to obtain the secondary associated object "movie M" corresponding to "cat". After that, an association operation is performed on the object feature information corresponding to "movie M" to obtain the secondary associated object "British Shorthair" corresponding to "movie M", and so on, until the stop association condition is met, thereby obtaining a more comprehensive secondary associated objects. In some embodiments, the stop association condition includes any condition for triggering the cessation of the association operation.

[0076] In some embodiments, the stop association condition includes any one of the following: the number of times the association operation is performed reaches a predetermined number threshold; the number of secondary associated objects obtained reaches a predetermined number threshold. In some embodiments, the number threshold can be set based on experience, and optionally, the number threshold can be adjusted based on user feedback information. In some embodiments, after the last association operation, the number of secondary associated objects obtained (that is, the total number of secondary associated objects obtained by multiple association operations that have been performed) is less than a predetermined number threshold, and after this association operation, the number of secondary associated objects obtained may be equal to or exceed the predetermined number threshold, then the stop association condition is deemed to be met.

[0077] In some embodiments, the method further includes: if the product detection information indicates that the target product is an unqualified product, the network device outputs at least one calibrated unqualified product that is connected to the target product. In some embodiments, the at least one calibrated unqualified product is obtained from a knowledge graph, and the calibrated unqualified product may be directly connected to the target product or may be indirectly connected to the target product. For example, if the product detection information output after product A is input into the product detection model indicates that product A is an unqualified product, then calibrated unqualified products B and C that are directly connected to product A are output. Optionally, at least one calibrated qualified product that is connected to the target product can also be output for further comparison or processing.

[0078] In some embodiments, if there are multiple calibrated unqualified products that have a connection relationship with the target product; wherein, the method further includes: the network device determines at least one calibrated unqualified product from the multiple calibrated unqualified products, wherein the connection hop count corresponding to the connection relationship between each calibrated unqualified product in the at least one calibrated unqualified product and the target product is less than or equal to a predetermined hop count threshold. In some embodiments, multiple calibrated unqualified products that have a connection relationship with the target product are obtained from the knowledge graph, and the connection hop count corresponding to the other connection relationship between the target product and each calibrated unqualified product is obtained, and then at least one calibrated unqualified product whose corresponding connection hop count is less than or equal to a predetermined hop count threshold is selected from the multiple calibrated unqualified products. In some embodiments, the hop count threshold can be set based on experience, and optionally, the hop count threshold can be adjusted based on feedback information for product detection information.

[0079] Figure 2 A flow chart of a method for constructing a knowledge graph according to one embodiment of the present application is shown, the method comprising step S21. In step S21, a network device constructs a knowledge graph based on product feature information corresponding to a plurality of products and object feature information corresponding to associated objects associated with the product feature information. The knowledge graph comprises a plurality of nodes, each node in the knowledge graph corresponding to a product or an associated object. The relevant operations in this embodiment have been described in detail in the preceding embodiments and will not be repeated here.

[0080] Figure 3A flow chart of a method for obtaining a commodity detection model through training according to an embodiment of the present application is shown, and the method includes steps S31 and S32. In step S31, the network device obtains the connection relationship feature information corresponding to the knowledge graph, wherein the knowledge graph is constructed based on the commodity feature information corresponding to a plurality of commodities and the object feature information corresponding to the associated objects associated with the commodity feature information, and the knowledge graph includes a plurality of nodes, each node in the knowledge graph corresponds to a commodity or an associated object, and the connection relationship feature information is used to characterize the connection relationship between the nodes in the knowledge graph; in step S32, the network device obtains the commodity detection model through training based on the commodity feature information, the object feature information, the connection relationship feature information and the commodity calibration information corresponding to the plurality of commodities, wherein the commodity calibration information is used to calibrate whether each of the plurality of commodities is an unqualified commodity. The relevant operations in this embodiment have been described in detail in the aforementioned embodiment and will not be repeated here.

[0081] Figure 4 A structural diagram of a network device for detecting unqualified products according to an embodiment of the present application is shown, and the device includes a first module 11, a second module 12, and a third module 13. The first module 11 is used to obtain connection relationship feature information corresponding to a knowledge graph, wherein the knowledge graph is constructed based on product feature information corresponding to multiple products and object feature information corresponding to associated objects associated with the product feature information, the knowledge graph includes multiple nodes, each node in the knowledge graph corresponds to a product or an associated object, and the connection relationship feature information is used to characterize the connection relationship between each node in the knowledge graph; the first module 12 is used to obtain a product detection model through training based on the product feature information, the object feature information, the connection relationship feature information, and product calibration information corresponding to the multiple products, wherein the product calibration information is used to calibrate whether each product in the multiple products is an unqualified product; the third module 13 is used to input the target product feature information corresponding to the target product into the product detection model to obtain product detection information corresponding to the target product output by the product detection model, wherein the product detection information is used to indicate whether the target product is an unqualified product.

[0082] Module 11 is used to obtain connection relationship feature information corresponding to the knowledge graph, wherein the knowledge graph is constructed based on product feature information corresponding to multiple products and object feature information corresponding to associated objects associated with the product feature information. The knowledge graph includes multiple nodes, and each node in the knowledge graph corresponds to a product or an associated object. The connection relationship feature information is used to characterize the connection relationship between each node in the knowledge graph.

[0083] In some embodiments, product feature information includes any information related to the characteristics of the product. Optionally, product feature information includes, but is not limited to, product title description, product category label, product price, corresponding merchant information for the product, product review information, product shipping location, product sales volume, product evaluation information, product description information, etc. In some embodiments, product feature information may have one or more associated objects. In some embodiments, the associated object associated with product feature information can be any object included in the product feature information. For example, if the product feature information of product A includes the merchant information "Merchant B" corresponding to product A, then the associated object associated with the product feature information can be Merchant B. For another example, if the product feature information of product A includes the product description information "Dogs like to use this product," then the associated object associated with the product feature information can be Dog. In some embodiments, objects associated with the semantic content of the product feature information can also be determined as the associated objects associated with the product feature information based on the semantic content of the product feature information. For example, if the semantic content of the product feature information of product A includes "High-end pet food brand," then the objects "Cat" and "Dog" associated with the semantic content can be determined as the associated objects associated with the product feature information. In some embodiments, the associated object may be any object in any form, preferably including but not limited to merchant objects, user objects, commodity objects, etc.

[0084] In some embodiments, the object feature information corresponding to the associated object includes any information related to the features of the associated object. When an associated object is a commodity, the object feature information corresponding to the associated object is the commodity feature information of the commodity; when an associated object is a user, the object feature information corresponding to the associated object includes but is not limited to the user's historical behavior information of uploading, editing, browsing, and purchasing commodities, the user's interest tag information, etc.; when an associated object is a merchant, the object feature information corresponding to the associated object includes but is not limited to other commodities sold by the merchant, merchant evaluation information, etc.

[0085] In some embodiments, a knowledge graph is constructed by using the product feature information corresponding to multiple products and the object feature information corresponding to the associated objects associated with the product feature information. The knowledge graph includes multiple nodes, each node corresponds to a product or an associated object, and the product feature information or object feature information is the attribute of the corresponding product node or associated object node. In some embodiments, the knowledge graph can reflect the relationship between each node. In the knowledge graph, a connection between two nodes (i.e., an association between products, an association between products and associated objects, and an association between associated objects) can be established through the respective attributes of the two nodes (i.e., the product feature information corresponding to the product node or the object feature information corresponding to the associated object node). The two nodes can be directly connected or indirectly connected through one or more other nodes.

[0086] For example, the node corresponding to product A is directly connected to the node of associated object B, and is indirectly connected to the node corresponding to product C through the node corresponding to associated object B; for another example, the product feature information of product A includes "once purchased by user U", and the object feature information of user B includes "once browsed product B". Therefore, through the knowledge graph, a direct association between product A and user U, a direct association between product B and user U, and an indirect association between product A and product B can be established (that is, the node corresponding to product A is directly connected to the node corresponding to user U, the node corresponding to product B is directly connected to the node corresponding to user U, and the node corresponding to product A is indirectly connected to the node corresponding to product B through the node corresponding to user U).

[0087] For another example, the product feature information of product C includes "has appeared many times in movie E", and the object feature information of movie E includes "product D has appeared many times". Therefore, through the knowledge graph, a direct association between product C and movie E, a direct association between product D and movie E, and an indirect association between product C and product D can be established (that is, the node corresponding to product C is directly connected to the node corresponding to movie E, the node corresponding to product D is directly connected to the node corresponding to movie E, and the node corresponding to product C is indirectly connected to the node corresponding to product D through the node corresponding to movie E).

[0088] For another example, the product feature information of product A includes "'dog' appears multiple times in the product description", the product feature information of product B includes "'cat' appears multiple times in the product description", the object feature information of the associated object "dog" includes "and cats are both common pets", and the object feature information of the associated object "cat" includes "and dogs are both common pets". Thus, through the knowledge graph, a direct association can be established between product A and the associated object "dog", a direct association between product B and the associated object "cat", a direct association between the associated object "dog" and the associated object "cat", and an indirect association between product A and product B (that is, the node corresponding to product A is directly connected to the node corresponding to the associated object "dog", the node corresponding to product B is directly connected to the node corresponding to the associated object "cat", and the node corresponding to product A and the node corresponding to product B are indirectly connected through the node corresponding to the associated object "dog" and the node corresponding to the associated object "cat").

[0089] In some embodiments, a connection relationship may be a direct connection relationship between two nodes, for example, node A is directly connected to node B, or a connection relationship may be an indirect connection relationship between two nodes, in which case the connection relationship includes the number of hops corresponding to the connection between the two nodes. For example, if node A is directly connected to node B, the number of hops from node A to node B is 1. For another example, if node A is indirectly connected to node C through node B, the number of hops from node A to node C is 2. In some embodiments, each node may have a connection relationship with only one node or with multiple nodes at the same time. In some embodiments, two nodes may have only one connection relationship or multiple connection relationships at the same time.

[0090] In some embodiments, the connection relationship feature information can be a collection of multiple connection relationships between various nodes in the knowledge graph. For example, the product feature information of product A includes "once purchased by user U1", and the object feature information of user U1 includes "once purchased product B", then the knowledge graph includes three nodes corresponding to product A, user U1, and product B respectively, and the connection relationship feature information corresponding to the three nodes can be obtained based on the knowledge graph. The connection relationship feature information is used to indicate that product A and user U1 have a direct connection relationship of "purchased", user U1 and product B have a direct connection relationship of "purchased", and product A and product B have an indirect connection relationship (an indirect connection relationship means that two nodes are not directly connected, but are connected through one or more other nodes, such as in this example, product A and product B are connected through user U1).

[0091] In some embodiments, the connection relationship between two nodes is directional. For example, in the above example, the connection relationship "purchased" from product A to user U1 is from the node corresponding to product A to the node corresponding to user U1, and the connection relationship "purchased" from user U1 to product B is from the node corresponding to user U1 to the node corresponding to product B.

[0092] Module 12 is used to obtain a product detection model through training based on the product feature information, the object feature information, the connection relationship feature information and the product calibration information corresponding to the multiple products, wherein the product calibration information is used to calibrate whether each of the multiple products is an unqualified product.

[0093] In some embodiments, unqualified products include, but are not limited to, counterfeit and shoddy products, illegal and prohibited products, etc. In some embodiments, product calibration information corresponding to the product is received as input. In some embodiments, at least one calibrated product connected to the product is obtained from the knowledge graph, and based on the similarity between the product feature information corresponding to the product and the product feature information corresponding to the at least one calibrated product, the product is calibrated and the corresponding product calibration information is obtained. In some embodiments, the product calibration information is manually annotated on the product by a model trainer.

[0094] In some embodiments, if at least one calibrated product that is connected to the product in the knowledge graph has similar product feature information such as similar price, similar purchase volume, similar view volume, etc., then the product can be calibrated as an unqualified product based on whether the calibrated product is an unqualified product. For example, if the calibrated product is an unqualified product, then the product can be calibrated as an unqualified product. For another example, if the calibrated product is a qualified product, then the product can be calibrated as a qualified product.

[0095] In some embodiments, multiple product samples and product calibration information corresponding to each product sample are collected to obtain object feature information corresponding to the associated object associated with each product sample and connection relationship feature information obtained through the knowledge graph. A product detection model is obtained through training based on this data. The product detection model can be generated through training based on this data, or the current product detection model can be updated through training based on this data, and the updated current product detection model is used as the product detection model. In some embodiments, the input of the product detection model is the product feature information corresponding to a certain product, and the output is product detection information used to indicate whether the product is an unqualified product.

[0096] Module 13 is configured to input target commodity feature information corresponding to the target commodity into the commodity detection model, and obtain commodity detection information corresponding to the target commodity output by the commodity detection model, wherein the commodity detection information is used to indicate whether the target commodity is an unqualified commodity. In some embodiments, the commodity detection information includes indication information for indicating whether the target commodity is an unqualified commodity, such as "1" indicating that the commodity is a qualified commodity and "0" indicating that the commodity is an unqualified commodity. In some embodiments, the commodity detection information includes probability information of the target commodity being an unqualified commodity, such as if the commodity detection information indicates that the target commodity has a 70% probability of being an unqualified commodity, or if the commodity detection information indicates that the target commodity has a 30% probability of being a qualified commodity.

[0097] In some embodiments, the device further includes a module 14 (not shown). The module 14 is configured to construct the knowledge graph based on the commodity feature information corresponding to the plurality of commodities and the object feature information corresponding to the associated objects associated with the commodity feature information. Figure 1 The embodiments shown are the same or similar and therefore will not be described in detail, but are incorporated herein by reference.

[0098] In some embodiments, obtaining a product detection model through training includes at least one of the following:

[0099] 1) Generate the product detection model through training

[0100] 2) Update the current commodity detection model through training, and use the updated current commodity detection model as the commodity detection model

[0101] Here, the relevant operations are Figure 1 The embodiments shown are the same or similar and therefore will not be described in detail, but are incorporated herein by reference.

[0102] In some embodiments, the device is further configured to: for each of the plurality of commodities, obtain commodity calibration information corresponding to the commodity. Figure 1 The embodiments shown are the same or similar and therefore will not be described in detail, but are incorporated herein by reference.

[0103] In some embodiments, obtaining the product calibration information corresponding to the product includes: obtaining at least one calibrated product that has a connection relationship with the product from the knowledge graph; if the similarity between the product feature information corresponding to the product and the product feature information corresponding to the at least one calibrated product meets a predetermined similarity threshold, determining the product calibration information corresponding to the product based on the product calibration information of the at least one calibrated product. Figure 1The embodiments shown are the same or similar and therefore will not be described in detail, but are incorporated herein by reference.

[0104] In some embodiments, the device is further configured to: for each of the plurality of commodities, obtain commodity feature information corresponding to the commodity, determine the associated object associated with the commodity feature information based on the commodity feature information, and obtain object feature information corresponding to the associated object. Figure 1 The embodiments shown are the same or similar and therefore will not be described in detail, but are incorporated herein by reference.

[0105] In some embodiments, for each of the plurality of commodities, the commodity characteristic information corresponding to the commodity includes one or more objects, wherein determining the associated object associated with the commodity characteristic information based on the commodity characteristic information includes: using at least one of the one or more objects as the associated object associated with the commodity characteristic information. Figure 1 The embodiments shown are the same or similar and therefore will not be described in detail, but are incorporated herein by reference.

[0106] In some embodiments, the step of using at least one of the one or more objects as an associated object associated with the product feature information includes: determining at least one object from the one or more objects, and using the at least one object as an associated object associated with the product feature information. Figure 1 The embodiments shown are the same or similar and therefore will not be described in detail, but are incorporated herein by reference.

[0107] In some embodiments, determining at least one object from the one or more objects includes: determining at least one object from the one or more objects based on the weight of each object in the product feature information, wherein the weight of each object in the at least one object in the product feature information satisfies a predetermined weight threshold. Figure 1 The embodiments shown are the same or similar and therefore will not be described in detail, but are incorporated herein by reference.

[0108] In some embodiments, the device is further configured to: determine the weight of each object in the product feature information based on the number of times each object appears in the product feature information. Figure 1 The embodiments shown are the same or similar and therefore will not be described in detail, but are incorporated herein by reference.

[0109] In some embodiments, the device is further configured to: determine the weight of each object in the product feature information according to the semantic importance of the object in the product feature information. Figure 1The embodiments shown are the same or similar and therefore will not be described in detail, but are incorporated herein by reference.

[0110] In some embodiments, the step of determining the associated object associated with the product feature information based on the product feature information includes: obtaining the semantic content of the product feature information; and determining, based on the semantic content, the object associated with the semantic content as the associated object associated with the product feature information. Figure 1 The embodiments shown are the same or similar and therefore will not be described in detail, but are incorporated herein by reference.

[0111] In some embodiments, the device is further configured to: for each associated object, determine the secondary associated object associated with the associated object based on the object feature information corresponding to the associated object, and obtain the object feature information corresponding to the secondary associated object; wherein, step S11 includes: constructing a knowledge graph based on the product feature information corresponding to a plurality of products, the object feature information corresponding to the associated object associated with the product feature information, and the object feature information corresponding to the secondary associated object associated with the associated object. Here, the relevant operations are the same as Figure 1 The embodiments shown are the same or similar and therefore will not be described in detail, but are incorporated herein by reference.

[0112] In some embodiments, the method of determining the secondary associated object corresponding to the associated object based on the object feature information corresponding to the associated object includes a five-module 15 (not shown), a six-module 16 (not shown) and a seven-module 17 (not shown). The five-module 15 is used to perform an association operation on the associated object based on the object feature information corresponding to the associated object, and obtain one or more secondary associated objects; the six-module 16 is used to perform an association operation again based on the object feature information corresponding to the secondary associated object for each secondary associated object obtained by this association operation, and obtain one or more secondary associated objects; the seven-module 17 is used to trigger the six-module 16 to repeat the operation until the stop association condition is met. Here, the specific implementation of the five-module 15, the six-module 16 and the seven-module 17 is the same as Figure 1 The embodiments of steps S15, S16 and S17 are the same or similar, so they are not repeated here and are included here by reference.

[0113] In some embodiments, the stop association condition includes any one of the following: the number of executions of the association operation reaches a predetermined number threshold; the number of secondary association objects obtained reaches a predetermined number threshold. Figure 1 The embodiments shown are the same or similar and therefore will not be described in detail, but are incorporated herein by reference.

[0114] In some embodiments, the device is further configured to: if the product detection information indicates that the target product is an unqualified product, output at least one calibrated unqualified product that is connected to the target product. Figure 1 The embodiments shown are the same or similar and therefore will not be described in detail, but are incorporated herein by reference.

[0115] In some embodiments, if there are multiple calibrated unqualified products that have a connection relationship with the target product; wherein the device is further used to: determine at least one calibrated unqualified product from the multiple calibrated unqualified products, wherein the connection hop count corresponding to the connection relationship between each calibrated unqualified product in the at least one calibrated unqualified product and the target product is less than or equal to a predetermined hop count threshold. Here, the relevant operations are the same as Figure 1 The embodiments shown are the same or similar and therefore will not be described in detail, but are incorporated herein by reference.

[0116] Figure 5 An exemplary system is shown that can be used to implement the various embodiments described herein.

[0117] like Figure 5 In some embodiments, the system 300 can function as any of the devices described in the various embodiments. In some embodiments, the system 300 can include one or more computer-readable media (e.g., system memory or NVM / storage device 320) having instructions and one or more processors (e.g., processor(s) 305) coupled to the one or more computer-readable media and configured to execute the instructions to implement the modules and thereby perform the actions described herein.

[0118] For one embodiment, system control module 310 may include any suitable interface controller to provide any suitable interface to at least one of processor(s) 305 and / or any suitable device or component in communication with system control module 310 .

[0119] The system control module 310 may include a memory controller module 330 to provide an interface to the system memory 315. The memory controller module 330 may be a hardware module, a software module, and / or a firmware module.

[0120] System memory 315 can be used, for example, to load and store data and / or instructions for system 300. For one embodiment, system memory 315 can include any suitable volatile memory, such as a suitable DRAM. In some embodiments, system memory 315 can include double data rate type four synchronous dynamic random access memory (DDR4 SDRAM).

[0121] For one embodiment, system control module 310 may include one or more input / output (I / O) controllers to provide interfaces to NVM / storage device 320 and communication interface(s) 325 .

[0122] For example, NVM / storage 320 may be used to store data and / or instructions. NVM / storage 320 may include any suitable non-volatile memory (e.g., flash memory) and / or may include any suitable non-volatile storage device(s) (e.g., one or more hard disk drives (HDDs), one or more compact disk (CD) drives, and / or one or more digital versatile disk (DVD) drives).

[0123] NVM / storage device 320 may include storage resources that are physically part of the device on which system 300 is installed, or it may be accessible to the device without being part of the device. For example, NVM / storage device 320 may be accessed over a network via communication interface(s) 325.

[0124] Communication interface(s) 325 may provide an interface for system 300 to communicate over one or more networks and / or with any other suitable devices. System 300 may wirelessly communicate with one or more components of a wireless network in accordance with any of one or more wireless network standards and / or protocols.

[0125] For one embodiment, at least one of the processor(s) 305 may be packaged together with the logic of one or more controllers of the system control module 310 (e.g., the memory controller module 330). For one embodiment, at least one of the processor(s) 305 may be packaged together with the logic of one or more controllers of the system control module 310 to form a system-in-package (SiP). For one embodiment, at least one of the processor(s) 305 may be integrated on the same die with the logic of one or more controllers of the system control module 310. For one embodiment, at least one of the processor(s) 305 may be integrated on the same die with the logic of one or more controllers of the system control module 310 to form a system-on-chip (SoC).

[0126] In various embodiments, system 300 may be, but is not limited to, a server, a workstation, a desktop computing device, or a mobile computing device (e.g., a laptop computing device, a handheld computing device, a tablet computer, a netbook, etc.). In various embodiments, system 300 may have more or fewer components and / or a different architecture. For example, in some embodiments, system 300 includes one or more cameras, a keyboard, a liquid crystal display (LCD) screen (including a touchscreen display), a non-volatile memory port, multiple antennas, a graphics chip, an application-specific integrated circuit (ASIC), and a speaker.

[0127] The present application also provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer code. When the computer code is executed, the method described in any one of the preceding items is executed.

[0128] The present application also provides a computer program product. When the computer program product is executed by a computer device, the method described in any one of the preceding items is executed.

[0129] The present application also provides a computer device, comprising:

[0130] one or more processors;

[0131] a memory for storing one or more computer programs;

[0132] When the one or more computer programs are executed by the one or more processors, the one or more processors are caused to implement the method as described in any one of the preceding items.

[0133] It should be noted that the application can be implemented in software and / or a combination of software and hardware, for example, can be implemented using an application specific integrated circuit (ASIC), a general purpose computer or any other similar hardware device. In one embodiment, the software program of the application can be executed by a processor to realize the steps or functions described above. Similarly, the software program of the application (including relevant data structures) can be stored in a computer-readable recording medium, for example, a RAM memory, a magnetic or optical drive or a floppy disk and similar devices. In addition, some steps or functions of the application can be implemented using hardware, for example, as a circuit that cooperates with a processor to perform each step or function.

[0134] In addition, a part of the present application may be applied as a computer program product, such as a computer program instruction, which, when executed by a computer, can call or provide the method and / or technical solution according to the present application through the operation of the computer. Those skilled in the art should understand that the form in which the computer program instruction exists in a computer-readable medium includes but is not limited to a source file, an executable file, an installation package file, etc. Accordingly, the way in which the computer program instruction is executed by the computer includes but is not limited to: the computer directly executes the instruction, or the computer compiles the instruction and then executes the corresponding compiled program, or the computer reads and executes the instruction, or the computer reads and installs the instruction and then executes the corresponding installed program. Here, the computer-readable medium can be any available computer-readable storage medium or communication medium that can be accessed by the computer.

[0135] Communication media include media by which communication signals containing, for example, computer-readable instructions, data structures, program modules, or other data are transmitted from one system to another. Communication media may include guided transmission media such as cables and wires (e.g., fiber optic, coaxial, etc.) and wireless (unguided transmission) media capable of propagating energy waves, such as acoustic, electromagnetic, RF, microwave, and infrared. Computer-readable instructions, data structures, program modules, or other data may be embodied as, for example, a modulated data signal in a wireless medium such as a carrier wave or similar mechanism such as that embodied as part of spread spectrum technology. The term "modulated data signal" refers to a signal that has one or more of its characteristics changed or set in such a manner as to encode information in the signal. Modulation may be analog, digital, or a hybrid modulation technique.

[0136] By way of example and not limitation, computer-readable storage media may include volatile and non-volatile, removable and non-removable media implemented in any method or technology for storage of information such as computer-readable instructions, data structures, program modules or other data. For example, computer-readable storage media include, but are not limited to, volatile memory, such as random access memory (RAM, DRAM, SRAM); and non-volatile memory, such as flash memory, various read-only memories (ROM, PROM, EPROM, EEPROM), magnetic and ferromagnetic / ferroelectric memories (MRAM, FeRAM); and magnetic and optical storage devices (hard disks, magnetic tapes, CDs, DVDs); or other media now known or later developed that can store computer-readable information / data for use by a computer system.

[0137] Here, according to one embodiment of the present application, a device is included, which includes a memory for storing computer program instructions and a processor for executing the program instructions, wherein, when the computer program instructions are executed by the processor, the device is triggered to run the methods and / or technical solutions based on the aforementioned multiple embodiments of the present application.

[0138] It is obvious to those skilled in the art that the present application is not limited to the details of the above-mentioned exemplary embodiments, and that the present application can be implemented in other specific forms without departing from the spirit or basic characteristics of the present application. Therefore, from any point of view, the embodiments should be regarded as exemplary and non-restrictive, and the scope of the present application is defined by the appended claims rather than the above description, and it is intended that all changes that fall within the meaning and scope of the equivalent elements of the claims are included in the present application. Any figure mark in the claims should not be regarded as limiting the claims involved. In addition, it is obvious that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices stated in the device claim can also be implemented by one unit or device through software or hardware. Words such as first and second are used to indicate names and do not indicate any particular order.

Claims

1. A method for detecting unqualified goods, wherein: The method comprises: Obtaining connection relationship feature information corresponding to a knowledge graph, wherein the knowledge graph is constructed based on product feature information corresponding to multiple products and object feature information corresponding to associated objects associated with the product feature information, the knowledge graph includes multiple nodes, each node in the knowledge graph corresponds to a product or an associated object, the connection relationship feature information is used to characterize the connection relationship between each node in the knowledge graph, the associated objects include objects included in the product feature information or objects associated with the semantic content of the product feature information, and the form of the associated objects includes at least one of a merchant object and a user object; Obtaining a product detection model through training based on the product feature information, the object feature information, the connection relationship feature information, and product calibration information corresponding to the multiple products, wherein the product calibration information is used to calibrate whether each of the multiple products is an unqualified product; The target product feature information corresponding to the target product is input into the product detection model to obtain product detection information corresponding to the target product output by the product detection model, wherein the product detection information is used to indicate whether the target product is an unqualified product, and the product detection model is used to detect whether the target product is qualified.

2. The method according to claim 1, wherein The step of obtaining the connection relationship feature information corresponding to the knowledge graph further includes: The knowledge graph is constructed based on product feature information corresponding to multiple products and object feature information corresponding to associated objects associated with the product feature information.

3. The method according to claim 2, wherein: The step of constructing the knowledge graph based on the product feature information corresponding to the plurality of products and the object feature information corresponding to the associated objects associated with the product feature information may also include: For each of the multiple commodities, commodity feature information corresponding to the commodity is obtained, an associated object associated with the commodity feature information is determined based on the commodity feature information, and object feature information corresponding to the associated object is obtained.

4. The method according to claim 3, wherein: For each of the multiple commodities, the commodity feature information corresponding to the commodity includes one or more objects; The determining of the associated object associated with the product characteristic information according to the product characteristic information includes: At least one of the one or more objects is used as an associated object associated with the product feature information.

5. The method according to claim 4, wherein The taking at least one of the one or more objects as an associated object associated with the product feature information includes: At least one object is determined from the one or more objects, and the at least one object is used as an associated object associated with the commodity feature information.

6. The method according to claim 5, wherein: The determining of at least one object from the one or more objects comprises: At least one object is determined from the one or more objects according to a proportion of each of the one or more objects in the product feature information, wherein a proportion of each of the at least one object in the product feature information satisfies a predetermined proportion threshold.

7. The method according to claim 6, wherein: The method further comprises: According to the number of occurrences of each object in the product feature information, the proportion of the object in the product feature information is determined.

8. The method according to claim 6, wherein: The method further comprises: According to the semantic importance of each object in the product feature information, the proportion of the object in the product feature information is determined.

9. The method according to claim 3, wherein: The determining, based on the product feature information, the associated object associated with the product feature information includes: Obtaining the semantic content of the product feature information; According to the semantic content, an object associated with the semantic content is determined as an associated object associated with the product feature information.

10. The method according to claim 2, wherein: The method further comprises: For each associated object, determine the secondary associated object associated with the associated object according to the object feature information corresponding to the associated object, and obtain the object feature information corresponding to the secondary associated object; The step of constructing the knowledge graph based on the product feature information corresponding to the plurality of products and the object feature information corresponding to the associated objects associated with the product feature information includes: The knowledge graph is constructed based on product feature information corresponding to multiple products, object feature information corresponding to associated objects associated with the product feature information, and object feature information corresponding to secondary associated objects associated with the associated objects.

11. The method according to claim 10, wherein: The determining, based on the object feature information corresponding to the associated object, a secondary associated object corresponding to the associated object includes: Performing an association operation on the associated object according to the object feature information corresponding to the associated object to obtain one or more secondary associated objects; For each secondary associated object obtained by this association operation, the association operation is performed again according to the object feature information corresponding to the secondary associated object to obtain one or more secondary associated objects; Repeat the step of re-performing the association operation until the stop association condition is met.

12. The method according to claim 11, wherein The stop association condition includes any one of the following: The number of executions of the associated operation reaches a predetermined threshold number; The number of the obtained secondary associated objects reaches a predetermined number threshold.

13. The method according to claim 1, wherein The product detection model obtained through training includes at least one of the following: Generating the commodity detection model through training; The current commodity detection model is updated through training, and the updated current commodity detection model is used as the commodity detection model.

14. The method according to claim 1, wherein The method further comprises: For each of the multiple commodities, commodity calibration information corresponding to the commodity is obtained.

15. The method according to claim 14, wherein The obtaining of the product calibration information corresponding to the product includes: Obtaining at least one marked product that has a connection relationship with the product from the knowledge graph; If the similarity between the product feature information corresponding to the product and the product feature information corresponding to the at least one calibrated product meets a predetermined similarity threshold, the product calibration information corresponding to the product is determined based on the product calibration information of the at least one calibrated product.

16. The method according to claim 1, wherein The method further comprises: If the commodity detection information indicates that the target commodity is an unqualified commodity, at least one calibrated unqualified commodity that is connected to the target commodity is output.

17. The method according to claim 16, wherein If there are multiple marked unqualified products that are connected to the target product; The method further comprises: At least one calibrated unqualified commodity is determined from the multiple calibrated unqualified commodities, wherein a connection hop count corresponding to a connection relationship between each calibrated unqualified commodity in the at least one calibrated unqualified commodity and the target commodity is less than or equal to a predetermined hop count threshold.

18. A device for detecting unqualified goods, wherein: The device comprises: processor; and A memory arranged to store computer executable instructions which, when executed, cause the processor to perform the method of any one of claims 1 to 17.

19. A computer-readable medium storing instructions that, when executed, cause a system to perform the operations of the method of any one of claims 1 to 17.

Images