Order identification method, device and electronic equipment

Abstract

The present invention relates to an order identification method, device and electronic device, the method comprising: obtaining a feature vector reflecting the degree of matching between a target transaction order and associated data based on a location area corresponding to a target transaction order when it is generated and set order information of the target transaction order; wherein the associated data reflects the association between the location area and an abnormal group for the set order information; and identifying whether the target transaction order is a risky order at least based on the feature vector.

Description

Technical Field

[0001] The present invention relates to the field of information processing, and more specifically, to an order identification method, an order identification device, an electronic device and a computer-readable storage medium. Background Art

[0002] At present, purchasing goods through e-commerce trading platforms has become the main consumption channel for people. When people purchase goods through e-commerce trading platforms, merchant ratings and product sales are the main reference factors for people to choose goods. Therefore, there is a large demand for artificially increasing product sales, which has spawned a black industry of cheating that relies on e-commerce transactions. In order to fight against the black industry of cheating, e-commerce trading platforms have developed risk control platforms to identify risky orders, so as to reduce the impact of false phenomena caused by fake orders on users.

[0003] As e-commerce trading platforms continue to grow stronger, the black industry of cheating that relies on e-commerce transactions is also growing rapidly. A series of cheating methods and behaviors, including fake transactions, fake orders, fake reviews, and risky traffic, have brought huge challenges to the fairness and credibility of the e-commerce ecosystem. In particular, there are now extremely concealed fake orders from the ground. Since fake orders from the ground are highly similar to ordinary orders, the traditional risk control platform has reduced its accuracy in identifying risky fake orders. It is necessary to provide a new method to improve the accuracy of identifying fake orders from the ground, thereby increasing the resistance strength of the risk control platform. Summary of the invention

[0004] One purpose of an embodiment of the present invention is to provide a new technical solution for identifying risky orders.

[0005] According to a first aspect of the present invention, there is provided an order identification method, comprising:

[0006] According to the location area corresponding to the target transaction order when it is generated and the set order information of the target transaction order, a feature vector reflecting the matching degree between the target transaction order and the associated data is obtained;

[0007] The association data reflects the association relationship between the location area and the abnormal group for setting the order information;

[0008] At least based on the feature vector, it is identified whether the target transaction order is a risky order.

[0009] Optionally, the method further comprises a step of obtaining the associated data, comprising:

[0010] Get the location area corresponding to the transaction order in the transaction order set when it was generated;

[0011] According to the location area corresponding to each of the transaction orders, an abnormal group for the set order information formed based on the geographical location is obtained;

[0012] An association relationship is established between each of the abnormal groups and the corresponding location area to obtain the association data.

[0013] Optionally, obtaining an abnormal group for setting order information based on a geographical location according to a location area corresponding to each transaction order includes:

[0014] According to the location area corresponding to each of the transaction orders, a location area that meets the set order aggregation requirements is obtained as a target area;

[0015] According to the distribution of the set order information of the transaction orders in each of the target areas, an abnormal group for the set order information formed based on the geographical location is obtained.

[0016] Optionally, obtaining a location area that meets a set order aggregation requirement according to the location area corresponding to each of the transaction orders includes:

[0017] According to the location area corresponding to each of the transaction orders, the location area where the number of transaction orders within the set time window reaches the set number is obtained as the location area that meets the set order aggregation requirement.

[0018] Optionally, obtaining an abnormal group for the set order information formed based on geographical location according to the distribution of the set order information of the transaction orders in each of the target areas includes:

[0019] For each of the target areas, clustering the set order information of the corresponding transaction orders;

[0020] The abnormal group is obtained according to the clustering result for each of the target areas.

[0021] Optionally, the setting order information includes transaction object information, and obtaining the abnormal group according to the clustering result for each of the target areas includes:

[0022] Screening out a first target clustering result satisfying a first distribution state from among all clustering results for transaction objects, wherein the first distribution state is at least two target object classes having different object categories, and the target object class is an object class in which the number of transaction objects reaches the first set number;

[0023] For each of the first target clustering results, aggregation is performed so that the corresponding first target clustering result satisfies the target object class of the first distribution state, and an abnormal group for transaction objects formed based on the corresponding location area is obtained.

[0024] Optionally, the order information includes merchant information, and for each target area, clustering the set order information of the corresponding transaction order includes:

[0025] For each of the target areas, clustering merchants according to the transaction object categories provided by the merchants involved in the corresponding transaction orders;

[0026] The step of obtaining the abnormal group according to the clustering result for each target area includes:

[0027] Among all the clustering results for merchants, a clustering result satisfying a second distribution state is screened out as a second target clustering result, wherein the second distribution state is at least two target merchant classes with different merchant categories, and the target merchant class is a merchant class whose sales volume for the corresponding transaction object category reaches a second set number;

[0028] For each of the second target clustering results, aggregation is performed so that the corresponding second target clustering result satisfies the target merchant class of the second distribution state, and an abnormal group for the merchant formed based on the corresponding location area is obtained.

[0029] Optionally, the order information includes user information, and the clustering of the set order information of the corresponding transaction order for each target area includes:

[0030] For each of the target areas, user clustering is performed according to the transaction object categories required by users of the corresponding transaction orders;

[0031] The obtaining of the abnormal group according to the clustering result for each of the location areas further comprises:

[0032] Among all clustering results for users, a clustering result satisfying a third distribution state is screened out as a third target clustering result, wherein the third distribution state may be a clustering result having at least one target user class, and the target user class is a user class whose sales volume of the corresponding transaction object class reaches a third set number;

[0033] For each of the third target clustering results, aggregation is performed so that the corresponding third target clustering result satisfies the target user class of the third distribution state, and an abnormal group for the user formed based on the corresponding location area is obtained.

[0034] Optionally, the method further comprises:

[0035] Detect the set processing events;

[0036] In response to detecting the triggering of any of the processing events, the step of obtaining the associated data is performed.

[0037] Optionally, the target transaction order is a transaction order in the transaction order set.

[0038] Optionally, obtaining a feature vector reflecting a degree of matching between the target transaction order and the associated data includes:

[0039] For each of the association relationships, according to the matching degree between the target transaction order and the corresponding association relationship, a matching feature value of the target transaction order for the corresponding association relationship is obtained;

[0040] The matching feature values ​​for each of the association relationships are arranged in a set order to obtain the feature vector.

[0041] Optionally, the identifying whether the target transaction order is a risky order at least according to the feature vector includes:

[0042] At least the feature vector is input into a recognition model to obtain a recognition result of whether the target transaction order is a risky order.

[0043] Optionally, the method further comprises a step of generating the recognition model, comprising:

[0044] Obtain standard samples;

[0045] Perturbing at least part of the standard samples to generate adversarial samples;

[0046] The standard sample and the adversarial sample are both used as training samples to train model parameters to obtain the recognition model.

[0047] Optionally, the method further comprises:

[0048] In the case where the target transaction order is identified as a risky order, a set penalty item is executed on at least one of the merchant, commodity and user corresponding to the target transaction order.

[0049] Optionally, executing a set penalty item on at least one of the merchant, commodity, and user corresponding to the target transaction order includes:

[0050] Report the information on the merchant and / or user corresponding to the target transaction order participating in the implementation of the risk order to the credit supervision department.

[0051] According to a second aspect of the present invention, there is also provided an order identification method for an e-commerce platform, comprising:

[0052] Obtaining a feature vector reflecting the degree of matching between the target transaction order and the associated data according to the location area corresponding to when the target transaction order is generated on the e-commerce platform and the set order information of the target transaction order, wherein the set order information includes at least one of product information, merchant information, and user information;

[0053] The association data reflects the association relationship between the location area and the abnormal group for setting the order information;

[0054] At least based on the feature vector, it is identified whether the target transaction order is a risky order.

[0055] According to a third aspect of the present invention, there is also provided an order identification method for a credit housing platform, comprising:

[0056] According to the location area corresponding to the target transaction order when it is generated on the CreditZhu platform and the set order information of the target transaction order, a feature vector reflecting the matching degree between the target transaction order and the associated data is obtained, wherein the set order information includes at least one of room information, merchant information and user information;

[0057] The association data reflects the association relationship between the location area and the abnormal group for setting the order information;

[0058] At least based on the feature vector, it is identified whether the target transaction order is a risky order.

[0059] According to a fourth aspect of the present invention, there is also provided an order identification method, which is implemented by a terminal device and includes:

[0060] Acquire order information of the transaction order, wherein the order information includes geographic location information of the terminal device when the transaction order is generated;

[0061] The order information is sent to a server for risk order identification, wherein the risk order identification comprises: obtaining a location area corresponding to the transaction order when it is generated according to the geographic location information, and obtaining a feature vector reflecting the degree of matching between the transaction order and associated data according to the location area and the set order information of the transaction order; and identifying whether the transaction order is a risk order at least according to the feature vector; the associated data reflects the association relationship between the location area and the abnormal group for the set order information;

[0062] In response to a processing result given by the server according to the positive recognition result, a prompt regarding the processing result is provided.

[0063] According to a fifth aspect of the present invention, there is also provided a credit processing method based on order identification, which is implemented by a terminal device and includes:

[0064] In response to the credit inquiry request, obtaining credit information of the requesting party, wherein the credit information includes information on the requesting party's participation in implementing the risk order;

[0065] A credit inquiry report is generated based on the acquired credit information and fed back to the requesting party.

[0066] According to a sixth aspect of the present invention, there is also provided an order identification device, comprising:

[0067] A matching module, used to obtain a feature vector reflecting the matching degree between the target transaction order and the associated data according to the location area corresponding to the target transaction order when it is generated and the set order information of the target transaction order;

[0068] The association data reflects the association relationship between the location area and the abnormal group for setting the order information;

[0069] An identification module is used to identify whether the target transaction order is a risky order based at least on the feature vector.

[0070] According to a seventh aspect of the present invention, there is further provided an electronic device, comprising the identification device according to the fifth aspect of the present invention, or the electronic device comprises:

[0071] A memory for storing executable commands;

[0072] A processor is used to execute the method described in any aspect of the first to fifth aspects of the present invention under the control of the executable command.

[0073] According to a seventh aspect of the present invention, there is further provided a computer-readable storage medium storing executable instructions, wherein when the executable instructions are executed by a processor, the method described in any one of the first to fifth aspects of the present invention is executed.

[0074] One beneficial effect of the present invention is that the method of this embodiment identifies whether the target transaction order is a risky order based on the matching of the target transaction order with each association relationship in the associated data. Here, since each association relationship is formed based on the geographical location, the method according to this embodiment can effectively identify fake orders involving real users that highly simulate normal transaction orders, thereby significantly improving the accuracy and timeliness of identifying risky orders.

[0075] Further features and advantages of the present invention will become apparent from the following detailed description of exemplary embodiments of the present invention with reference to the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0076] The accompanying drawings, which are incorporated in and constitute a part of the specification, illustrate embodiments of the invention and, together with the description, serve to explain the principles of the invention.

[0077] Figure 1a A schematic diagram of a scenario for implementing the method of the embodiment of the present invention;

[0078] Figure 1b A hardware configuration structure diagram of a risk control system that can be used to implement the order identification method of an embodiment of the present invention;

[0079] Figure 2 A schematic diagram showing a flow chart of an order identification method according to an embodiment of the present invention is shown;

[0080] Figure 3 A schematic diagram showing a flow chart of an order identification method according to another embodiment of the present invention;

[0081] Figure 4 A schematic diagram showing a flow chart of an order identification method according to a third embodiment of the present invention;

[0082] Figure 5 A schematic diagram showing a flow chart of an order identification method according to a fourth embodiment of the present invention;

[0083] Figure 6 A schematic diagram of an order identification device according to an embodiment of the present invention is shown;

[0084] Figure 7 A schematic diagram of an electronic device according to an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0085] Various exemplary embodiments of the present invention will now be described in detail with reference to the accompanying drawings. It should be noted that the relative arrangement of components and steps, numerical expressions and numerical values ​​set forth in these embodiments do not limit the scope of the present invention unless otherwise specifically stated.

[0086] The following description of at least one exemplary embodiment is merely illustrative in nature and is in no way intended to limit the invention, its application, or uses.

[0087] Technologies, methods, and apparatuses known to persons of ordinary skill in the relevant art may not be discussed in detail, but where appropriate, such technologies, methods, and apparatuses should be considered part of the specification.

[0088] In all examples shown and discussed herein, any specific values ​​should be interpreted as merely exemplary and not limiting. Therefore, other examples of the exemplary embodiments may have different values.

[0089] It should be noted that like reference numerals and letters refer to similar items in the following figures, and therefore, once an item is defined in one figure, it need not be further discussed in subsequent figures.

[0090] <Hardware Configuration>

[0091] With the rapid development of the black industry that relies on online transactions, there is now an industry of fake orders that can highly simulate normal orders and are participated in by real users. Due to the high concealment of such fake orders, the existing risk control platform cannot effectively identify the risky orders generated by such fake orders. Figure 1a As shown, this type of order brushing has the following characteristics, namely, the locations of the orders are concentrated when they are generated, and a large number of transactions of unrelated transaction objects (such as goods) occur in a specific location area and in a relatively compact time. For example, a large number of transaction orders distributed in the east gate area of ​​** School are generated in a relatively compact time, and these transaction orders involve transactions of unrelated goods or merchants. The embodiment of the present invention is based on the above-mentioned characteristics of this type of order brushing, and adds feature vectors reflecting the similarity between the order and these characteristics when performing order identification, thereby greatly improving the probability of identifying ground-based order brushing as risky orders, and improving the risk control platform's recognition accuracy of risky orders.

[0092] See also Figure 1a As shown, according to an embodiment of the present invention, after users generate transaction orders through their respective terminal devices 1200, any transaction order may be a normal order or a false order. The terminal device 1200 needs to upload the order information to the server for background processing according to the settings. The order information includes the geographical location information of the terminal device when the transaction order is generated, the transaction object (for example, e-commerce products, credit cards, face-to-face payments, etc.), the order issuer information (also referred to as user information) and the order recipient information (also referred to as merchant information), etc. After obtaining the order information uploaded by the terminal device 1200, the server will process the order information. Data processing is performed to extract the feature vector of each transaction order and input it into the recognition model. The recognition model is pre-trained with sample data. The recognition model outputs a recognition result indicating whether the corresponding transaction order is a risky order based on the input feature vector. Furthermore, the server can feedback the processing result to the associated electronic device 2000 based on the recognition result that the corresponding transaction order is a risky order. The electronic device 2000 may include a terminal device of the corresponding merchant, a terminal device of the corresponding user, and a terminal device or server of an administrative functional department such as a corresponding credit supervision department (for example, the Ministry of Public Security, a credit reporting center, etc.), which is not limited here.

[0093] Figure 1bThe present invention is a schematic diagram of the structure of a risk control system to which the order identification method according to an embodiment of the present invention can be applied.

[0094] As shown in FIG. 1 , the risk control system 1000 of this embodiment includes a server 1100 , a terminal device 1200 , and a network 1300 .

[0095] The server 1100 may be, for example, a blade server, a rack server, etc. The server 1100 may also be a server cluster deployed in the cloud, which is not limited here. The server may be a server providing online transaction platform services, or a server of the above administrative functional departments, which is not limited here.

[0096] As shown in FIG1 , the server 1100 may include a processor 1110, a memory 1120, an interface device 1130, a communication device 1140, a display device 1150, and an input device 1160. The processor 1110 may be, for example, a central processing unit CPU, etc. The memory 1120 may include, for example, a ROM (read-only memory), a RAM (random access memory), a non-volatile memory such as a hard disk, etc. The interface device 1130 may include, for example, a USB interface, a serial interface, etc. The communication device 1140 may be, for example, capable of wired or wireless communication. The display device 1150 may be, for example, a liquid crystal display screen. The input device 1160 may include, for example, a touch screen, a keyboard, etc.

[0097] In this embodiment, the server 1100 may be used to participate in implementing the order identification method according to any embodiment of the present invention.

[0098] In the embodiment of the present invention, the memory 1120 of the server 1100 is used to store instructions, and the instructions are used to control the processor 1110 to operate to support the implementation of the order identification method according to any embodiment of the present invention. A technician can design instructions according to the scheme disclosed in the present invention. How instructions control the processor to operate is well known in the art, so it will not be described in detail here.

[0099] Those skilled in the art should understand that, although multiple devices of the server 1100 are shown in FIG. 1 , the server 1100 of an embodiment of the present invention may involve only some of the devices, for example, only the processor 1110 and the memory 1120 .

[0100] As shown in FIG. 1 , the terminal device 1200 may include a processor 1210, a memory 1220, an interface device 1230, a communication device 1240, a display device 1250, an input device 1260, an audio output device 1270, an audio input device 1280, and the like. The processor 1210 may be a central processing unit CPU, a microprocessor MCU, and the like. The memory 1220 may include, for example, a ROM (read-only memory), a RAM (random access memory), a non-volatile memory such as a hard disk, and the like. The interface device 1230 may include, for example, a USB interface, an earphone interface, and the like. The communication device 1240 may, for example, be capable of wired or wireless communication. The display device 1250 may, for example, be a liquid crystal display screen, a touch display screen, and the like. The input device 1260 may, for example, include a touch screen, a keyboard, and the like. The terminal device 1200 may output audio information through an audio output device 1270, and the audio output device 1270 may, for example, include a speaker. The terminal device 1200 may pick up voice information input by a user through an audio pickup device 1280, and the audio pickup device 1280 may, for example, include a microphone.

[0101] The terminal device 1200 can be any device that can support the use of e-commerce platform applications, such as a smart phone, a portable computer, a desktop computer, a tablet computer, etc.

[0102] In this embodiment, the terminal device 1200 can be used to obtain the geographic location information of the transaction order when the transaction order is generated, and send the geographic location information together with other order information to the server 1100, so that the server 1100 can identify whether it is a risky order based on the order information of the transaction order.

[0103] In the embodiment of the present invention, the memory 1220 of the terminal device 1200 is used to store instructions, and the instructions are used to control the processor 1210 to operate to support the implementation of the order identification method according to any embodiment of the present invention. The technician can design instructions according to the scheme disclosed in the present invention. How the instructions control the processor to operate is well known in the art, so it will not be described in detail here.

[0104] Those skilled in the art should understand that, although multiple devices of the terminal device 1200 are shown in FIG1 , the terminal device 1200 of an embodiment of the present invention may involve only some of the devices, for example, only the processor 1210, the memory 1220, the display device 1250, the input device 1260, etc.

[0105] The communication network 1300 may be a wireless network or a wired network, a local area network or a wide area network. The terminal device 1200 may communicate with the server 1100 via the communication network 1300 .

[0106] The risk control system 1000 shown in FIG1 is merely illustrative and is by no means intended to limit the present invention, its applications or uses. For example, although FIG1 only shows one server 1100 and one terminal device 1200, it does not mean to limit the number of each, and the risk control system 1000 may include multiple servers 1100 and / or multiple terminal devices 1200.

[0107] <Method Example>

[0108] This embodiment provides an order identification method, which can be implemented by the server in FIG1 . Figure 2 As shown, the method includes the following steps S1100-S1200:

[0109] Step S1100, obtaining a feature vector reflecting the degree of matching between the target transaction order and the associated data according to the location area corresponding to the target transaction order when it is generated and the set order information of the target transaction order; wherein the associated data reflects the association relationship between the location area and the abnormal group for the set order information.

[0110] The transaction order in this embodiment is: an online order generated when a buyer purchases goods through an online platform.

[0111] The online platform may be, for example, an e-commerce platform, a platform that provides credit accommodation, or a platform that provides face-to-face payment, etc., and is not limited here.

[0112] If the online platform is an e-commerce platform, the transaction object of the transaction order is the commodity. If the online platform is a platform that provides credit accommodation, the transaction object of the transaction order is the room. If the online platform is a platform that provides face-to-face payment, the transaction object of the transaction order is virtual currency.

[0113] In this embodiment, the location area corresponding to the transaction order when it is generated can be determined based on the geographical location information of the corresponding terminal device when the transaction order is generated.

[0114] The geographical location information of the corresponding terminal device when the transaction order is generated can be provided by the positioning device of the terminal device, or can be determined based on the IP address of the terminal device when the transaction order is generated, which is not limited here. Correspondingly, when the buyer generates a transaction order through the terminal device, the terminal device will obtain its own geographical location information from the positioning device or based on the current IP address, and upload the geographical location information as part of the order information of the transaction order to the server, and the server can determine the location area corresponding to the transaction order when it is generated based on the geographical location information.

[0115] In this embodiment, each location area can be obtained by dividing according to a set scale, such as dividing according to a street scale, or dividing according to a grid scale, etc. For the grid, the grid size can be preset, such as setting a grid of 100m x 100m, and the target geographical range is divided according to the grid size to obtain each location area.

[0116] In one embodiment, the method of the present invention may further include, before step S1100: obtaining geographic location information corresponding to the target transaction order when it is generated, and determining the location area corresponding to the target transaction order when it is generated based on the geographic location information.

[0117] In this embodiment, the terminal device may upload the geographic location information as a part of the order information to the server so that the server can determine the location area corresponding to the target transaction order when it is generated.

[0118] In this embodiment, for a transaction order generated on an e-commerce platform, the set order information may include, for example, any one or more of product information, merchant information, and user information. The set order information may also include other order information, which is not limited here.

[0119] In this embodiment, for a transaction order generated on a platform providing credit accommodation, the set order information may include, for example, any one or more of room information, merchant information, and user information. The set order information may also include other order information, which is not limited here.

[0120] In this embodiment, the abnormal group for setting order information is an abnormal group for setting order information based on the geographical location, so an association relationship can be generated between the location area and the abnormal group. This means that for any abnormal group in the associated data, all members in the same group are associated with the same location area. For example, for an abnormal group for transaction objects in the associated data, the location area corresponding to the transaction order of each transaction object in the same group when it is generated is the same location area, that is, each transaction object in the same group is associated with the same location area.

[0121] In this embodiment, the abnormal group may include at least one or more of an abnormal group for transaction objects formed based on geographic location, an abnormal group for merchants formed based on geographic location, and an abnormal group for users formed based on geographic location.

[0122] In the case where the abnormal group includes an abnormal group for a transaction object, the associated data includes an associated relationship between the abnormal group for the transaction object and the location area, which is called a first associated relationship. For example, the first abnormal group for the transaction object corresponds to the first location area, that is, the first abnormal group for the transaction object has an associated relationship with the first location area. Taking the transaction object as a commodity as an example, the associated data includes an associated relationship between the abnormal group for the commodity and the location area as the first associated relationship. In the case where the abnormal group includes an abnormal group for a merchant, the associated data includes an associated relationship between the abnormal group for the merchant and the location area, which is called a second associated relationship. For example, the first abnormal group for the merchant corresponds to the first location area, that is, the first abnormal group for the merchant has an associated relationship with the first location area.

[0123] In the case where the abnormal group includes an abnormal group for the user, the associated data includes an associated relationship between the abnormal group for the user and the location area, which is called a third associated relationship. For example, the first abnormal group for the user corresponds to the first location area, that is, the first abnormal group for the user has an associated relationship with the first location area.

[0124] When the abnormal group includes at least two of an abnormal group for transaction objects, an abnormal group for merchants, and an abnormal group for users, these abnormal groups may also be associated with each other based on corresponding to the same location area.

[0125] In this embodiment, since the feature vector reflects the matching degree between the target transaction order and the associated data, the feature vector is also a feature vector related to the geographical location.

[0126] In one example, obtaining a feature vector reflecting the degree of matching between the target transaction order and the associated data may include the following steps S1110 and S1120:

[0127] Step S1110: for each association relationship, according to the matching degree between the target transaction order and the corresponding association relationship, a matching feature value of the target transaction order for the corresponding association relationship is obtained.

[0128] In this embodiment, the feature value of a successful match can be set to 1, and the feature value of a failed match can be set to 0, so that a feature vector composed of 0 and / or 1 can be finally obtained.

[0129] For example, it can be determined whether the first location area corresponding to the target transaction order when it is generated is a location area with an abnormal group. If not, the matching feature value for each association relationship will be 0; if yes, the next matching operation is performed, which may include any one or more of the following:

[0130] The first item: whether the transaction object or transaction object category of the target transaction order belongs to the abnormal group for transaction objects corresponding to the first location area. If so, it means that the target transaction order is successfully matched with the first association relationship, and the corresponding matching feature value is 1; if not, it means that the target transaction order is unsuccessful in matching with the first association relationship, and the corresponding matching feature value is 0.

[0131] The second item: whether the merchant of the target transaction order belongs to the abnormal group of merchants corresponding to the first location area. If so, it means that the target transaction order is successfully matched with the second association relationship, and the corresponding matching feature value is 1; if not, it means that the target transaction order is unsuccessful in matching with the second association relationship, and the corresponding matching feature value is 0.

[0132] The third item: whether the user of the target transaction order belongs to the abnormal group of users corresponding to the first location area. If so, it means that the target transaction order is successfully matched with the third association relationship, and the corresponding matching feature value is 1; if not, it means that the target transaction order is unsuccessful in matching with the third association relationship, and the corresponding matching feature value is 0.

[0133] In addition, in this embodiment, a matching feature value may be added to reflect whether the first location area is a location area with an abnormal group, that is, the added matching feature value is included in the feature vector finally obtained.

[0134] Step S1120 , arranging the matching feature values ​​for each association relationship according to a set order to obtain the feature vector.

[0135] In step S1120, for example, the feature vector is a multidimensional vector of n×1, where n represents the number of matching feature values. When the associated data includes a first association relationship, a second association relationship, and a third association relationship, the value of n is greater than or equal to 3.

[0136] Step S1200, identifying whether the target transaction order is a risky order based at least on the feature vector obtained in step S1100.

[0137] In this embodiment, mapping data reflecting the correspondence between the input vector and the recognition result can be pre-set, wherein the input vector includes a feature vector, so as to identify whether the target transaction order is a risky order based on at least the feature vector obtained in step S1100 and the mapping data.

[0138] In this embodiment, in addition to the above feature vectors, the input vector may also include features related to user behavior, merchant timing, transaction object attributes, etc., which are not limited here.

[0139] In one embodiment, identifying whether the target transaction order is a risky order in step S1200 at least based on the feature vector obtained in step S1100 may include: inputting at least the feature vector obtained in step S1100 into a recognition model to obtain a recognition result of whether the target transaction order is a risky order through the recognition model.

[0140] The recognition model can be various neural network models, such as a deep neural network model DNN, which is not limited here. The recognition model is pre-trained based on sample data for the purpose of identifying risky orders.

[0141] According to the above steps S1100 and S1200, the method of this embodiment identifies whether the target transaction order is a risky order based on the matching of the target transaction order with each association relationship in the associated data. Here, since each association relationship is formed based on the geographical location, the method according to this embodiment can effectively identify fake orders participated by real users that highly simulate normal transaction orders, such as the highly concealed fake orders promoted by ground users on e-commerce platforms, which significantly improves the accuracy and timeliness of risky order identification.

[0142] When the server identifies a target transaction order as a risky order according to the method of this embodiment, the server may execute a set penalty item on at least one of the merchant, transaction object and user corresponding to the target transaction order. For example, the penalty item may include lowering the merchant's level; for another example, the penalty item may include demoting the transaction object so that the recommendation weight of the transaction object is reduced; for another example, the penalty item may include lowering the user's level, or closing the user's account, etc., which are not limited here.

[0143] In one embodiment, executing the set penalty item on at least one of the merchant, transaction object and user corresponding to the target transaction order may also include: reporting information on the merchant and / or user corresponding to the target transaction order participating in the implementation of the risk order to the credit supervision department.

[0144] The credit supervision department is, for example, the Ministry of Public Security, a credit or credit reporting center, or other departments that are qualified to provide personal or corporate credit inquiries.

[0145] This embodiment may further include: when the number of merchants and / or users corresponding to the target transaction order participating in the implementation of the risk order reaches a set number, the information on the merchants and / or users corresponding to the target transaction order participating in the implementation of the risk order is reported to the credit supervision department.

[0146] According to this embodiment, when the online trading platform reports the information of merchants and / or users participating in the implementation of risky orders to the credit supervision department, the behavior of merchants and / or users participating in the implementation of risky orders will affect the credit rating of the corresponding merchants and / or users, thereby increasing the penalty for merchants and / or users participating in the implementation of risky orders.

[0147] Figure 3 A schematic flow chart of an order identification method according to another embodiment of the present invention is shown.

[0148] according to Figure 3 As shown, in this embodiment Figure 2 Based on the illustrated embodiment, an alternative method for obtaining the above associated data is provided, and the method includes the following steps S3100 to S3300:

[0149] Step S3100, obtaining the location area corresponding to each transaction order in the transaction order set when it is generated.

[0150] The implementation of obtaining the location area corresponding to each transaction order in the transaction order set when it is generated can be the same as the implementation of obtaining the location area corresponding to the target transaction order when it is generated, which will not be repeated here.

[0151] In this embodiment, the transaction order set obtained in step S3100 may include all transaction orders generated when the acquisition operation is performed; it may also include transaction orders generated within a set time range, for example, all transaction orders generated on the day or week when the acquisition operation is performed, or for example, transaction orders generated within a set time length from the current time when the acquisition operation is performed, etc., which are not limited here.

[0152] The above target transaction order may belong to the transaction order set or may not belong to the transaction order set, which is not limited here. In the case where the target transaction order belongs to the transaction order set, the identification of the target transaction order will be more targeted and thus more accurate.

[0153] Step S3200, according to the location area corresponding to each transaction order, obtain an abnormal group for the set order information formed based on the geographical location.

[0154] The set order information may include, for example, any one or more of transaction object information, merchant information, and buyer information. The set order information may also include other order information, which is not limited here.

[0155] In the present embodiment, an abnormal group for setting order information is formed based on the geographic location, that is, for any abnormal group obtained, all members in the same group are associated with the same location area. For example, for the abnormal group of transaction objects obtained, the location area corresponding to the transaction order of each transaction object in the same group when it is generated is the same location area, that is, each transaction object in the same group is associated with the same location area.

[0156] In order to reduce the amount of data processing for obtaining abnormal groups and speed up the processing and response speed for risk order identification, for all transaction orders obtained in step S3100, the concentrated transaction orders can be first screened out according to their corresponding location areas as high-risk orders, and further, according to the location areas corresponding to the high-risk orders, abnormal groups for set order information formed based on geographical locations are obtained. Therefore, in one example, in step S3200, according to the location areas corresponding to each transaction order, abnormal groups for set order information formed based on geographical locations are obtained, which can include the following steps S3211 and S3212:

[0157] Step S3211, based on the location area corresponding to each transaction order, obtain the location area that meets the set order concentration distribution requirements as the target area.

[0158] The order concentration distribution requirement may be, for example: the number of trading orders within a set time window reaches a set number. In this embodiment, "reach" means greater than or equal to.

[0159] The set time window may be in hours or other time units, which are not limited here. For example, the set time window may be 2 hours, 4 hours, etc.

[0160] In view of the above order concentration distribution requirements, step S3211 may be: according to the location area corresponding to each transaction order, obtain the location area where the number of transaction orders within the set time window reaches the set number as the target area.

[0161] The set quantity can be determined based on a reasonable number of transaction orders that may be generated in the same location area within a set time window. For example, the set quantity can be a value greater than the reasonable number, which is not limited here.

[0162] In the process of executing step S3211, for any location area, the corresponding transaction orders can be arranged in the order of generation time to obtain an order sequence, and then slide on the order sequence by setting a time window. If there is at least one sliding position that meets the order concentration distribution requirements, then the location area meets the order concentration distribution requirements.

[0163] Step S3212, according to the distribution of the set order information of the transaction orders in each target area, obtain an abnormal group for the set order information formed based on the geographical location.

[0164] In this example, the trading orders in the target area are also the target orders that may become risky orders.

[0165] For example, if the order information is set to include transaction object information, obtaining abnormal groups for the set order information based on geographical locations according to the distribution of the set order information of the transaction orders in each target area in step S3212 may include: obtaining abnormal groups for the transaction objects based on geographical locations according to the distribution of the transaction objects of the transaction orders in each target area.

[0166] For another example, the set order information includes merchant information. In step S3212, according to the distribution of the set order information of the transaction orders in each target area, obtaining the abnormal group for the set order information based on the geographical location may include: according to the distribution of merchants of the transaction orders in each target area, obtaining the abnormal group for the merchants based on the geographical location.

[0167] For another example, the set order information includes user information. In step S3212, according to the distribution of the set order information of the transaction orders in each target area, obtaining the abnormal group for the set order information based on the geographical location may include: according to the distribution of users of the transaction orders in each target area, obtaining the abnormal group for the users based on the geographical location.

[0168] The step S3212 may perform cluster analysis on the set order information of the transaction orders in the target area to obtain an abnormal group for the set order information formed based on the geographical location. Therefore, in one example, the step S3212 may include: clustering the set order information of the transaction orders corresponding to each target area; and obtaining the abnormal group according to the clustering result for each target area.

[0169] In this example, the set order information may include one order information or multiple order information. For a specific information in the set order information, for example, for transaction object information, one target area may correspond to one clustering result, and one clustering result may include multiple clusters.

[0170] For example, if the order information is set to include transaction object information, merchant information and user information, then in this example, each of the information can be clustered separately to obtain three clustering results for each target area, namely, the clustering result for the transaction object, the clustering result for the merchant and the clustering result for the user, and each clustering result contains multiple classes generated in the cluster. Taking the transaction object as a commodity as an example, the clustering results for the transaction object may include dresses, home improvement building materials, men's casual pants, sunglasses, etc.

[0171] In this example, for any target area, clustering the set order information of the corresponding transaction orders can be: dividing the set order information of the corresponding transaction orders into multiple classes composed of similar objects. For example, the set order information includes transaction object information, and the clustered transaction object classes include dress class, home improvement building materials class, etc.

[0172] In this example, when clustering, the determination of similar objects can not only require similarity in the category of specific information, but also require similarity in the corresponding order generation time. The time similarity can be, for example, that the corresponding order generation time is concentrated in a certain time period. In this way, for the transaction objects in a transaction object class obtained by clustering, both the object type and the corresponding transaction generation time will have similarity. For example, in an object class obtained by clustering, the object type is a dress class, and the corresponding order generation time is concentrated in the afternoon time period from 14:00 to 16:00.

[0173] In this example, it can be determined whether it is necessary to add time similarity for clustering according to the time length of the transaction order obtained in step S3100. When the time length of the transaction order obtained in step S3100 is short, for example, the transaction order generated on the day when the operation is performed is obtained, the obtained transaction order itself has the time similarity feature for setting order information, so clustering can be directly performed according to the category similarity of specific information.

[0174] In this example, it is set that the order information may include transaction object information such as commodity information, and the transaction objects involved in the corresponding transaction order may be clustered for each target area. Further, when obtaining an abnormal group based on the clustering result for each target area, the following steps S3212-A1 and S3212-A2 may be included:

[0175] Step S3212-A1, among all clustering results for transaction objects, screen out a first target clustering result that satisfies a first distribution state, wherein the first distribution state is at least two target object classes with different transaction object categories, and the target object class is an object class in which the number of transaction objects reaches a first set number.

[0176] For example, a first target clustering result selected has two target object classes with different object categories, namely, a dress class and a home decoration building materials class.

[0177] In this example, if time similarity is added to cluster transaction objects, the first distribution state may also be at least two target object classes with different object categories but the same or similar time categories. The same time category may mean that the time categories to which these target object classes belong are the same, that is, the corresponding time ranges are the same. The similar time categories may mean that the overlap of the time ranges corresponding to these target object classes reaches a set value.

[0178] Step S3212-A2: for each first target clustering result, aggregate the target object class corresponding to the first target clustering result to satisfy the first distribution state, and obtain an abnormal group for the transaction object formed based on the corresponding location area.

[0179] For example, according to step S3212-A1, it is determined that the clustering result corresponding to the first position area includes the first target clustering result, so that the clustering result satisfies the first distribution state and thus becomes the target object class of the first target clustering result, including the dress class and the home improvement building materials class. Therefore, according to step S3212-A2, it can be obtained that the abnormal group for the transaction object formed based on the first position area includes the dress class and the home improvement building materials class.

[0180] In this example, since a large number of transactions of unrelated transaction objects occur in the same location area, for example, the location area including the east gate of a certain university in a certain city, and the location area including the entrance of a specific subway station in a certain city, and the probability of time compactness is extremely low, the first distribution state reflects the characteristics of concentrated order brushing. Establishing the first association relationship based on the first distribution state will enable concentrated order brushing involving real users to be identified in terms of the transaction object dimension.

[0181] In this example, it is set that the order information can include merchant information, and then for each target area, the merchants involved in the corresponding transaction order can be clustered, for example, the merchants are clustered according to the transaction object categories provided by the merchants, that is, the transaction object categories provided by all merchants in a merchant class should be the same. Further, when obtaining an abnormal group based on the clustering results for each target area, the following steps S3212-B1 and S3212-B2 can be included:

[0182] Step S3212-B1, among all clustering results for merchants, screen out clustering results that satisfy the second distribution state as second target clustering results, wherein the second distribution state is at least two target merchant classes with different merchant categories, and the target merchant class is a merchant class whose sales volume for the corresponding transaction object category reaches a second set number.

[0183] For example, a second target clustering result selected has two target merchant classes with different merchant categories, namely, a merchant class selling dresses and a merchant class selling home improvement building materials.

[0184] In this example, if time similarity is added to perform merchant clustering, the second distribution state may also be at least two target merchant classes with different merchant categories but the same or similar time categories. The same time category may mean that the time categories to which these target merchant classes belong are the same, that is, the generation time of the transaction orders to which these target merchant classes belong is within the same time range. The similar time category may mean that the overlap of the time ranges corresponding to these target object classes reaches a set value.

[0185] Step S3212-B2: for each second target clustering result, aggregate the target merchant class corresponding to the second target clustering result to satisfy the second distribution state, and obtain an abnormal group for the merchant formed based on the corresponding location area.

[0186] For example, according to step S3212-B1, it is determined that the clustering result corresponding to the first location area includes the second target clustering result, so that the clustering result satisfies the second distribution state and thus becomes the target merchant class of the second target clustering result, including the merchant class selling dresses and the merchant class selling home improvement building materials. Therefore, according to step S3212-B2, it is possible to obtain an abnormal group for merchants formed based on the first location area, including the merchant class selling dresses and the merchant class selling home improvement building materials.

[0187] Similar to the first distribution state, the second distribution state also reflects the characteristics of concentrated order brushing involving real users. Therefore, by establishing a second association relationship based on the second distribution state, it will be possible to identify this type of order brushing at the merchant dimension, such as identifying offline order brushing.

[0188] In this example, it is set that the order information can include user information, and then for each target area, the users involved in the corresponding transaction order can be clustered, for example, the users are clustered according to the category of the transaction object purchased by the user. Further, when obtaining the abnormal group according to the clustering result for each target area, the following steps S3212-C1 and S3212-C2 can be included:

[0189] Step S3212-C1, among all clustering results for users, screen out clustering results that satisfy the third distribution state as the third target clustering results, wherein the third distribution state may be a clustering result having at least one target user class, the target user class being a user class whose sales volume of the corresponding transaction object category reaches a third set number, and the target user class may further correspond to at least two object categories, that is, users in the target user class have all purchased goods of the at least two object categories.

[0190] In the case where the order information is set to also include transaction object information, the transaction object category corresponding to the target user class may be an object category of a target object class corresponding to the same target area.

[0191] For example, in a filtered clustering result, there is a user class that purchased both a dress and home improvement building materials, and the sales volume of the dress and home improvement building materials in the corresponding location area both reached the third set quantity. Therefore, this user class is the target user class, and the clustering result is the third target clustering result.

[0192] Step S3212-C2: for each third target clustering result, aggregate the target user class corresponding to the third target clustering result to satisfy the third distribution state, and obtain an abnormal group for the user formed based on the corresponding location area.

[0193] Similar to the first distribution state, the third distribution state also reflects the characteristics of concentrated order brushing. Therefore, by establishing a third association relationship based on the third distribution state, it will be possible to identify concentrated order brushing involving real users in the user dimension.

[0194] Step S3300 , establishing an association relationship between each abnormal group and the corresponding location area, and obtaining the association data required to be used in step S1100 .

[0195] In this embodiment, the established association relationship may include one or more of the above-mentioned first association relationship, second association relationship and third association relationship.

[0196] In this embodiment, after obtaining the associated data, an operation of identifying whether each transaction order in the order set is a risky order may be performed.

[0197] According to the above steps S3100 to S3300, the method of this embodiment takes each location area where a large number of transaction orders in the transaction order set are located as a unit, analyzes the distribution of the set order information of the transaction orders in each location area, and obtains the abnormal group for the set order information based on the geographical location based on the distribution, and then establishes a corresponding association relationship between the abnormal group and the location area, and obtains the association data that conforms to the characteristics of concentrated order brushing participated by real users, and then when performing order risk identification, it can effectively identify the transaction orders belonging to concentrated order brushing (such as ground order brushing) participated by real users by adding the above feature vectors.

[0198] In one embodiment, the method of the present invention may further include the following steps: detecting a set processing event; and, in response to detecting the triggering of any processing event, executing Figure 3 The operation of obtaining associated data in the illustrated embodiment.

[0199] In this embodiment, at least one processing event may be set. The at least one processing event may include, for example, the expiration of a set time, the receipt of an external instruction, etc., which are not limited here.

[0200] According to this embodiment, a processing event for triggering the execution of the operation of obtaining associated data may be preset, so that the operation of obtaining associated data is automatically executed when the processing event is detected, and a risk order identification operation based on the associated data is further executed.

[0201] In one embodiment, the method of the present invention may further include the following steps: providing a setting entry; acquiring and saving any one or more of the above set quantity values ​​input through the setting entry.

[0202] In another embodiment, the numerical value of at least one of the above set quantities may also adopt a default fixed value, which is not limited here.

[0203] According to the method of this embodiment, the operator can be supported to adjust the values ​​of various set quantities through the preset setting entrance according to the actual situation, thereby improving the flexibility of the application.

[0204] In one embodiment, identifying whether the target transaction order is a risky order based at least on the feature vector in the above step S1200 may include: inputting at least the feature vector into a recognition model to obtain a recognition result of whether the target transaction order is a risky order.

[0205] In this embodiment, the input vector input to the recognition model, in addition to the above-mentioned feature vector reflecting the matching degree between the target transaction order and the related data, may also include features of other dimensions, for example, it may also include features related to user behavior, merchant timing, transaction object attributes, etc.

[0206] The method of this embodiment also includes the step of generating the recognition model, which may include: obtaining standard samples; perturbing at least part of the standard samples to generate adversarial samples; and using both the standard samples and the adversarial samples as training samples to train model parameters to obtain the recognition model.

[0207] In this embodiment, normal samples include positive samples and negative samples. Positive samples are samples belonging to risky orders, negative samples are samples that do not belong to risky orders, and normal samples are order samples used in traditional model training.

[0208] For traditional recognition models, if the order-brushing operator artificially disturbs the specific dimensional features of the model, for example, disturbing the user behavior click feature, increasing the behavior click frequency before the transaction occurs, disturbing the user browsing time feature, and / or increasing the user browsing time before the transaction occurs, etc., the recognition accuracy and recall rate of the recognition model for risky orders will drop significantly. Therefore, in this embodiment, the perturbed samples are added to participate in the model training to improve the robustness and robustness of the recognition model, so that risky orders can be accurately identified despite the artificial disturbance added by the order-brushing operator.

[0209] In this embodiment, the adversarial sample may include adversarial samples that perturb the above matching feature values, and / or adversarial samples that perturb other features in the input vector, for example, including adversarial samples that perturb user behavior, etc., which are not limited here.

[0210] In this embodiment, adversarial samples may be generated based on normal samples and based on an iterative gradient signal algorithm, or may be generated based on other algorithms, which are not limited here.

[0211] In the step of training the model parameters through training samples to obtain the recognition model in this embodiment, the training samples can be firstly processed to extract the input vector of each training sample, where the input vector includes a feature vector reflecting the degree of match between the target transaction order and the associated data, and can also include features related to user behavior, merchant timing, transaction object attributes, etc., which are not limited here; thereafter, the parameters of the model can be trained and set according to the actual results of the extracted input vectors of the training samples and whether the training samples belong to risky orders, thereby obtaining a recognition module for use in online order recognition.

[0212] Figure 4 It is a flowchart of an order identification method according to another embodiment of the present invention. This embodiment is implemented by a terminal device, for example, by a terminal device 1200 that logs in to a user account.

[0213] like Figure 4As shown, the method of this embodiment may include the following steps S4100 to S4200:

[0214] Step S4100, obtaining order information of the transaction order, wherein the order information includes geographic location information of the terminal device when the transaction order is generated.

[0215] In this embodiment, when a transaction order is generated, the terminal device obtains order information of the transaction order and sends it to the server.

[0216] The geographic location information may be provided by a positioning device of the terminal device, such as a GPS positioning device of the terminal device, or may be determined based on an IP address of the terminal device when the transaction order is generated, which is not limited here.

[0217] The order information may also include transaction object information, merchant information, user information, etc.

[0218] Step S4200, sending the order information to the server for risky order identification.

[0219] In this embodiment, the server's risk order identification may include: obtaining the location area corresponding to the transaction order when it is generated based on the received geographic location information, and obtaining a feature vector reflecting the degree of matching between the transaction order and the associated data based on the location area and the set order information of the transaction order; and, at least based on the feature vector, identifying whether the transaction order is a risky order.

[0220] In this embodiment, the association data reflects the association relationship between the location area and the abnormal group for setting the order information.

[0221] The steps for the above server to identify risky orders can be found in the above method embodiments and will not be repeated here.

[0222] When the server obtains the identification result that the transaction order is a risky order, it will execute the set penalty item on at least one of the merchant, transaction object and user corresponding to the transaction order, and feedback the corresponding processing result to the corresponding terminal device, for example, feedback the processing result for the merchant and / or product to the terminal device logged in to the merchant account, and feedback the processing result for the user to the terminal device logged in to the user account.

[0223] Step S4300, in response to the processing result given by the server according to the positive recognition result, a prompt is given regarding the processing result.

[0224] In this embodiment, for the terminal device that logs in the user account, a prompt about the corresponding processing result can be given so that the user can know the corresponding processing result, and a complaint entrance can be further provided so that the user can file a complaint to the e-commerce platform according to the prompt.

[0225] Figure 5 It is a flowchart of a credit processing method based on order identification according to another embodiment of the present invention. In this embodiment, for example, the credit processing method is implemented by an electronic device of a corresponding credit supervision department.

[0226] The electronic device may be a terminal device corresponding to the credit supervision department, or may be a server corresponding to the credit supervision department, which is not limited here.

[0227] like Figure 5 As shown, the credit processing method of this embodiment may include the following steps S5100 and S5200:

[0228] Step S5100, in response to the credit inquiry request, obtaining the credit information of the requesting party, wherein the credit information includes information on the requesting party's participation in implementing the risk order.

[0229] The requesting party may be an individual or an enterprise, and there is no limitation here.

[0230] In this embodiment, the credit inquiry request may be triggered via a terminal device of the corresponding credit supervision department, and the terminal device may obtain information on the requesting party's participation in implementing risk orders from a server of the corresponding credit supervision department or a server of the corresponding online trading platform.

[0231] In this embodiment, the credit inquiry request may also be triggered via the terminal device of the requester, wherein the terminal device of the requester is installed with a credit inquiry application provided by the credit supervision department, or can log in to the web application of the credit supervision department. In response to the credit inquiry request, the server of the credit supervision department obtains information on the requester's participation in the implementation of risk orders from its own database or from the server of the corresponding online trading platform to form the credit information of the requester.

[0232] In this embodiment, the information of the requesting party participating in the implementation of the risk order is collected by the server of the corresponding online trading platform according to the above Figure 2 The embodiments shown are obtained and will not be described in detail here.

[0233] Step S5200: Generate a credit inquiry report based on the acquired credit information and feed it back to the requesting party.

[0234] According to this step, for example, the server of the credit supervision department can send a credit inquiry request to the terminal device of the requesting party; for another example, the credit inquiry report can also be generated and displayed by the terminal device of the credit supervision department for printing the report.

[0235] According to this embodiment, any user and / or merchant participating in the implementation of risky orders will affect the credit rating of the corresponding user and / or merchant, thereby increasing the penalty for participating in fake orders.

[0236] <Device Example>

[0237] This embodiment provides an order recognition device, which is, for example, Figure 6 The order identification device 600 shown includes a matching module 610 and an identification module 620.

[0238] The matching module 610 is used to obtain a feature vector reflecting the degree of matching between the target transaction order and the associated data based on the location area corresponding to the target transaction order when it is generated and the set order information of the target transaction order; wherein the associated data reflects the association relationship between the location area and the abnormal group for the set order information.

[0239] For example, the matching module 610 is used to obtain a feature vector reflecting the degree of matching between the target transaction order and the associated data based on the location area corresponding to when the target transaction order is generated on the e-commerce platform and the set order information of the target transaction order, wherein the set order information includes at least one of product information (object information), merchant information and user information.

[0240] For another example, the matching module 610 is used to obtain a feature vector reflecting the degree of matching between the target transaction order and the associated data based on the location area corresponding to the target transaction order when it is generated on the CreditZhu platform and the set order information of the target transaction order, wherein the set order information includes at least one of room information (object information), merchant information and user information.

[0241] The identification module 620 is used to identify whether the target transaction order is a risky order based on at least the feature vector.

[0242] In one embodiment, the order identification device 600 further includes an associated data acquisition module.

[0243] The associated data acquisition module can be used to: obtain the location area corresponding to the transaction order in the transaction order set when it is generated; obtain the abnormal group for the set order information based on the geographical location according to the location area corresponding to each transaction order; establish the association relationship between each abnormal group and the corresponding location area to obtain the associated data. In one embodiment, when the associated data acquisition module obtains the abnormal group for the set order information based on the geographical location according to the location area corresponding to each transaction order, it can be used to: obtain the location area that meets the set order aggregation requirements as the target area according to the location area corresponding to each transaction order; obtain the abnormal group for the set order information based on the geographical location according to the distribution of the set order information of the transaction orders in each target area.

[0244] In one embodiment, when the associated data acquisition module obtains the location area that meets the set order aggregation requirements based on the location area corresponding to each transaction order: it can be used to obtain the location area where the number of transaction orders within the set time window reaches a set number based on the location area corresponding to each transaction order, as the location area that meets the set order aggregation requirements.

[0245] In one embodiment, when the associated data acquisition module obtains the abnormal group for the set order information based on the geographical location according to the distribution of the set order information of the transaction orders in each target area, it can be used to: cluster the set order information of the corresponding transaction orders for each target area; and obtain the abnormal group according to the clustering results for each target area.

[0246] In one embodiment, when the associated data acquisition module obtains an abnormal group based on the clustering results for each target area, it can be used to: screen out first target clustering results that satisfy a first distribution state from all clustering results for transaction objects, wherein the first distribution state is at least two target object classes with different transaction object categories, and the target object class is an object class in which the number of transaction objects reaches a first set number; for each first target clustering result, aggregate the target object class that satisfies the first distribution state so that the corresponding first target clustering result satisfies the first distribution state, to obtain an abnormal group for the transaction object formed based on the corresponding location area.

[0247] In one embodiment, when the associated data acquisition module clusters the set order information of the corresponding transaction order, it can be used to: for each target area, cluster the merchants for the transaction object categories sold by the merchants involved in the corresponding transaction order; and when obtaining the abnormal group according to the clustering result for each target area, it can be used to: among all the clustering results for the merchants, screen out the clustering results that satisfy the second distribution state as the second target clustering results, wherein the second distribution state is at least two target merchant classes with different merchant categories, and the target merchant class is a merchant class whose sales volume for the corresponding transaction object category reaches a second set number; for each second target clustering result, aggregate the target merchant class so that the corresponding second target clustering result satisfies the second distribution state, and obtain the abnormal group for the merchant formed based on the corresponding location area.

[0248] In one example, when the associated data acquisition module clusters the set order information of the corresponding transaction order, it can be used to: for each target area, cluster users according to the transaction object category required by the user of the corresponding transaction order; and when obtaining abnormal groups based on the clustering results for each target area, it can be used to: among all clustering results for users, screen out clustering results that satisfy a third distribution state as third target clustering results, wherein the third distribution state may be a user class having at least one target user class, and the target user class is a user class whose sales volume of the corresponding transaction object category reaches a third set number; for each third target clustering result, aggregate the target user class so that the corresponding third target clustering result satisfies the third distribution state, and obtain an abnormal group for the user formed based on the corresponding location area.

[0249] In one embodiment, the associated data acquisition module may also be used to: detect a set processing event; and execute the step of obtaining associated data in response to detecting the triggering of any processing event.

[0250] In one embodiment, the target transaction order is a transaction order in the transaction order set.

[0251] In one embodiment, when obtaining a feature vector reflecting the degree of matching between a target transaction order and associated data, the matching module 610 can be used to: for each associated relationship, obtain a matching feature value of the target transaction order for the corresponding associated relationship based on the degree of matching between the target transaction order and the corresponding associated relationship; and arrange the matching feature values ​​for each associated relationship in a set order to obtain a feature vector.

[0252] In one embodiment, when the identification module 620 identifies whether the target transaction order is a risky order based at least on the feature vector, it can be used to: at least input the feature vector into the identification model to obtain an identification result of whether the target transaction order is a risky order.

[0253] In one embodiment, the order identification device 600 may also include a recognition model generation module, which is used to: obtain standard samples; perturb at least part of the standard samples to generate adversarial samples; use both the standard samples and the adversarial samples as training samples to train model parameters to obtain a recognition model.

[0254] In one embodiment, the order identification device 600 may further include a penalty processing module, which is used to: when the target transaction order is identified as a risky order, execute a set penalty item on at least one of the merchant, transaction object and user corresponding to the target transaction order.

[0255] In one embodiment, when the penalty processing module executes the set penalty item on at least one of the merchant, transaction object and user corresponding to the target transaction order, it can be used to report the information of the merchant and / or user corresponding to the target transaction order participating in the implementation of the risk order to the credit supervision department.

[0256] <Equipment Embodiment>

[0257] This embodiment provides an electronic device, which includes the identification device described in the device embodiment of the present invention; or the electronic device is Figure 7 The electronic device 700 shown includes:

[0258] The memory 710 is used to store executable commands.

[0259] The processor 720 is used to execute the method described in any method embodiment of the present invention under the control of the executable instructions stored in the memory 710.

[0260] The implementation subject of the method embodiment executed in the electronic device may be a server or a terminal device.

[0261] <Computer Readable Storage Medium Embodiment>

[0262] This embodiment provides a computer-readable storage medium, in which executable commands are stored. When the executable commands are executed by a processor, the method described in any method embodiment of the present invention is executed.

[0263] The present invention may be a system, a method and / or a computer program product. The computer program product may include a computer-readable storage medium carrying computer-readable program instructions for causing a processor to implement various aspects of the present invention.

[0264] A computer-readable storage medium may be a tangible device that can hold and store instructions used by an instruction execution device. A computer-readable storage medium may be, for example, but not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing. More specific examples of computer-readable storage media (a non-exhaustive list) include: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a static random access memory (SRAM), a portable compact disk read-only memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, a mechanical encoding device, such as a punch card or a raised structure in a groove on which instructions are stored, and any suitable combination of the foregoing. As used herein, a computer-readable storage medium is not to be interpreted as a transient signal per se, such as a radio wave or other freely propagating electromagnetic wave, an electromagnetic wave propagating through a waveguide or other transmission medium (e.g., a light pulse through a fiber optic cable), or an electrical signal transmitted through a wire.

[0265] The computer-readable program instructions described herein can be downloaded from a computer-readable storage medium to each computing / processing device, or downloaded to an external computer or external storage device via a network, such as the Internet, a local area network, a wide area network, and / or a wireless network. The network can include copper transmission cables, optical fiber transmissions, wireless transmissions, routers, firewalls, switches, gateway computers, and / or edge servers. The network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards the computer-readable program instructions for storage in the computer-readable storage medium in each computing / processing device.

[0266] The computer program instructions for performing the operation of the present invention may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, state setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages, such as Smalltalk, C++, etc., and conventional procedural programming languages, such as "C" language or similar programming languages. Computer-readable program instructions may be executed entirely on a user's computer, partially on a user's computer, as an independent software package, partially on a user's computer, partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., using an Internet service provider to connect via the Internet). In some embodiments, an electronic circuit, such as a programmable logic circuit, a field programmable gate array (FPGA), or a programmable logic array (PLA), may be personalized by utilizing the state information of the computer-readable program instructions, and the electronic circuit may execute the computer-readable program instructions, thereby realizing various aspects of the present invention.

[0267] Various aspects of the present invention are described herein with reference to the flow charts and / or block diagrams of the methods, devices (systems) and computer program products according to embodiments of the present invention. It should be understood that each box of the flow chart and / or block diagram and the combination of each box in the flow chart and / or block diagram can be implemented by computer-readable program instructions.

[0268] These computer-readable program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, thereby producing a machine, so that when these instructions are executed by the processor of the computer or other programmable data processing device, a device that implements the functions / actions specified in one or more boxes in the flowchart and / or block diagram is generated. These computer-readable program instructions can also be stored in a computer-readable storage medium, and these instructions cause the computer, programmable data processing device, and / or other equipment to work in a specific manner, so that the computer-readable medium storing the instructions includes a manufactured product, which includes instructions for implementing various aspects of the functions / actions specified in one or more boxes in the flowchart and / or block diagram.

[0269] Computer-readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device so that a series of operating steps are performed on the computer, other programmable data processing apparatus, or other device to produce a computer-implemented process, thereby causing the instructions executed on the computer, other programmable data processing apparatus, or other device to implement the functions / actions specified in one or more boxes in the flowchart and / or block diagram.

[0270] The flowcharts and block diagrams in the accompanying drawings show the possible architecture, functions and operations of the systems, methods and computer program products according to multiple embodiments of the present invention. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of an instruction, and a part of the module, program segment or instruction contains one or more executable instructions for realizing the specified logical function. In some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or the flowchart, and the combination of the boxes in the block diagram and / or the flowchart can be implemented by a dedicated hardware-based system that performs the specified function or action, or can be implemented by a combination of dedicated hardware and computer instructions. It is well known to those skilled in the art that it is equivalent to implement it by hardware, implement it by software, and implement it by combining software and hardware.

[0271] Embodiments of the present invention have been described above, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The terms used herein are selected to best explain the principles of the embodiments, practical applications, or technical improvements in the marketplace, or to enable other persons of ordinary skill in the art to understand the embodiments disclosed herein. The scope of the present invention is defined by the appended claims.

Claims

1. An order identification method, comprising: According to the location area corresponding to the target transaction order when it is generated and the set order information of the target transaction order, a feature vector reflecting the matching degree between the target transaction order and the associated data is obtained; The location area corresponding to the target transaction order when it is generated is determined according to the geographical location information of the terminal device when the target transaction order is generated, the set order information includes the geographical location information of the terminal device when the target transaction order is generated, at least one of the transaction object, user information and merchant information, the associated data reflects the association relationship between the location area and the abnormal group for the set order information, the abnormal group includes at least one of the abnormal group for the transaction object formed based on the geographical location, the abnormal group for the merchant formed based on the geographical location, and the abnormal group for the user formed based on the geographical location, and all members in the same abnormal group are associated with the same location area; At least input the feature vector into a recognition model to obtain a recognition result of whether the target transaction order is a risky order; The obtaining of a feature vector reflecting the degree of matching between the target transaction order and the associated data includes: For each of the association relationships, according to the matching degree between the target transaction order and the corresponding association relationship, a matching feature value of the target transaction order for the corresponding association relationship is obtained; The matching feature values ​​for each of the association relationships are arranged in a set order to obtain the feature vector.

2. The method according to claim 1, wherein: The method further comprises the step of obtaining the associated data, comprising: Get the location area corresponding to the transaction order in the transaction order set when it was generated; According to the location area corresponding to each of the transaction orders, an abnormal group for the set order information formed based on the geographical location is obtained; An association relationship is established between each of the abnormal groups and the corresponding location area to obtain the association data.

3. The method according to claim 2, wherein: The step of obtaining, according to the location area corresponding to each of the transaction orders, an abnormal group for setting the order information based on the geographical location, includes: According to the location area corresponding to each of the transaction orders, a location area that meets the set order aggregation requirements is obtained as a target area; According to the distribution of the set order information of the transaction orders in each of the target areas, an abnormal group for the set order information formed based on the geographical location is obtained.

4. The method according to claim 3, wherein: The obtaining, according to the location area corresponding to each of the transaction orders, a location area that meets the set order aggregation requirements, includes: According to the location area corresponding to each of the transaction orders, the location area where the number of transaction orders within the set time window reaches the set number is obtained as the location area that meets the set order aggregation requirement.

5. The method according to claim 3, wherein: The obtaining, according to the distribution of the set order information of the transaction orders in each of the target areas, an abnormal group for the set order information formed based on the geographical location, comprises: For each of the target areas, clustering the set order information of the corresponding transaction orders; The abnormal group is obtained according to the clustering result for each of the target areas.

6. The method according to claim 5, wherein: The setting order information includes transaction object information, and obtaining the abnormal group according to the clustering result for each target area includes: Screening out a first target clustering result satisfying a first distribution state from among all clustering results for transaction objects, wherein the first distribution state is at least two target object classes having different object categories, and the target object class is an object class in which the number of transaction objects reaches a first set number; For each of the first target clustering results, aggregation is performed so that the corresponding first target clustering result satisfies the target object class of the first distribution state, and an abnormal group for transaction objects formed based on the corresponding location area is obtained.

7. The method according to claim 5, wherein: The order information includes merchant information, and for each target area, clustering the set order information of the corresponding transaction order includes: For each of the target areas, clustering merchants according to the transaction object categories provided by the merchants involved in the corresponding transaction orders; The step of obtaining the abnormal group according to the clustering result for each target area includes: Among all the clustering results for merchants, a clustering result satisfying a second distribution state is screened out as a second target clustering result, wherein the second distribution state is at least two target merchant classes with different merchant categories, and the target merchant class is a merchant class whose sales volume for the corresponding transaction object category reaches a second set number; For each of the second target clustering results, aggregation is performed so that the corresponding second target clustering result satisfies the target merchant class of the second distribution state, and an abnormal group for the merchant formed based on the corresponding location area is obtained.

8. The method according to claim 5, wherein: The order information includes user information, and for each target area, clustering the set order information of the corresponding transaction order includes: For each of the target areas, user clustering is performed according to the transaction object categories required by users of the corresponding transaction orders; The obtaining of the abnormal group according to the clustering result for each of the location areas further comprises: Among all clustering results for users, a clustering result satisfying a third distribution state is screened out as a third target clustering result, wherein the third distribution state is having at least one target user class, and the target user class is a user class whose sales volume of the corresponding transaction object class reaches a third set number; For each of the third target clustering results, aggregation is performed so that the corresponding third target clustering result satisfies the target user class of the third distribution state, and an abnormal group for the user formed based on the corresponding location area is obtained.

9. The method according to claim 2, wherein: The method further comprises: Detect the set processing events; In response to detecting the triggering of any of the processing events, the step of obtaining the associated data is performed.

10. The method according to claim 2, wherein: The target transaction order is a transaction order in the transaction order set.

11. The method according to claim 1, wherein: The method further comprises the step of generating the recognition model, comprising: Obtain standard samples; Perturbing at least part of the standard samples to generate adversarial samples; The standard sample and the adversarial sample are both used as training samples to train model parameters to obtain the recognition model.

12. The method according to any one of claims 1 to 10, wherein: The method further comprises: In the case where the target transaction order is identified as a risky order, a set penalty item is executed on at least one of the merchant, commodity and user corresponding to the target transaction order.

13. The method according to claim 12, wherein: The executing of a set penalty item on at least one of the merchant, commodity and user corresponding to the target transaction order includes: Report the information on the merchant and / or user corresponding to the target transaction order participating in the implementation of the risk order to the credit supervision department.

14. An order identification method for an e-commerce platform, comprising: According to the location area corresponding to the target transaction order when it is generated on the e-commerce platform and the set order information of the target transaction order, a feature vector reflecting the degree of matching between the target transaction order and the associated data is obtained, wherein the location area corresponding to the target transaction order when it is generated on the e-commerce platform is determined according to the geographical location information of the terminal device when the target transaction order is generated, and the set order information includes at least one of product information, merchant information and user information; The association data reflects the association relationship between the location area and the abnormal group for setting the order information, wherein the abnormal group includes at least one of an abnormal group for transaction objects formed based on the geographical location, an abnormal group for merchants formed based on the geographical location, and an abnormal group for users formed based on the geographical location, and all members in the abnormal group are associated with the same location area; At least input the feature vector into a recognition model to obtain a recognition result of whether the target transaction order is a risky order; The obtaining of a feature vector reflecting the degree of matching between the target transaction order and the associated data includes: For each of the association relationships, according to the matching degree between the target transaction order and the corresponding association relationship, a matching feature value of the target transaction order for the corresponding association relationship is obtained; The matching feature values ​​for each of the association relationships are arranged in a set order to obtain the feature vector.

15. An order recognition method for a credit housing platform, comprising: According to the location area corresponding to the target transaction order when it is generated on the CreditZhu platform and the set order information of the target transaction order, a feature vector reflecting the degree of matching between the target transaction order and the associated data is obtained, wherein the location area corresponding to the target transaction order when it is generated on the CreditZhu platform is determined according to the geographical location information of the terminal device when the target transaction order is generated, and the set order information includes at least one of room information, merchant information and user information; The association data reflects the association relationship between the location area and the abnormal group for setting the order information, wherein the abnormal group includes at least one of an abnormal group for transaction objects formed based on the geographical location, an abnormal group for merchants formed based on the geographical location, and an abnormal group for users formed based on the geographical location, and all members in the abnormal group are associated with the same location area; At least input the feature vector into a recognition model to obtain a recognition result of whether the target transaction order is a risky order; The obtaining of a feature vector reflecting the degree of matching between the target transaction order and the associated data includes: For each of the association relationships, according to the matching degree between the target transaction order and the corresponding association relationship, a matching feature value of the target transaction order for the corresponding association relationship is obtained; The matching feature values ​​for each of the association relationships are arranged in a set order to obtain the feature vector.

16. An order identification method, implemented by a terminal device, comprising: Acquire order information of the transaction order, wherein the order information includes geographic location information of the terminal device when the transaction order is generated; The order information is sent to a server for risk order identification, wherein the risk order identification comprises: obtaining the location area corresponding to the transaction order when it is generated according to the geographic location information, and obtaining a feature vector reflecting the degree of matching between the transaction order and the associated data according to the location area and the set order information of the transaction order; and at least inputting the feature vector into a recognition model to obtain an identification result of whether the transaction order is a risk order; the set order information comprises at least one of the geographic location information, transaction object, user information and merchant information of the terminal device when the transaction order is generated, the associated data reflects the association relationship between the location area and the abnormal group for the set order information, the abnormal group comprises at least one of the abnormal group for the transaction object formed based on the geographic location, the abnormal group for the merchant formed based on the geographic location, and the abnormal group for the user formed based on the geographic location, and all members in the same abnormal group are associated with the same location area; In response to a processing result given by the server according to the recognition result, recording and / or prompting the processing result; The obtaining of a feature vector reflecting the degree of matching between the transaction order and the associated data includes: For each of the association relationships, according to the matching degree between the transaction order and the corresponding association relationship, a matching feature value of the transaction order for the corresponding association relationship is obtained; The matching feature values ​​for each of the association relationships are arranged in a set order to obtain the feature vector.

17. A credit processing method based on order identification, implemented by an electronic device of a credit supervision department, comprising: In response to a credit query request, the credit information of the requester is obtained, wherein the credit information includes information on the participation of the requester in implementing a risk order, and the information on the participation of the requester in implementing a risk order is determined by the server of the online trading platform executing the following method for identifying risk orders: obtaining a feature vector reflecting the degree of matching between the target transaction order and the associated data according to the location area corresponding to the target transaction order when it is generated and the set order information of the target transaction order; at least inputting the feature vector into a recognition model to obtain an identification result of whether the target transaction order is a risk order; wherein the location area corresponding to the target transaction order when it is generated is determined according to the geographical location information of the terminal device when the target transaction order is generated, the set order information includes at least one of the geographical location information of the terminal device when the target transaction order is generated, the transaction object, the user information and the merchant information, the associated data reflects the association relationship between the location area and the abnormal group for the set order information, the abnormal group includes at least one of the abnormal group for the transaction object formed based on the geographical location, the abnormal group for the merchant formed based on the geographical location, and the abnormal group for the user formed based on the geographical location, and all members in the same abnormal group are associated with the same location area; The obtaining of a feature vector reflecting the degree of matching between the target transaction order and the associated data includes: For each of the association relationships, according to the matching degree between the target transaction order and the corresponding association relationship, a matching feature value of the target transaction order for the corresponding association relationship is obtained; Arranging the matching feature values ​​for each of the association relationships in a set order to obtain the feature vector; A credit inquiry report is generated based on the acquired credit information and fed back to the requesting party.

18. An order identification device, comprising: A matching module, used to obtain a feature vector reflecting the matching degree between the target transaction order and the associated data according to the location area corresponding to the target transaction order when it is generated and the set order information of the target transaction order; The location area corresponding to the target transaction order when it is generated is determined according to the geographical location information of the terminal device when the target transaction order is generated, the set order information includes the geographical location information of the terminal device when the target transaction order is generated, at least one of the transaction object, user information and merchant information, the associated data reflects the association relationship between the location area and the abnormal group for the set order information, the abnormal group includes at least one of the abnormal group for the transaction object formed based on the geographical location, the abnormal group for the merchant formed based on the geographical location, and the abnormal group for the user formed based on the geographical location, and all members in the same abnormal group are associated with the same location area; an identification module, configured to input at least the feature vector into an identification model to obtain an identification result of whether the target transaction order is a risky order; The matching module is also used to obtain, for each of the association relationships, a matching feature value of the target transaction order for the corresponding association relationship based on the matching degree between the target transaction order and the corresponding association relationship; and to arrange the matching feature values ​​for each of the association relationships in a set order to obtain the feature vector.

19. An electronic device, comprising the identification device according to claim 18, or the electronic device comprising: A memory for storing executable commands; A processor, configured to execute the method according to any one of claims 1 to 17 under the control of the executable command.

20. A computer-readable storage medium storing executable instructions, wherein when the executable instructions are executed by a processor, the method according to any one of claims 1 to 17 is performed.

Images