Object information determination method and device, equipment and readable medium

Abstract

The embodiment of the invention provides an object information determination method and device, equipment and a readable medium. The method comprises the steps that an object information set is determined, the object information set comprises object information corresponding to at least one object, the object information corresponding to each object is represented as a graph comprising nodes and edges, each node represents the object information used by each object in different service links of a service, and each edge represents the object information used by each object in different service links of the service; each edge represents the relationship between the service links corresponding to the connected nodes; providing the object information set to a machine learning model to extract node feature representations corresponding to the nodes and edge feature representations corresponding to the edges; based on the node feature representation and the edge feature representation, utilizing a machine learning model to determine an object feature representation corresponding to the object information; providing the object feature representation to a classification machine learning model to determine a prediction result of each object in the object information set; and selecting a target object information set from the object information sets based on the prediction result.

Description

Technical Field

[0001] Example embodiments of the present disclosure generally relate to the field of computer technology, and more particularly, to methods, apparatuses, devices, and computer-readable storage media for object information determination. Background Art

[0002] When logistics companies provide discounts to merchants after taking into account factors such as business volume and profit, some risky parties will seek a portion of the profits through some means. Risky parties refer to those who use the low-discount accounts they have obtained to other parties who have not reached the discount in order to seek a portion of the profits. Risky parties use this method to earn price difference profits, collect price arbitrage in the middle, interfere with market pricing, and damage the deserved interests of logistics companies. How to accurately identify risky parties based on existing data information and prevent or interrupt risky behaviors in a timely manner is an urgent problem to be solved. Summary of the invention

[0003] In a first aspect of the present disclosure, a method for determining object information is provided. The method includes: determining an object information set, the object information set includes object information corresponding to at least one object, and the object information corresponding to each object is represented as a graph including nodes and edges, each node represents the object information used by each object in different business links of the business, and each edge represents the relationship between the business links corresponding to the connected nodes; providing the object information set to a machine learning model to extract the node feature representation corresponding to the node and the edge feature representation corresponding to the edge; based on the node feature representation and the edge feature representation, using the machine learning model to determine the object feature representation corresponding to the object information, the object feature representation characterizing the risk feature of each object; providing the object feature representation to a classification machine learning model to determine the prediction result for each object in the object information set, the prediction result indicating whether each object is classified as a risk object; and based on the prediction result, selecting a target object information set from the object information set, the target object information set including the object information corresponding to at least one risk object.

[0004] In the second aspect of the present disclosure, a device for determining object information is provided. The device includes: a determination module, configured to determine an object information set, the object information set includes object information corresponding to at least one object, and the object information corresponding to each object is represented as a graph including nodes and edges, each node represents the object information used by each object in different business links of the business, and each edge represents the relationship between the business links corresponding to the connected nodes; an extraction module, configured to provide the object information set to a machine learning model to extract the node feature representation corresponding to the node and the edge feature representation corresponding to the edge; a utilization module, configured to determine the object feature representation corresponding to the object information based on the node feature representation and the edge feature representation, using the machine learning model, the object feature representation characterizing the risk feature of each object; a provision module, configured to provide the object feature representation to a classification machine learning model to determine the prediction result for each object in the object information set, the prediction result indicating whether each object is classified as a risk object; and an execution module, configured to select a target object information set from the object information set based on the prediction result, the target object information set including object information corresponding to at least one risk object.

[0005] In a third aspect of the present disclosure, an electronic device is provided. The electronic device includes at least one processing unit; and at least one memory, the at least one memory is coupled to the at least one processing unit and stores instructions for execution by the at least one processing unit, and when the instructions are executed by the at least one processing unit, the electronic device executes the method of the first aspect of the present disclosure.

[0006] In a fourth aspect of the present disclosure, a computer-readable storage medium is provided, wherein a computer program is stored on the computer-readable storage medium, and the computer program can be executed by a processor to perform the method according to the first aspect of the present disclosure.

[0007] In a fifth aspect of the present disclosure, a computer program product is provided, comprising computer executable instructions, wherein the computer executable instructions can be executed by a processor to perform the method according to the first aspect of the present disclosure.

[0008] It should be understood that the contents described in the summary of the invention are not intended to limit the key features or important features of the embodiments of the present disclosure, nor are they intended to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] In the following, in conjunction with the accompanying drawings and with reference to the following detailed description, the above and other features, advantages and aspects of various implementations of the present disclosure will become more apparent. In the accompanying drawings, the same or similar reference numerals represent the same or similar elements, wherein:

[0010] Figure 1A schematic diagram showing an example environment in which embodiments of the present disclosure can be implemented;

[0011] Figure 2 A flowchart showing a process of object information determination according to some embodiments of the present disclosure;

[0012] Figure 3 A schematic diagram showing a process of determining object information according to some embodiments of the present disclosure;

[0013] Figure 4 A schematic diagram showing an example of a representation form of object information according to some embodiments of the present disclosure;

[0014] Figure 5 A schematic structural block diagram of an apparatus for determining object information according to some embodiments of the present disclosure is shown; and

[0015] Figure 6 A block diagram of an electronic device is shown in which one or more embodiments of the present disclosure may be implemented. DETAILED DESCRIPTION

[0016] Embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although some embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be construed as being limited to the embodiments described herein. On the contrary, these embodiments are provided to provide a more thorough and complete understanding of the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are only for exemplary purposes and are not intended to limit the scope of protection of the present disclosure.

[0017] In the description of the embodiments of the present disclosure, the term "including" and similar terms should be understood as open inclusion, that is, "including but not limited to". The term "based on" should be understood as "based at least in part on". The term "one embodiment" or "the embodiment" should be understood as "at least one embodiment". The term "some embodiments" should be understood as "at least some embodiments". Other explicit and implicit definitions may also be included below.

[0018] It should be noted that the acquisition, storage and application of user personal information involved in the technical solution of the present disclosure are in compliance with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0019] It is understandable that before using the technical solutions disclosed in the embodiments of the present disclosure, the types, scope of use, usage scenarios, etc. of the personal information involved in the present disclosure should be informed to the user and the user's authorization should be obtained in an appropriate manner in accordance with relevant laws and regulations.

[0020] For example, in response to receiving an active request from a user, a prompt message is sent to the user to clearly prompt the user that the operation requested to be performed will require obtaining and using the user's personal information, so that the user can independently choose whether to provide personal information to software or hardware such as electronic devices, applications, servers or storage media that execute operations of the technical solution of the present disclosure based on the prompt message.

[0021] As an optional but non-limiting implementation, in response to receiving an active request from the user, the prompt information is sent to the user in a manner such as a pop-up window, in which the prompt information can be presented in text form. In addition, the pop-up window can also carry a selection control for the user to choose "agree" or "disagree" to provide personal information to the electronic device.

[0022] The traditional scheme for determining object information is through regional or external reporting. The event report is jointly investigated and verified by the relevant departments and the supervisory department. After the risk object is finally determined, the risky merchants are dealt with accordingly. This scheme cannot actively identify. At the same time, the auditor needs to go to the region to investigate and collect evidence in person. The evidence is passively obtained, and the effectiveness is low (the efficiency of manual review is low and the cost of issuing regional verification is high). Alternatively, a scoring model is built through the registration information risk and operation risk of the object, and different risk indicators are assigned different weights to calculate the merchant risk score, that is, by simply calculating the registered mobile phone number and email overlap of different merchants (shared phone number / number of email addresses) and the overlap of the shipping mobile phone number and shipping account number of the shipping bill (shared shipping mobile phone number / number of shipping accounts), the two risks of merchant registration and operation are defined, and a scoring model is built to calculate the merchant risk score, thereby identifying high-risk suspected scalper merchants. In general, the traditional scheme for determining object information relies on people's subjective experience and considers fewer dimensions. It only considers the association within a single type of information, but does not consider the cross-information association between merchants.

[0023] In view of this, an embodiment of the present disclosure provides a solution for determining object information. The solution includes: determining an object information set; providing the object information set to a machine learning model to extract node feature representations corresponding to nodes and edge feature representations corresponding to edges; based on the node feature representations and edge feature representations, using a machine learning model to determine the object feature representations corresponding to the object information; providing the object feature representations to a classification machine learning model to determine the prediction results for each object in the object information set; and based on the prediction results, selecting a target object information set from the object information set. In this way, the accuracy of object information recognition can be improved, the efficiency of object information determination can be improved, and business risks can be reduced.

[0024] Figure 1 1 is a schematic diagram of an example environment 100 in which embodiments of the present disclosure can be implemented. Figure 1In the environment 100 , an information processing system 110 is deployed for selecting a target object information set 135 from the object information set 105 .

[0025] In an embodiment of the present disclosure, the object information set 105 may include object information corresponding to at least one object, and the object information corresponding to each object is represented as a graph including nodes and edges, each node represents the object information used by each object in different business links of the business, and each edge represents the relationship between the business links corresponding to the connected nodes. In some examples, the object may represent a mailing merchant, and the object information may represent mailing information. It is understood that this is merely exemplary and not restrictive. There may be other examples, and the scope of the embodiments of the present disclosure is not limited to this. In other examples, the scheduled business may be a logistics business. It is understood that this is merely exemplary and not restrictive, and there may be other examples, and the scope of the embodiments of the present disclosure is not limited to this. In more examples, when the object information represents the information of the mailing merchant and the scheduled business is an express business or an express business, the business link may include a mailing link and a registration link. It is understood that this is merely exemplary and not restrictive. There may be other examples, and the scope of the embodiments of the present disclosure is not limited to this.

[0026] In an embodiment of the present disclosure, the information processing system 110 may provide the object information set 105 to the machine learning model 120 to extract the node feature representation corresponding to the node and the edge feature representation corresponding to the edge. The information processing system 110 may also determine the object feature representation 124 corresponding to the object information based on the node feature representation and the edge feature representation using the machine learning model 120, and the object feature representation 124 may characterize the risk characteristics of each object. In some examples, when the object information represents the information of the sender and the scheduled business is an express business or an express business, the risk characteristics may indicate that a certain sender has at least one of the following identical information with one or more other senders, such as a sender account number, a sender mobile phone number, a registered mobile phone number, a registered email address, and a unique identity identifier, etc. It will be understood that this is merely exemplary and not restrictive. There may be other examples, and the scope of the embodiments of the present disclosure is not limited to this.

[0027] In the embodiment of the present disclosure, the information processing system 110 can provide the object feature representation 124 to the classification machine learning model 130 to determine the prediction result for each object in the object information set 105, and the prediction result can indicate whether each object is classified as a risk object. In some examples, when the object information represents the information of the sender and the scheduled business is an express business or an express business, the sender reaches a certain agreement with the logistics company. When the shipment volume of the sender reaches the predetermined volume, the logistics company can provide the merchant with a certain discount on logistics costs. In this case, some merchants with lower discounts provide services such as replacement shipments for other merchants who have not reached the lower discounts, and then make profits from it. We regard such merchants as risk objects. It can be understood that this is merely exemplary and not restrictive. There may be other examples, and the scope of the embodiments of the present disclosure is not limited to this.

[0028] In the embodiment of the present disclosure, the information processing system 110 may further select a target object information set 135 from the object information set 105 based on the prediction result. The target object information set 135 may include object information corresponding to at least one risk object.

[0029] The information processing system 110 may include any computing system with computing capabilities, such as various computing devices / systems, terminal devices, server devices, etc. The terminal device may be any type of mobile terminal, fixed terminal or portable terminal, including a mobile phone, a desktop computer, a laptop computer, a notebook computer, a netbook computer, a tablet computer, a media computer, a multimedia tablet, a handheld computer, a portable game terminal, a VR / AR device, a personal communication system (PCS) device, a personal navigation device, a personal digital assistant (PDA), an audio / video player, a digital camera / camcorder, a positioning device, a television receiver, a radio broadcast receiver, an e-book device, a game device, or any combination of the foregoing, including accessories and peripherals of these devices or any combination thereof.

[0030] It should be understood that the structure and function of the various elements in the environment 100 are described for exemplary purposes only and do not imply any limitation on the scope of the present disclosure.

[0031] Some example embodiments of the present disclosure will be described below with continued reference to the accompanying drawings.

[0032] Figure 2 200 is a flowchart showing a process of determining object information according to some embodiments of the present disclosure. Process 200 may be implemented at information processing system 110. For ease of discussion, reference will be made to Figure 1The process 200 is described with reference to the environment 100 of FIG.

[0033] like Figure 2 As shown, the information processing system 110 can determine the object information set 105. Further, the object information set 105 may include object information corresponding to at least one object. The object information corresponding to each object is represented in the form of a graph, which has multiple nodes and edges for connecting the nodes. Each node in the graph represents the object information used by each object in different business links of the business, and each edge represents the relationship between the business links corresponding to the connected nodes. In some embodiments, when the object information represents the information of the sender and the scheduled business is an express business or an express business, the sender reaches a certain agreement with the logistics company. When the shipment volume of the sender reaches a predetermined amount, the logistics company provides the merchant with a certain discount on logistics costs. In this way, the relationship between merchants and the association between different types of information can be considered, so that the target object information set 135 is obtained more accurately.

[0034] In some examples, a large number of associations between different merchants are represented by a graph. By analyzing the basic risks and operational risks related to the merchants, the basic risks may include the risks of the corresponding registered mobile phone number and the corresponding registered email information, that is, different merchants use the same registered mobile phone number or registered email. Operational risks may include the risks of associations between different merchants' shipping behaviors. Specifically, the associations between different merchants' shipping behaviors may include the sending mobile phone number and the sending account number, that is, different merchants use the same sending mobile phone number or sending account number. The K code represents the unique identifier corresponding to the merchant, and the K code includes object information such as the registered mobile phone number, registered email address, sending mobile phone number, and sending account number. In some examples, each node may represent object information such as the K code, registered mobile phone number, sending mobile phone number, sending account number, and registered email address. In other examples, each edge may represent registration information or sending information.

[0035] In some embodiments, the object information set 105 may include object information corresponding to at least one object related to the predetermined business and having a risk level greater than a risk threshold. In some examples, when the object is a mailing merchant related to express delivery or express transportation and having a risk level greater than a risk threshold, the object information may include unique identification information, registered mobile phone number, registered email address, mailing mobile phone number, mailing account number, merchant mailing record, etc. It will be understood that this is merely exemplary and not restrictive. There may also be other examples, and the scope of the embodiments of the present disclosure is not limited to this. In this way, the number of objects to be confirmed can be reduced, and the target object information can be identified more accurately.

[0036] In some embodiments, the information processing system 110 obtains the object information set 105 to be screened and the business type of the associated predetermined business. Further, the information processing system 110 can determine the risk condition for the object information set 105 to be screened based on the business type. Furthermore, the information processing system 110 can determine the object information that meets the risk condition in the object information set 105 to be screened as an object whose risk level is greater than the risk threshold. In addition, the information processing system 110 can also determine the object information set 105 based on the object whose risk level is greater than the risk threshold. In this way, the number of samples can be reduced, and the recognition range of the machine learning model 120 can be reduced while ensuring a high recall, thereby further improving the accuracy of the machine learning model 120.

[0037] In more examples, in the scenario where the object information is the information of the sender and the scheduled business is the logistics business, by combining the screening conditions, the merchants with low risks are removed and the merchant information with higher risks is retained. In some examples, the screening conditions can be determined based on some business experience and merchant delivery characteristics. Since the discount rates of merchants in different product lines vary greatly under the same delivery volume, the delivery volume of the express business is approximately 10 times that of the express business under the same discount. Therefore, the merchants are first classified based on different businesses and certain discount rate thresholds are set for merchants of different business types to filter out merchants below the business discount threshold corresponding to the discount rate. For example, for merchants in the express business, there may be risky merchants when the business discount is less than 45%, but for merchants in the express business, when the business discount is less than 60%, machine learning is used. Therefore, the first discount threshold of 0.45 can be selected as the discount threshold for merchants in the express business, and the second discount threshold of 0.6 can be selected as the discount threshold for merchants in the express business. It can be understood that this is only exemplary and not restrictive. In addition, there may be reverse services in express delivery or express transportation business, that is, when the consignee returns the goods, the consignee or other merchants provide delivery services. We do not consider the reverse service situation to reduce the risk of misidentification.

[0038] In some embodiments, the business type includes a first type and a second type. Further, if the business type indicates the first type, the information processing system 110 may determine that the risk condition for the object information set 105 to be screened is the first condition. Specifically, the first type may indicate that the business volume of at least one object in the object information set 105 to be screened is less than the first business threshold. The first condition here indicates that the business index in the object information set 105 to be screened is less than the first index threshold. Further, if the business type indicates the second type, the information processing system 110 may determine that the risk condition for the object information set 105 to be screened is the second condition. Specifically, the second type may indicate that the business volume of at least one object in the object information set 105 to be screened is greater than the second business threshold, and the second condition may indicate that the business index in the object information set 105 to be screened is less than the second index threshold. The first business threshold here is less than or equal to the second business threshold, and the first index threshold is less than the second index threshold. In some examples, when the scheduled business is a logistics business, the first type may indicate an express business, and the second type may indicate an express business. It can be understood that this is merely exemplary and not restrictive. There may be other examples, which are not limited by the scope of the embodiments of the present disclosure. In this way, a part of the object information is first screened out to reduce the scope of the objects to be confirmed and improve efficiency.

[0039] In other embodiments, the information processing system 110 may provide the object information set 105 to the machine learning model 120 to extract the node feature representation corresponding to the node and the edge feature representation corresponding to the edge. Further, the information processing system 110 may determine the object feature representation 124 corresponding to the object information based on the node feature representation and the edge feature representation using the machine learning model 120, and the object feature representation 124 may characterize the risk characteristics of each object. In some examples, the node feature representation may be a node embedding vector. In other examples, the edge feature representation may be an edge embedding vector. It will be understood that this is merely exemplary and not restrictive. In this way, a more accurate object feature representation 124 is obtained, and then a more accurate target object information set 135 is obtained.

[0040] In some embodiments, the machine learning model 120 may be determined in the following manner. The information processing system 110 obtains a historical object information set. Further, the information processing system 110 may determine a training set from the historical object information set, the training set including historical object information corresponding to at least one object. Further, the information processing system 110 may perform training on an untrained model based on the training set to determine the machine learning model 120. In this way, a more accurate target object information set 135 may be obtained through the trained machine learning model 120.

[0041] In some examples, due to the consideration of the segmented association relationships between merchants, a heterogeneous relationship approach is used to define the graph, and at the same time, if only K codes are selected for the nodes, all association relationships are considered to be the same (basic risk and operational risk cannot be distinguished). At the same time, if a media identifier (ID) used by multiple merchants appears, a larger level of edges needs to be inserted, and the magnitude of the graph will be very large. Graph algorithms with heterogeneous nodes and edges are selected to determine the machine learning model 120. In other examples, the machine learning model 120 can be a GATNE-T model, which uses a general attributed multiplex heterogeneous network embedding learning framework (General Attributed Multiplex HeTerogeneous Network Embedding, GATNE), which is a framework that can capture rich attribute information of nodes and multiplexed topological structures from different node types. The GATNE-T model is a direct learning method. If a new data node is added to the test data set, the model needs to be retrained. In addition, the edge type is taken into account and modeled as MHEN (multiple generators). Furthermore, the attributes of the nodes are also taken into account and modeled as AMHEN (Attributed Multiple Heterogeneous Network). In this way, all information can be fully utilized, which can achieve better results.

[0042] In other examples, the feature space of the GATNE-T model includes two parts, one is the base space, and the other is the heterogeneous feature space other than the base. Therefore, the sources of node aggregation information in the GATNE-T model are the above two parts. That is to say, the embedding representation (embedding) of the node on the edge type is divided into two parts: base embedding (base embedding) and edge embedding (edge ​​embedding). In addition, the base embedding can be shared among different types of edges, and the base embedding is a common type. Edge embedding is the aggregation of node features under different edge types. More importantly, in the heterogeneous feature space, the final representation of the node in the K layer is the fusion of the embedding representations of each edge type. Since different types of edges will have an impact on the node, but the importance of different types of edges to the node is different, the GATNE-T model can adaptively adjust the weights of different types of edges through the self-attention mechanism (self-attention).

[0043] In some embodiments, for the first object in at least one risk object, if the information processing system 110 identifies a newly added node, the first node information corresponding to the newly added node can be determined in the object information corresponding to the first object. Further, the information processing system 110 can update the target object information set 135 based on the first node information. Furthermore, the information processing system 110 can also retrain the machine learning model 120 based on the updated target object information set 135 to update the parameters of the machine learning model 120. In this way, the parameters of the machine learning model 120 can be made more accurate in the presence of newly added nodes, thereby obtaining a more accurate target object information set 135, and improving efficiency.

[0044] In some embodiments, if the information processing system 110 identifies a newly added second object related to the business and having a risk level greater than a risk threshold, the second object information corresponding to the second object can be determined. Further, the information processing system 110 can update the target object information set 135 based on the second object information. Furthermore, the information processing system 110 can also retrain the machine learning model 120 based on the updated target object information set 135 to update the parameters of the machine learning model 120. In this way, the parameters of the machine learning model 120 can be made more accurate when there are newly added objects to be determined, thereby obtaining a more accurate target object information set 135, and improving accuracy.

[0045] In some embodiments, the information processing system 110 may provide the object feature representation 124 to the classification machine learning model 130 to determine the prediction results for each object in the object information set 105. Specifically, the prediction results may indicate whether each object is classified as a risk object. In some examples, the classification machine learning model 130 may be a multilayer perceptron (MLP) classifier. In other examples, the multilayer perceptron classifier may be based on the acquired feature representation as the input layer, and perform feature extraction and nonlinear mapping through the hidden layer, and then output the prediction results for the object information. In this way, an accurate prediction result for the object information set 105 can be obtained through the classification machine learning model 130, and then a target object information set 135 with high accuracy can be obtained, thereby improving the accuracy.

[0046] In some embodiments, the information processing system 110 may select a target object information set 135 from the object information set 105 based on the prediction result. Specifically, the target object information set 135 includes object information corresponding to at least one risk object. In this way, corresponding prevention measures may be taken for the risk object according to the obtained target object information set 135 to reduce business risks.

[0047] In some embodiments, the information processing system 110 may select risk object information indicated as a risk object by the prediction result from the object information set 105. Further, the information processing system 110 may determine the audit result of the risk object information based on the first audit rule. Further, the information processing system 110 may also select a target object information set 135 from the risk object information based on the audit result.

[0048] In some embodiments, the information processing system 110 determines a second audit result of the object information of each object in the object information set 105 based on the second audit rule. Specifically, the second audit result may indicate whether each object is a risk object or a non-risk object. Further, the information processing system 110 may discard the object information of a non-risk object indicated by the second audit result from the object information set 105 to update the object information set 105. Further, the information processing system 110 may select a target object information set 135 from the updated object information set 105 based on the prediction result. In some examples, the second audit rule may be determined based on the business experience of relevant personnel or the experience of relevant experts. It will be understood that this is merely exemplary and not restrictive. In this way, a more accurate target object information set 135 may be obtained through the second audit rule, thereby improving the accuracy of determining the target object information.

[0049] In some embodiments, the information processing system 110 may provide the target object information set 135 to the object management system to trigger an early warning notification for at least one risk object in the object management system. In this way, an early warning notification may be issued in a timely manner, so that corresponding deterrent measures may be taken against the risk object, and business risks may be effectively reduced.

[0050] Continue to refer to Figure 3 The schematic diagram of the process of determining the object information is further described. It can be understood that this is only exemplary and not restrictive.

[0051] Figure 3 3200 is a schematic diagram showing a process of determining object information according to some embodiments of the present disclosure. Process 3200 may be implemented at information processing system 110. For ease of discussion, reference will be made to Figure 1 The process 300 is described with reference to the environment 100 of FIG.

[0052] In some embodiments, on the algorithm side, according to the offline strategy, the information processing system 110 may provide the object information set 105 to the graph model 320 to extract the node feature representation corresponding to the node and the edge feature representation corresponding to the edge. Further, the information processing system 110 may determine the object feature representation 124 corresponding to the object information using the graph model 320 based on the node feature representation and the edge feature representation. Further, the information processing system 110 may provide the object feature representation 124 to the multi-layer perceptron classifier 330, and the multi-layer perceptron classifier 330 may determine the prediction result for each object in the object information set 105 based on the object feature representation 124 received from the graph model 320. In some examples, the prediction result for each object may indicate that the object is classified as a risk object. In other examples, the prediction result for each object may also indicate that the object is not classified as a risk object. In other examples, the graph model 320 may be a GATNE-T model.

[0053] Furthermore, the information processing system 110 may select risk object information indicated as a risk object by the prediction result from the object information set 105. The information processing system 110 may determine the audit result of the risk object information based on the first audit rule 340. The information processing system 110 may also select a target object information set 135 from the risk object information based on the audit result. In some embodiments, the target information set 135 may be uploaded to an operation support system (OSS) 360 for storage. In other embodiments, part of the object information of each object in the target information set 135 may also be uploaded to a customer relationship management (CRM) system 370.

[0054] In some embodiments, on the business side, the information processing system 110 may also determine a second audit result of partial object information of each object in the object information set 105 based on the second audit rule 380. In some examples, the second audit result may indicate that each object is a risk object. In other examples, the second audit result may indicate that each object is a non-risk object. Further, the information processing system 110 may discard object information that the second audit result indicates is a non-risk object from the object information set 105 to update the object information set 105. The information processing system 110 selects a target object information set 135 from the updated object information set 105 based on the prediction result. In some examples, when the scheduled business is a logistics business and the object is a consignor merchant, the object information set 105 may include at least one candidate risk merchant, and the target object information set 135 includes at least one risk merchant.

[0055] Continue to refer to Figure 4 The example representation of the object information when the object information is the information of the shipping merchant and the scheduled service is the logistics service is further described. It should be understood that this is merely exemplary and not restrictive.

[0056] Figure 4 A schematic diagram showing an example of a representation form 400 of object information according to some embodiments of the present disclosure is shown. The form 400 may be implemented at the information processing system 110. For ease of discussion, reference will be made to Figure 1 The form 400 is described with respect to the environment 100 .

[0057] like Figure 4 As shown, each merchant who has reached an agreement with the logistics company will have a unique identification code (K code) in the logistics system. In some examples, the K code can be expressed as K010, K021, K040, etc. In some embodiments, each merchant will have a registered email address and a registered mobile phone number when registering a mailing account. In other embodiments, when the merchant uses the mailing account for express or express transportation, there will be a retained mailing mobile phone number. In this case, the above K code, mailing account, registered email address, registered mobile phone number, and mailing mobile phone number can all be presented in the form of nodes. In some embodiments, the connected nodes can form edges, and the edges can represent the relationship between the business links corresponding to the connected nodes. In some examples, the edges can represent mailing information or registration information. In other examples, the business links may include mailing links and registration links. In this case, if the edge represents the mailing information, the mailing links corresponding to the connected nodes have the same information. If the edge represents the registration information, the registration links corresponding to the connected nodes have the same information. It can be understood that this is only exemplary and not restrictive.

[0058] For example, assuming there is a sending mobile phone number A, a registered mobile phone number G, a registered mailbox B, a registered mailbox H, a sending account number C, and a sending account number D, the K code includes K010, K012, K034, K033, K042, K040, K023, K018, K020, K0183, and K134. In some examples, the information contained in the K code may include a registered mobile phone number, a registered mailbox, a sending account number, a sending mobile phone number, a registered mobile phone number, etc. In other examples, the above object information is represented by a graph including nodes and edges. Specifically, node 410 represents the sending mobile phone number A, node 430 represents the registered mailbox B, node 440 represents the sending account number C, node 450 represents the registered mobile phone number G, node 460 represents the registered mailbox H, and node 470 represents the sending account number D. Further, node 421 represents K010, node 422 represents K040, node 423 represents K012, node 424 represents K023, node 431 represents K018, node 432 represents K034, node 433 represents K020, node 434 represents K033, node 441 represents K0183, node 442 represents K042, and node 443 represents K134. It can be understood that this is merely exemplary and not restrictive.

[0059] In some examples, the edge 401 connecting node 421 and node 420 can indicate that the mobile phone number for sending mail contained in K010 is the same as the mobile phone number for sending mail A, and the type of edge 401 is the information for sending mail. In some examples, the edge 402 connecting node 422 and node 420 can indicate that the mobile phone number for sending mail contained in K040 is the same as the mobile phone number for sending mail A, and the type of edge 402 is the information for sending mail. In other examples, the edge 403 connecting node 423 and node 420 can indicate that the mobile phone number for sending mail contained in K012 is the same as the mobile phone number for sending mail A, and the type of edge 403 is the information for sending mail. In more examples, the edge 404 connecting node 424 and node 420 can indicate that the mobile phone number for sending mail contained in K424 is the same as the mobile phone number for sending mail A, and the type of edge 404 is the information for sending mail. Furthermore, edge 405 connecting node 434 and node 420 may indicate that the sender mobile phone number included in K033 is the same as sender mobile phone number A, and the type of edge 405 is sender information. In addition, edge 406 connecting node 424 and node 433 may indicate that the sender mobile phone number included in K020 is the same as sender mobile phone number A, and the type of edge 406 is sender information.

[0060] More, edge 407 connecting node 423 and node 430 can indicate that the registration mailbox contained in K012 is the same as registration mailbox B, and the type of edge 407 is registration information. In some examples, edge 408 connecting node 431 and node 430 can indicate that the registration mailbox contained in K018 is the same as registration mailbox B, and the type of edge 408 is registration information. In other examples, edge 409 connecting node 432 and node 430 can indicate that the registration mailbox contained in K034 is the same as registration mailbox B, and the type of edge 409 is registration information. In more examples, edge 410 connecting node 433 and node 430 can indicate that the registration mailbox contained in K020 is the same as registration mailbox B, and the type of edge 410 is registration information. More, edge 410 connecting node 433 and node 430 can indicate that the registration mailbox contained in K020 is the same as registration mailbox B, and the type of edge 410 is registration information. In addition, the edge 411 connecting the node 443 and the node 430 may indicate that the registered email address included in K134 is the same as the registered email address B, and the type of the edge 411 is registration information.

[0061] In addition, the edge 412 connecting the node 432 and the node 440 can indicate that the mailing account number contained in K034 is the same as the mailing account number C, and the type of the edge 412 is mailing information. In some examples, the edge 413 connecting the node 424 and the node 440 can indicate that the mailing account number contained in K018 is the same as the mailing account number C, and the type of the edge 413 is mailing information. In other examples, the edge 414 connecting the node 441 and the node 440 can indicate that the mailing account number contained in K0183 is the same as the mailing account number C, and the type of the edge 414 is mailing information. In more examples, the edge 415 connecting the node 442 and the node 440 can indicate that the mailing account number contained in K042 is the same as the mailing account number C, and the type of the edge 415 is mailing information. More, the edge 416 connecting the node 443 and the node 440 can indicate that the mailing account number contained in K134 is the same as the mailing account number C, and the type of the edge 416 is mailing information.

[0062] In some examples, edge 417 connecting node 434 and node 470 may indicate that the mailing account number included in K033 is the same as mailing account number D, and the type of edge 417 is mailing information. Further, edge 418 connecting node 431 and node 450 may indicate that the registered mobile phone number included in K018 is the same as registered mobile phone number G, and the type of edge 418 is registration information. Further, edge 419 connecting node 441 and node 460 may indicate that the registered mailbox included in K0183 is the same as registered mailbox H, and the type of edge 418 is registration information.

[0063] In summary, the embodiments of the present disclosure can accurately determine the target object information set from the object information set based on the machine learning model and the classification machine learning model in the information processing system, and can also send early warning notifications in a timely manner. In this way, the accuracy and efficiency of determining the target object information set can be improved, and business risks can also be reduced.

[0064] The embodiments of the present disclosure also provide corresponding devices for implementing the above methods or processes.

[0065] Figure 5 A schematic structural block diagram of an apparatus 500 for determining object information according to some embodiments of the present disclosure is shown. The apparatus 500 may be implemented as or included in the information processing system 110. Each module / component in the apparatus 500 may be implemented by hardware, software, firmware, or any combination thereof.

[0066] like Figure 5 As shown, the device 500 includes a determination module 510, which is configured to determine an object information set, the object information set includes object information corresponding to at least one object, and the object information corresponding to each object is represented as a graph including nodes and edges, each node represents the object information used by each object in different business links of the business, and each edge represents the relationship between the business links corresponding to the connected nodes; the device 500 also includes an extraction module 520, which is configured to provide the object information set to the machine learning model to extract the node feature representation corresponding to the node and the edge feature representation corresponding to the edge; the device 500 also includes a utilization module 530, which is configured to determine the object feature representation corresponding to the object information based on the node feature representation and the edge feature representation using the machine learning model, and the object feature representation characterizes the risk feature of each object; the device 500 also includes a provision module 540, which is configured to provide the object feature representation to the classification machine learning model to determine the prediction result for each object in the object information set, and the prediction result indicates whether each object is classified as a risk object; the device 500 also includes an execution module 550, which is configured to select a target object information set from the object information set based on the prediction result, and the target object information set includes object information corresponding to at least one risk object.

[0067] In some embodiments, the execution module 550 is also configured to select risk object information indicated as a risk object by the prediction result from the object information set; determine the audit result of the risk object information based on the first audit rule; and select a target object information set from the risk object information based on the audit result.

[0068] In some embodiments, the object information set includes object information corresponding to at least one object that is related to a predetermined business and has a risk level greater than a risk threshold.

[0069] In some embodiments, the determination module 510 is also configured to obtain the object information set to be screened and the business type of the associated predetermined business; based on the business type, determine the risk condition for the object information set to be screened; and determine the object information in the object information set to be screened that meets the risk condition as an object with a risk level greater than a risk threshold; based on the object with a risk level greater than the risk threshold, determine the object information set.

[0070] In some embodiments, the business type includes a first type and a second type, and the determination module 510 is also configured to determine, in response to the business type indicating the first type, that the risk condition for the object information set to be screened is a first condition, the first type indicating that the business volume of at least one object in the object information set to be screened is less than a first business threshold, and the first condition indicating that the business indicator in the object information set to be screened is less than the first indicator threshold; and in response to the business type indicating the second type, determine that the risk condition for the object information set to be screened is a second condition, the second type indicating that the business volume of at least one object in the object information set to be screened is greater than a second business threshold, the second condition indicating that the business indicator in the object information set to be screened is less than the second indicator threshold, the first business threshold is less than or equal to the second business threshold, and the first indicator threshold is less than the second indicator threshold.

[0071] In some embodiments, the extraction module 520 is further configured to obtain a historical object information set; determine a training set from the historical object information set, the training set including historical object information corresponding to at least one object; and perform training on an untrained model based on the training set to determine a machine learning model.

[0072] In some embodiments, the extraction module 520 is also configured to, for a first object among at least one risk object, in response to identifying a newly added node, determine first node information corresponding to the newly added node in the object information corresponding to the first object; update the target object information set based on the first node information; and retrain the machine learning model based on the updated target object information set to update the parameters of the machine learning model.

[0073] In some embodiments, the extraction module 520 is also configured to determine second object information corresponding to the second object in response to identifying a newly added second object related to the business and having a risk level greater than a risk threshold; update the target object information set based on the second object information; and retrain the machine learning model based on the updated target object information set to update the parameters of the machine learning model.

[0074] In some embodiments, the execution module 550 is also configured to determine a second audit result of the object information of each object in the object information set based on a second audit rule, the second audit result indicating whether each object is a risk object or a non-risk object; discard the object information that the second audit result indicates is a non-risk object from the object information set to update the object information set; and select a target object information set from the updated object information set based on the prediction result.

[0075] In some embodiments, the execution module 550 is further configured to provide the target object information set to the object management system to trigger an early warning notification for at least one risk object in the object management system.

[0076] The units and / or modules included in the device 500 can be implemented in various ways, including software, hardware, firmware, or any combination thereof. In some embodiments, one or more units and / or modules can be implemented using software and / or firmware, such as machine executable instructions stored on a storage medium. In addition to or as an alternative to machine executable instructions, some or all of the units and / or modules in the device 500 can be implemented at least in part by one or more hardware logic components. As an example and not limitation, exemplary types of hardware logic components that can be used include field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), and the like.

[0077] It should be understood that one or more steps in the above method can be performed by a suitable electronic device or a combination of electronic devices. Such an electronic device or a combination of electronic devices may include, for example, Figure 1 The information processing system 110 in FIG.

[0078] Figure 6 1 shows a block diagram of an electronic device 600 in which one or more embodiments of the present disclosure may be implemented. It should be understood that Figure 6 The electronic device 600 shown is merely exemplary and should not constitute any limitation on the functionality and scope of the embodiments described herein. Figure 6 The electronic device 600 shown can be used to implement Figure 1 Information processing system 110 or Figure 5 Device 500.

[0079] like Figure 6As shown, the electronic device 600 is in the form of a general electronic device. The components of the electronic device 600 may include, but are not limited to, one or more processors or processing units 610, a memory 620, a storage device 630, one or more communication units 640, one or more input devices 650, and one or more output devices 660. The processing unit 610 may be an actual or virtual processor and is capable of performing various processes according to a program stored in the memory 620. In a multi-processor system, multiple processing units execute computer executable instructions in parallel to improve the parallel processing capability of the electronic device 600.

[0080] The electronic device 600 typically includes a plurality of computer storage media. Such media can be any available media accessible to the electronic device 600, including but not limited to volatile and non-volatile media, removable and non-removable media. The memory 620 can be a volatile memory (e.g., registers, caches, random access memory (RAM)), a non-volatile memory (e.g., a read-only memory (ROM), an electrically erasable programmable read-only memory (EEPROM), flash memory) or some combination thereof. The storage device 630 can be a removable or non-removable medium, and can include a machine-readable medium, such as a flash drive, a disk, or any other medium, which can be used to store information and / or data and can be accessed within the electronic device 600.

[0081] The electronic device 600 may further include additional removable / non-removable, volatile / non-volatile storage media. Figure 6 As shown in , a disk drive for reading or writing from a removable, non-volatile disk (e.g., a "floppy disk") and an optical drive for reading or writing from a removable, non-volatile optical disk may be provided. In these cases, each drive may be connected to the bus (not shown) by one or more data media interfaces. The memory 620 may include a computer program product 625 having one or more program modules that are configured to perform various methods or actions of various embodiments of the present disclosure.

[0082] The communication unit 640 implements communication with other electronic devices through a communication medium. Additionally, the functions of the components of the electronic device 600 can be implemented with a single computing cluster or multiple computing machines that can communicate through a communication connection. Therefore, the electronic device 600 can operate in a networked environment using a logical connection with one or more other servers, a network personal computer (PC), or another network node.

[0083] The input device 650 may be one or more input devices, such as a mouse, a keyboard, a tracking ball, etc. The output device 660 may be one or more output devices, such as a display, a speaker, a printer, etc. The electronic device 600 may also communicate with one or more external devices (not shown) through the communication unit 640 as needed, such as a storage device, a display device, etc., communicate with one or more devices that allow a user to interact with the electronic device 600, or communicate with any device that allows the electronic device 600 to communicate with one or more other electronic devices (e.g., a network card, a modem, etc.). Such communication may be performed via an input / output (I / O) interface (not shown).

[0084] According to an exemplary implementation of the present disclosure, a computer-readable storage medium is provided, on which computer-executable instructions are stored, wherein the computer-executable instructions are executed by a processor to implement the method described above. According to an exemplary implementation of the present disclosure, a computer program product is also provided, which is tangibly stored on a non-transitory computer-readable medium and includes computer-executable instructions, and the computer-executable instructions are executed by a processor to implement the method described above.

[0085] Various aspects of the present disclosure are described herein with reference to the flowcharts and / or block diagrams of the methods, devices, equipment, and computer program products implemented according to the present disclosure. It should be understood that each box in the flowchart and / or block diagram and the combination of each box in the flowchart and / or block diagram can be implemented by computer-readable program instructions.

[0086] These computer-readable program instructions can be provided to a processing unit 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 processing unit 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.

[0087] Computer-readable program instructions can be loaded onto a computer, other programmable data processing apparatus, or other device so that a series of operational steps are performed on the computer, other programmable data processing apparatus, or other device to produce a computer-implemented process, so that the instructions executed on the computer, other programmable data processing apparatus, or other device implement the functions / actions specified in one or more boxes in the flowchart and / or block diagram.

[0088] The flow chart and block diagram in the accompanying drawings show the possible architecture, function and operation of the system, method and computer program product according to multiple implementations of the present disclosure. In this regard, each square box in the flow chart or block diagram can represent a part of a module, program segment or instruction, and a part of a module, program segment or instruction includes one or more executable instructions for realizing the logical function of the specification. In some implementations as replacements, the function marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two continuous square boxes can actually be executed substantially in parallel, and they can sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be realized by a special hardware-based system that performs the function or action of the specification, or can be realized by a combination of special hardware and computer instructions.

[0089] The above descriptions of various implementations of the present disclosure are exemplary, non-exhaustive, and not limited to the disclosed implementations. 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 implementations. The selection of terms used herein is intended to best explain the principles of the implementations, practical applications, or improvements to the technology in the market, or to enable other persons of ordinary skill in the art to understand the various implementations disclosed herein.

Claims

1. A method for determining object information, comprising: Determine an object information set, wherein the object information set includes object information corresponding to at least one object, and the object information corresponding to each object is represented as a graph including nodes and edges, each node represents object information used by each object in different business links of the business, and each edge represents a relationship between business links corresponding to the connected nodes; Providing the object information set to a machine learning model to extract node feature representations corresponding to the nodes and edge feature representations corresponding to the edges; Based on the node feature representation and the edge feature representation, determining the object feature representation corresponding to the object information using the machine learning model, wherein the object feature representation represents the risk feature of each object; providing the object feature representation to a classification machine learning model to determine a prediction result for each object in the object information set, the prediction result indicating whether each object is classified as a risk object; as well as Based on the prediction result, a target object information set is selected from the object information set, the target object information set including object information corresponding to at least one risk object.

2. The method according to claim 1, wherein based on the prediction result, selecting a target object information set from the object information set comprises: selecting risk object information indicated as a risk object by the prediction result from the object information set; Based on the first audit rule, determining the audit result of the risk object information; as well as Based on the audit result, a target object information set is selected from the risk object information. 3 . The method according to claim 1 , wherein the object information set comprises object information corresponding to at least one object related to the predetermined business and having a risk level greater than a risk threshold.

4. The method of claim 3, wherein determining the object information set comprises: Acquire the object information set to be screened and the business type of the associated predetermined business; Based on the business type, determining a risk condition for the object information set to be screened; as well as Determine the object information that meets the risk condition in the object information set to be screened as objects whose risk level is greater than a risk threshold; The object information set is determined based on the objects whose risk levels are greater than a risk threshold.

5. The method according to claim 4, wherein the business type includes a first type and a second type, and based on the business type, determining the risk condition for the object information set to be screened includes: In response to the business type indicating a first type, determining that the risk condition for the object information set to be screened is a first condition, the first type indicating that the business volume of the at least one object in the object information set to be screened is less than a first business threshold, and the first condition indicating that the business index in the object information set to be screened is less than a first index threshold; as well as In response to the business type indicating a second type, the risk condition for the object information set to be screened is determined to be a second condition, the second type indicating that the business volume of at least one object in the object information set to be screened is greater than a second business threshold, the second condition indicating that the business indicator in the object information set to be screened is less than a second indicator threshold, the first business threshold is less than or equal to the second business threshold, and the first indicator threshold is less than the second indicator threshold.

6. The method according to claim 1, wherein the machine learning model is determined in the following manner: Get the historical object information set; Determine a training set from the historical object information set, the training set including the historical object information corresponding to the at least one object; and Training is performed on an untrained model based on the training set to determine the machine learning model.

7. The method according to claim 1, further comprising: For a first object among the at least one risk object, in response to identifying a newly added node, determining first node information corresponding to the newly added node in the object information corresponding to the first object; updating the target object information set based on the first node information; as well as Retrain the machine learning model based on the updated target object information set to update the parameters of the machine learning model.

8. The method according to claim 1, further comprising: In response to identifying a newly added second object related to the business and having a risk level greater than a risk threshold, determining second object information corresponding to the second object; updating the target object information set based on the second object information; as well as Retrain the machine learning model based on the updated target object information set to update the parameters of the machine learning model.

9. The method according to claim 1, wherein selecting a target object information set from the object information set comprises: Based on the second audit rule, determining a second audit result of the object information of each object in the object information set, the second audit result indicating whether each object is a risk object or a non-risk object; discarding object information indicating that the second audit result is a non-risk object from the object information set to update the object information set; as well as Based on the prediction result, the target object information set is selected from the updated object information set.

10. The method according to claim 1, further comprising: The target object information set is provided to an object management system to trigger an early warning notification for the at least one risk object in the object management system.

11. A device for determining object information, comprising: A determination module is configured to determine an object information set, wherein the object information set includes object information corresponding to at least one object, and the object information corresponding to each object is represented as a graph including nodes and edges, each node represents object information used by each object in different business links of the business, and each edge represents a relationship between business links corresponding to the connected nodes; an extraction module, configured to provide the object information set to a machine learning model to extract a node feature representation corresponding to the node and an edge feature representation corresponding to the edge; A utilizing module is configured to determine, based on the node feature representation and the edge feature representation, an object feature representation corresponding to the object information using the machine learning model, wherein the object feature representation represents a risk feature of each object; providing a module configured to provide the object feature representation to a classification machine learning model to determine a prediction result for each object in the object information set, the prediction result indicating whether each object is classified as a risk object; as well as The execution module is configured to select a target object information set from the object information set based on the prediction result, wherein the target object information set includes object information corresponding to at least one risk object.

12. An electronic device comprising: at least one processing unit; as well as At least one memory, the at least one memory is coupled to the at least one processing unit and stores instructions for execution by the at least one processing unit, the instructions, when executed by the at least one processing unit, causing the electronic device to perform the method according to any one of claims 1 to 10.

13. A computer-readable storage medium having a computer program stored thereon, wherein the computer program can be executed by a processor to implement the method according to any one of claims 1 to 10.

14. A computer program product comprising computer executable instructions, wherein the computer executable instructions, when executed by a processor, implement the method according to any one of claims 1 to 10.

Images