Information processing method, device, equipment and storage medium

Abstract

The present application provides an information processing method, apparatus, device, and storage medium, relating to the field of computer technology. The information processing method comprises: obtaining a data set to be processed, generating a relationship graph based on the data set; obtaining the characteristics of each node in the relationship graph and the characteristic value corresponding to the characteristic of each node, wherein the characteristic of each node corresponds to a target characteristic in a pre-trained decision tree model, and the decision tree model includes multiple tree nodes, and the tree nodes correspond to the target characteristics one-to-one; inputting the characteristic value corresponding to the characteristic of each node into the pre-trained decision tree model for decision processing, and obtaining the blacklist ratio corresponding to each node; and determining the risk level of the object corresponding to each node based on the blacklist ratio corresponding to each node. The present application can greatly improve the efficiency of screening and ensure the accuracy of the screening results.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to an information processing method, apparatus, device, and storage medium. Background Art

[0002] Relevant institutions store extensive data on personnel relationships, with each person corresponding to a node. This node can contain attributes such as identity information, contact information, and social media accounts. Nodes can be connected through relationships such as kinship, calling, and affiliation. The nodes and their relationships form a vast, network-like graph. If a node in the graph is implicated in a related incident, other nodes with close connections to it are likely also implicated. This allows for the identification of suspicious individuals and their subsequent investigation.

[0003] Currently, if someone is suspected of being involved in an incident, the process of investigating whether their associates are involved is typically done manually in a query system that uses the graph data to identify relevant nodes based on their relationships. This process then involves querying other nodes, one by one. If all the nodes surrounding a particular node are found to be involved in the incident, the probability that the person associated with that node is also involved is very high, and this person is then prioritized for investigation. However, this manual investigation method is inefficient. Summary of the Invention

[0004] The present application provides an information processing method, apparatus, device and storage medium to solve the problem of low efficiency of manual investigation.

[0005] In a first aspect, the present application provides an information processing method, comprising:

[0006] Obtaining a data set to be processed, where the data set includes multiple objects to be processed;

[0007] Generate a relationship graph based on the data set. The relationship graph contains nodes and the relationship between nodes, where the nodes correspond one-to-one to the objects in the data set.

[0008] Obtain the features of each node in the relationship graph and the feature values ​​corresponding to the features of each node, where the features of each node correspond to the target features in the pre-trained decision tree model. The decision tree model contains multiple tree nodes, and the tree nodes correspond to the target features one by one.

[0009] The feature value corresponding to each node's feature is input into the pre-trained decision tree model for decision processing to obtain the blacklist ratio corresponding to each node;

[0010] According to the blacklist ratio corresponding to each node, the risk level of the object corresponding to each node is determined.

[0011] Optionally, the characteristic values ​​corresponding to the features of each node are input into a pre-trained decision tree model for decision processing to obtain the blacklist ratio corresponding to each node, including: inputting the characteristic values ​​corresponding to the features of each node into a pre-trained decision tree model, comparing the characteristic values ​​corresponding to the features of each node with the cutoff values ​​corresponding to the target features in the pre-trained decision tree model, and determining the decision path corresponding to each node; and obtaining the blacklist ratio corresponding to each node based on the blacklist objects corresponding to the leaf nodes in the decision path.

[0012] Optionally, the features of each node in the relationship graph and the feature values ​​corresponding to the features of each node are obtained, including: clustering each node in the relationship graph to obtain the clustering results corresponding to each node in the relationship graph; based on the clustering results and the target features, the features of each node in the relationship graph and the feature values ​​corresponding to the features of each node are obtained.

[0013] Optionally, the risk level of the object corresponding to each node is determined based on the blacklist ratio corresponding to each node, including: if the blacklist ratio is greater than or equal to a first threshold, the risk level is determined to be a high risk level; if the blacklist ratio is less than the first threshold and greater than or equal to a second threshold, the risk level is determined to be a medium risk level, and the second threshold is less than the first threshold; if the blacklist ratio is less than the second threshold, the risk level is determined to be a low risk level.

[0014] In a second aspect, the present application provides a method for training a decision tree model, comprising:

[0015] Obtain a training data set, which contains positive samples and negative samples. Positive samples are samples that do not contain blacklisted samples, and negative samples are blacklisted samples.

[0016] Generate a sample relationship graph based on the positive and negative samples in the training data set;

[0017] Perform clustering processing on the nodes in the sample relationship graph to obtain the sample clustering results corresponding to the nodes in the sample relationship graph;

[0018] According to the sample clustering results and the preset sample features, the sample features of each node in the sample relationship graph and the corresponding feature values ​​of the sample features are obtained. The sample features of each node correspond to the preset sample features one by one.

[0019] Determine the information value corresponding to the preset sample feature according to the feature value corresponding to the sample feature of each node;

[0020] Determine the target feature from the preset sample features according to the information value corresponding to the preset sample feature;

[0021] According to the training data set and the target feature, the initial decision tree model is iteratively trained to obtain a decision tree model, wherein the decision tree model includes the target feature and the cutoff value corresponding to the target feature.

[0022] Optionally, after obtaining the training data set, the training method of the decision tree model further includes: preprocessing the positive samples in the training data set to obtain preprocessed positive samples, wherein the preprocessing is credibility filtering of the source channels of the positive samples and data cleaning of the positive samples.

[0023] In a third aspect, the present application provides an information processing device, comprising:

[0024] A first acquisition module is used to acquire a data set to be processed, wherein the data set includes a plurality of objects to be processed;

[0025] A generation module is used to generate a relationship graph based on the data set, where the relationship graph includes nodes and the relationship between nodes, where the nodes correspond one-to-one to the objects in the data set;

[0026] The second acquisition module is used to obtain the features of each node in the relationship graph and the feature value corresponding to the feature of each node, wherein the feature of each node corresponds to the target feature in the pre-trained decision tree model, and the decision tree model includes multiple tree nodes, and the tree nodes correspond to the target features one by one;

[0027] A processing module is used to input the feature value corresponding to each node into a pre-trained decision tree model for decision processing to obtain the blacklist ratio corresponding to each node;

[0028] The determination module is used to determine the risk level of the object corresponding to each node based on the blacklist ratio corresponding to each node.

[0029] Optionally, the processing module is specifically used to: input the characteristic value corresponding to the feature of each node into a pre-trained decision tree model, compare the characteristic value corresponding to the feature of each node with the cutoff value corresponding to the target feature in the pre-trained decision tree model, and determine the decision path corresponding to each node; according to the blacklist object corresponding to the leaf node in the decision path, obtain the blacklist ratio corresponding to each node.

[0030] Optionally, the second acquisition module is specifically used to: perform clustering processing on each node in the relationship graph to obtain the clustering results corresponding to each node in the relationship graph; based on the clustering results and the target features, obtain the features of each node in the relationship graph and the feature values ​​corresponding to the features of each node.

[0031] Optionally, the determination module is specifically used to: if the blacklist ratio is greater than or equal to a first threshold, determine the risk level as a high risk level; if the blacklist ratio is less than the first threshold and greater than or equal to a second threshold, determine the risk level as a medium risk level, and the second threshold is less than the first threshold; if the blacklist ratio is less than the second threshold, determine the risk level as a low risk level.

[0032] In a fourth aspect, the present application provides a training device for a decision tree model, comprising:

[0033] The first acquisition module is used to obtain a training data set, where the training data set includes positive samples and negative samples. The positive samples are samples that do not contain blacklist items, and the negative samples are blacklist samples.

[0034] A generation module is used to generate a sample relationship graph based on the positive samples and negative samples in the training data set;

[0035] The second acquisition module is used to perform clustering processing on the nodes in the sample relationship graph to obtain the sample clustering results corresponding to the nodes in the sample relationship graph;

[0036] The third acquisition module is used to obtain the sample features and feature values ​​corresponding to each node in the sample relationship graph based on the sample clustering results and the preset sample features. The sample features of each node correspond to the preset sample features one-to-one;

[0037] A first determining module, configured to determine an information value corresponding to a preset sample feature according to a feature value corresponding to the sample feature of each node;

[0038] A second determining module is used to determine the target feature from the preset sample features according to the information value corresponding to the preset sample feature;

[0039] The training module is used to iteratively train the initial decision tree model according to the training data set and the target feature to obtain a decision tree model, wherein the decision tree model includes the target feature and the cutoff value corresponding to the target feature.

[0040] Optionally, after acquiring the training data set, the first acquisition module is further used to: preprocess the positive samples in the training data set to obtain preprocessed positive samples, wherein the preprocessing is to perform credibility filtering on the source channels of the positive samples and to perform data cleaning on the positive samples.

[0041] In a fifth aspect, the present application provides an electronic device, comprising: a processor, and a memory communicatively connected to the processor;

[0042] Memory stores computer-executable instructions;

[0043] The processor executes the computer-executable instructions stored in the memory to implement the information processing method as described in the first aspect of the present application.

[0044] In a sixth aspect, the present application provides an electronic device, comprising: a processor, and a memory communicatively connected to the processor;

[0045] Memory stores computer-executable instructions;

[0046] The processor executes the computer-executable instructions stored in the memory to implement the training method of the decision tree model as described in the second aspect of the present application.

[0047] In a seventh aspect, the present application provides a computer-readable storage medium, in which computer program instructions are stored. When the computer program instructions are executed by a processor, the information processing method described in the first aspect of the present application is implemented.

[0048] In an eighth aspect, the present application provides a computer-readable storage medium, in which computer program instructions are stored. When the computer program instructions are executed by a processor, the training method of the decision tree model as described in the second aspect of the present application is implemented.

[0049] In a ninth aspect, the present application provides a computer program product, comprising a computer program, which, when executed by a processor, implements the information processing method as described in the first aspect of the present application.

[0050] In a tenth aspect, the present application provides a computer program product, comprising a computer program, which, when executed by a processor, implements the training method of the decision tree model as described in the second aspect of the present application.

[0051] The information processing method, apparatus, device and storage medium provided by the present application obtain a data set to be processed, and generate a relationship graph based on the data set; obtain the characteristics of each node in the relationship graph and the characteristic value corresponding to the characteristics of each node, wherein the characteristics of each node correspond to the target characteristics in a pre-trained decision tree model, and the decision tree model includes multiple tree nodes, and the tree nodes correspond to the target characteristics one by one; the characteristic value corresponding to the characteristics of each node is input into the pre-trained decision tree model for decision processing to obtain the blacklist ratio corresponding to each node; according to the blacklist ratio corresponding to each node, the risk level of the object corresponding to each node is determined. Since the present application automatically determines the risk level of the object in the data set based on the pre-trained decision tree model, without the need for manual screening, the screening efficiency can be greatly improved; the pre-trained decision tree model can output the blacklist ratio corresponding to each node based on the characteristic value corresponding to each node in the relationship graph generated by the data set, so as to determine the risk level of the object in the data set corresponding to the node, instead of directly using the yes or no calculation result output by the current decision tree model, so as to ensure the accuracy of the screening result. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0053] Figure 1 A schematic diagram of an application scenario provided in an embodiment of the present application;

[0054] Figure 2 A flowchart of an information processing method provided in one embodiment of the present application;

[0055] Figure 3 A schematic diagram of a pre-trained decision tree model provided in one embodiment of the present application;

[0056] Figure 4 A flowchart of an information processing method provided in another embodiment of the present application;

[0057] Figure 5 The process of the training method of the decision tree model provided in one embodiment of the present application;

[0058] Figure 6 A schematic diagram of the structure of an information processing device provided in one embodiment of the present application;

[0059] Figure 7A schematic diagram of the structure of a training device for a decision tree model provided in one embodiment of the present application;

[0060] Figure 8 A schematic diagram of the structure of an electronic device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0061] To make the purpose, technical solutions, and advantages of the embodiments of this application more clear, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0062] In the technical solution of this application, the collection, storage, use, processing, transmission, provision and disclosure of information such as financial data or user data involved comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0063] First, some technical terms involved in this application are explained:

[0064] Graph computing: A graph is an abstract data structure used to represent the relationship between objects. It is described using vertices and edges. Vertices represent objects, and edges represent the relationship between objects. Data that can be abstracted into a graph is called graph data. Graph computing is the process of using graphs as data models to express and solve problems.

[0065] The Louvain clustering algorithm is a modularity-based community discovery algorithm that is efficient and effective, and can discover hierarchical community structures. Its optimization goal is to maximize the modularity of the entire community network.

[0066] PageRank, also known as Google's page ranking algorithm, is used to measure the importance of a particular web page relative to other web pages in the search engine index. It is used in graph computing to represent the importance of a node in the graph.

[0067] N-degree association, that is, all node relationships that have a first-degree relationship with the selected node are called first-degree association, all node relationships that have a second-degree relationship with the selected node are called second-degree association, and so on;

[0068] A decision tree is a tree structure in which each non-leaf node represents a test on a feature attribute, each branch represents the output of this feature attribute over a certain value range, and each leaf node stores a category. The decision-making process using a decision tree starts from the root node, tests the corresponding feature attributes of the item to be classified, and selects the output branch according to its value until a leaf node is reached. The category stored in the leaf node is used as the decision result.

[0069] Information Value (IV) is mainly used to encode input variables and evaluate their predictive power in binary classification problems in machine learning. The size of the information value indicates the strength of the input variable's predictive power, and the value range of the information value is [0, positive infinity). If the current group only contains responding customers or non-responding customers, the information value is positive infinity. The specific quantitative indicators include: if the information value is <0.02, it means that the input variable is not suitable for prediction; if 0.02≤information value<0.1, it means that the input variable has weak predictive power; if 0.1≤information value<0.3, it means that the input variable has medium predictive power; if 0.3≤information value<0.5, it means that the input variable has strong predictive power; if the information value is >0.5, it means that the input variable has super strong predictive power.

[0070] Currently, if someone is identified as being involved in an incident, such as being a non-compliant or illegal individual, manual screening is typically performed to determine if their connections are also non-compliant or illegal. When there are many connections, checking each one individually takes a significant amount of time. Furthermore, searching for secondary connections takes even more time, resulting in low efficiency. Furthermore, when there are many connected nodes, manual screening doesn't reveal which nodes pose the greatest risk. Without a quantitative risk score for each node, the process is blind and inefficient, preventing the rapid and accurate identification of suspicious individuals.

[0071] Based on the above problems, the present application provides an information processing method, apparatus, device and storage medium, which generates a relationship graph based on the data set to be processed. Based on the relationship graph, a pre-trained decision tree model is used to determine the risk level of the object in the data set corresponding to each node in the relationship graph, which can greatly improve the screening efficiency.

[0072] Below, the application scenarios of the solution provided in this application are first illustrated.

[0073] Figure 1 This is a schematic diagram of an application scenario provided by an embodiment of the present application. Figure 1As shown, in this application scenario, server 102 obtains a dataset to be processed from client 101. Server 102 determines the risk level of objects in the dataset based on the dataset to be processed, and sends the risk level of the objects in the dataset to client 101 for display. The specific implementation process of server 102 determining the risk level of objects in the dataset based on the dataset to be processed can be found in the solutions of the following embodiments.

[0074] It should be noted that Figure 1 This is only a schematic diagram of an application scenario provided by the embodiment of the present application. Figure 1 The equipment included in the Figure 1 The positional relationship between the devices in the Figure 1 The application scenario shown may also include a data storage device, which may be an external memory relative to the client 101 or the server 102 , or an internal memory integrated in the client 101 or the server 102 .

[0075] Next, the information processing method is introduced through specific embodiments.

[0076] Figure 2 This is a flow chart of an information processing method provided in one embodiment of the present application. The method of the embodiment of the present application can be applied to an electronic device, which can be a server or a server cluster. Figure 2 As shown, the method of the embodiment of the present application includes:

[0077] S201: Obtain a data set to be processed.

[0078] The data set includes multiple objects to be processed.

[0079] In the embodiment of the present application, the data set to be processed is, for example, sent by another device to the electronic device executing the embodiment of the present method. The data set to be processed includes multiple objects to be processed, wherein the objects to be processed are, for example, blacklist objects, specifically, non-compliant or illegal persons; the objects to be processed are, for example, suspicious objects, specifically, potentially non-compliant or illegal persons; and the objects to be processed are, for example, compliant and legal objects.

[0080] S202: Generate a relationship diagram based on the data set.

[0081] The relationship graph includes nodes and the relationships between nodes, where nodes correspond one-to-one to objects in the dataset.

[0082] In this step, after obtaining the dataset to be processed, a relationship graph can be generated based on the dataset. For example, each object in the dataset is treated as a node in the relationship graph; the associations between objects in the dataset are treated as the associations between nodes in the relationship graph. The associations between nodes in the relationship graph can be, for example, kinship, calling, or subordinate relationships, which are not specifically limited in this application; based on the blacklist objects in the dataset, the nodes corresponding to the blacklist objects are marked as blacklist nodes; and the relationship graph is generated based on the nodes and the associations between them.

[0083] S203. Obtain the features of each node in the relationship graph and the feature values ​​corresponding to the features of each node, wherein the features of each node correspond to the target features in a pre-trained decision tree model. The decision tree model includes multiple tree nodes, and the tree nodes correspond to the target features one by one.

[0084] In this step, the decision tree model is pre-trained. For details on how to train the decision tree model, please refer to the subsequent embodiments and will not be described here. The tree nodes in the decision tree model may include root nodes, internal nodes, and leaf nodes. After the relationship graph is generated, the features of each node in the relationship graph and the feature values ​​corresponding to the features of each node can be obtained, wherein the features of each node correspond one-to-one to the target features in the pre-trained decision tree model. For example, Figure 3 A schematic diagram of a pre-trained decision tree model provided in an embodiment of the present application is shown in FIG. Figure 3 As shown, the pre-trained decision tree model includes tree nodes a, b, and c, where a is the root node, b and c are both leaf nodes, and the target features corresponding to the tree nodes are a, b, and c. Correspondingly, the features of each node in the relationship graph are a, b, and c. After determining that the features of each node in the relationship graph are a, b, and c, the feature values ​​corresponding to the three features a, b, and c for each node in the relationship graph can be obtained. For specific information on how to obtain the feature values ​​corresponding to the features of each node in the relationship graph, please refer to the subsequent embodiments and will not be repeated here.

[0085] S204: Input the feature value corresponding to the feature of each node into a pre-trained decision tree model for decision processing to obtain the blacklist ratio corresponding to each node.

[0086] In this step, after obtaining the characteristic values ​​corresponding to the features of each node in the relationship graph, the characteristic values ​​corresponding to the features of each node can be input into a pre-trained decision tree model for decision processing to obtain the blacklist ratio corresponding to each node. For example, a decision processing is performed based on the characteristic values ​​corresponding to the features of each node and the cutoff value corresponding to the target feature in the pre-trained decision tree model to obtain the blacklist ratio corresponding to each node. For details on how to obtain the blacklist ratio corresponding to each node, please refer to the subsequent embodiments and will not be repeated here.

[0087] S205: Determine the risk level of the object corresponding to each node according to the blacklist ratio corresponding to each node.

[0088] In this step, after obtaining the blacklist ratio corresponding to each node, the risk level of the object in the data set corresponding to the node can be determined based on the blacklist ratio. For example, the risk level of the object in the data set corresponding to the node can be determined based on the threshold range to which the blacklist ratio belongs, wherein different threshold ranges correspond to different risk levels. For specific information on how to determine the risk level of the object corresponding to the node based on the blacklist ratio, please refer to the subsequent embodiments and will not be repeated here. After determining the risk level of the object corresponding to the node, the risk level of the object in the data set is obtained, and the risk level of the object in the data set can be sent to the client for display, so as to determine suspicious persons based on the risk level of the object in the data set and conduct focused investigation on the suspicious persons.

[0089] The information processing method provided in the embodiment of the present application obtains a data set to be processed, generates a relationship graph based on the data set; obtains the characteristics of each node in the relationship graph and the characteristic value corresponding to the characteristics of each node, wherein the characteristics of each node correspond to the target characteristics in a pre-trained decision tree model, and the decision tree model includes multiple tree nodes, and the tree nodes correspond to the target characteristics one by one; the characteristic value corresponding to the characteristics of each node is input into the pre-trained decision tree model for decision processing to obtain the blacklist ratio corresponding to each node; and the risk level of the object corresponding to each node is determined based on the blacklist ratio corresponding to each node. Since the embodiment of the present application automatically determines the risk level of the object in the data set based on the pre-trained decision tree model, without the need for manual screening, the screening efficiency can be greatly improved; the pre-trained decision tree model can output the blacklist ratio corresponding to each node based on the characteristic value corresponding to each node in the relationship graph generated by the data set, so as to determine the risk level of the object in the data set corresponding to the node, instead of directly using the yes or no calculation result output by the current decision tree model, so as to ensure the accuracy of the screening result.

[0090] Figure 4 This is a flowchart of an information processing method provided by another embodiment of the present application. Based on the above embodiment, this embodiment of the present application further explains how to perform information processing. Figure 4 As shown, the method of the embodiment of the present application may include:

[0091] S401: Obtain a data set to be processed.

[0092] The detailed description of this step can be found in Figure 2 The relevant description of S201 in the illustrated embodiment will not be repeated here.

[0093] S402: Preprocess the data set to obtain a preprocessed data set.

[0094] Among them, preprocessing includes credibility filtering of the source channels of the data set and data cleaning of the data set.

[0095] For example, the source of a dataset may be a carrier. The dataset can be filtered for credibility based on its source. Data cleaning can include filtering objects with empty information in the dataset to remove invalid data. By preprocessing the dataset, a high-quality dataset can be obtained, which can be considered a credible dataset.

[0096] S403: Generate a relationship graph based on the preprocessed data set.

[0097] In this step, after obtaining the preprocessed data set, a relationship graph can be generated based on the preprocessed data set. For details on how to generate the relationship graph, see Figure 2 The relevant description of S202 in the illustrated embodiment will not be repeated here.

[0098] In the embodiment of this application, Figure 2 The step S203 may further include the following two steps S404 and S405:

[0099] S404: Perform clustering processing on each node in the relationship graph to obtain a clustering result corresponding to each node in the relationship graph.

[0100] In this step, after the relationship graph is generated, each node in the relationship graph can be clustered to obtain the corresponding clustering results.

[0101] Optionally, performing clustering processing on each node in the relationship graph to obtain a corresponding clustering result may include: performing clustering processing on each node in the relationship graph using a Louvain clustering algorithm to obtain a corresponding clustering result.

[0102] For example, a Louvain clustering algorithm uses a relationship graph as input and clusters each node in the graph using the Louvain clustering algorithm to obtain the corresponding clustering results. For example, if the relationship graph contains 100 nodes, the Louvain clustering algorithm can obtain 5 clusters, each containing the corresponding node. It can be understood that clustering can group closely connected nodes into a single cluster.

[0103] S405: Based on the clustering result and the target feature, obtain the feature of each node in the relationship graph and the feature value corresponding to the feature of each node.

[0104] In this step, after obtaining the clustering results, the features of each node in the relationship graph and the feature values ​​corresponding to the features of each node can be obtained based on the clustering results and the target features in the pre-trained decision tree model. It should be noted that the target features in the pre-trained decision tree model may include in-group features or out-of-group features, wherein the in-group features include, for example, the number of people in the group to which the node belongs, the number of people on the key list of the group to which the node belongs, the proportion of the key list of the group to which the node belongs, the number of people who are first-degree associated with the group to which the node belongs, or at least one of the number of people on the key list of the group to which the node belongs; out-of-group features include, for example, the number of all first-degree associated with the node, the number of all first-degree associated key lists of the node, the PageRank value of the node, or the PageRank value ranking of the group to which the node belongs. The key list is, for example, a blacklist. For specific information on how to obtain the target features in the pre-trained decision tree model, please refer to the subsequent embodiments, which will not be repeated here.

[0105] For example, refer to Figure 3 The pre-trained decision tree model contains three target features: a, b, and c. Assume that a represents the number of blacklisted members of the node's group, b represents the number of first-degree connections within the node's group, and c represents the node's PageRank value. Assume that the pre-processed dataset contains 100 objects, 8 of which are blacklisted. Accordingly, the relationship graph contains 100 nodes, 8 of which are blacklisted. Each node in the relationship graph is clustered to obtain the corresponding clustering result. For example, if it is divided into 5 clusters, taking the first cluster as an example, assuming that the first cluster contains 10 nodes, 8 of which are blacklisted nodes, then the feature value of target feature a for each node in the first cluster is 8, that is, the number of blacklisted members in the first cluster is 8. Based on the number of first-degree connections of each node in the first cluster, the feature value of target feature b for each node can be determined. Through the PageRank algorithm, the feature value of target feature c for each node can be obtained.

[0106] In the embodiment of this application, Figure 2 Step S204 may further include the following two steps S406 and S407:

[0107] S406. Input the feature value corresponding to the feature of each node into the pre-trained decision tree model, compare the feature value corresponding to the feature of each node with the cutoff value corresponding to the target feature in the pre-trained decision tree model, and determine the decision path corresponding to each node.

[0108] S407 : Obtain the blacklist ratio corresponding to each node according to the blacklist objects corresponding to the leaf nodes in the decision path.

[0109] For example, refer to Figure 3In the pre-trained decision tree model, each target feature has a corresponding cutoff value, that is, the cutoff value corresponding to a is 100, the cutoff value corresponding to b is 50, and the cutoff value corresponding to c is 200. Each cutoff value is determined during the training of the decision tree model. If the eigenvalue corresponding to a input into the pre-trained decision tree model is greater than the cutoff value 100 corresponding to a, the result of the corresponding branch is: not blacklisted; if the eigenvalue corresponding to a input into the pre-trained decision tree model is less than or equal to the cutoff value 100 corresponding to a, then go from the root node a to the leaf node b; if the eigenvalue corresponding to b input into the pre-trained decision tree model is greater than or equal to the cutoff value 50 corresponding to b, then the result of the corresponding branch is: it is blacklisted, and the blacklist ratio is output; if the eigenvalue corresponding to b input into the pre-trained decision tree model is less than the cutoff value 50 corresponding to b, then go from the leaf node b to the leaf node c; if the eigenvalue corresponding to c input into the pre-trained decision tree model is greater than the cutoff value 200 corresponding to c, then the result of the corresponding branch is: it is not blacklisted; if the eigenvalue corresponding to c input into the pre-trained decision tree model is less than or equal to the cutoff value 200 corresponding to c, then the result of the corresponding branch is: it is blacklisted, and the blacklist ratio is output. In this step, the characteristic values ​​of the three target features a, b, and c corresponding to each node are respectively input into the pre-trained decision tree model for decision processing, and the blacklist ratio corresponding to each node can be obtained. Specifically, for example, if the characteristic value of a corresponding to a node is 98, the characteristic value of b corresponding to b is 55, and the characteristic value of c corresponding to c is 80, the characteristic values ​​of the three target features a, b, and c corresponding to the node are compared with the cutoff values ​​corresponding to a, b, and c respectively. Since the characteristic value of a corresponding to the node is 98, which is less than the cutoff value of a corresponding to 100, then when going from the root node a to the leaf node b, the characteristic value of b corresponding to the node is 55, which is greater than the cutoff value of b corresponding to 50. Therefore, the result of the corresponding branch is: it is blacklisted, that is, it can be determined that the node corresponds to the blacklist branch of the pre-trained decision tree model, which is the decision path from the root node a to the leaf node b. Suppose a dataset contains eight blacklisted objects. The decision path for the nodes corresponding to the ten objects in the dataset, including the eight blacklisted objects, is as follows: from the root node a to the blacklist branch of the pre-trained decision tree model at leaf node b. It can be determined that the blacklist proportion corresponding to the blacklist branch at leaf node b is 80%, meaning that the blacklist proportion for all ten objects in the dataset is 80%. If the feature value corresponding to a for a node is 105, then that node is not a blacklisted node.

[0110] It can be understood that in a pre-trained decision tree model, each node from the root to each leaf corresponds to a different risk rule. For example, the blacklist branch from root node a to leaf node b has a corresponding blacklist ratio of 80%, which is the highest blacklist ratio. Therefore, the decision path from root node a to leaf node b can be considered a high-risk rule, i.e., a <= 100 and b >= 50. The blacklist branch from root node a to leaf node c has a corresponding blacklist ratio of 40%. Therefore, the decision path from root node a to leaf node c can be considered a low-risk rule. Therefore, the decision results of the pre-trained decision tree model include whether the node is blacklisted and the blacklist ratio. Based on the blacklist ratio corresponding to the leaf node, high-risk rules, medium-risk rules, and low-risk rules can be determined, respectively.

[0111] In the embodiment of this application, Figure 2 Step S205 may further include the following three steps S408 to S410:

[0112] S408: If the blacklist ratio is greater than or equal to the first threshold, determine that the risk level is a high risk level.

[0113] S409: If the blacklist ratio is less than the first threshold and greater than or equal to the second threshold, determine that the risk level is a medium risk level.

[0114] The second threshold is smaller than the first threshold.

[0115] S410: If the blacklist ratio is less than a second threshold, the risk level is determined to be a low risk level.

[0116] For example, refer to Figure 3 , the first threshold is 70%, the second threshold is 50%, if the blacklist ratio of the node is 80%, then the risk level of the object corresponding to the node can be determined to be a high risk level; if the blacklist ratio of the node is 40%, then the risk level of the object corresponding to the node can be determined to be a low risk level; if the blacklist ratio of the node is 60% ( Figure 3 (not shown), the risk level of the object corresponding to the node can be determined to be medium. After determining the risk level of the object corresponding to the node, the risk level of the object in the dataset is obtained. The risk level of the object in the dataset can be sent to the client for display. Suspicious individuals can be identified based on the risk level of the object in the dataset and then be screened for priority. Specifically, for example, high-risk individuals are screened for priority.

[0117] The information processing method provided in the embodiment of the present application obtains a data set to be processed, preprocesses the data set, and obtains a preprocessed data set; generates a relationship graph based on the preprocessed data set; performs clustering processing on each node in the relationship graph to obtain a clustering result corresponding to each node in the relationship graph, and obtains the feature of each node in the relationship graph and the feature value corresponding to the feature of each node based on the clustering result and the target feature; inputs the feature value corresponding to the feature of each node into a pre-trained decision tree model, compares the feature value corresponding to the feature of each node with the cutoff value corresponding to the target feature in the pre-trained decision tree model, and determines the decision path corresponding to each node; obtains the blacklist ratio corresponding to each node based on the blacklist object corresponding to the leaf node in the decision path; if the blacklist ratio is greater than or equal to the first threshold, determines that the risk level of the object corresponding to the node is a high risk level; if the blacklist ratio is less than the first threshold and greater than or equal to the second threshold, determines that the risk level of the object corresponding to the node is a medium risk level; if the blacklist ratio is less than the second threshold, determines that the risk level of the object corresponding to the node is a low risk level. Since the embodiment of the present application performs clustering processing on each node in the relationship graph, closely connected nodes can be grouped into one cluster, and the characteristic value of the target feature in the pre-trained decision tree model corresponding to each node is determined based on the clustering result, and the characteristic value of each node is compared with the cutoff value corresponding to the target feature in the pre-trained decision tree model to determine the decision path corresponding to each node, and then determine the blacklist ratio corresponding to each node, thereby determining the risk level of the object in the data set corresponding to the node, thereby greatly improving the screening efficiency; based on the cutoff value corresponding to the target feature in the pre-trained decision tree model, the decision path corresponding to each node can be determined, and the blacklist ratio corresponding to each node can be output to determine the risk level of the object in the data set corresponding to the node, instead of directly using the yes or no calculation result output by the current decision tree model, thereby ensuring the accuracy of the screening results.

[0118] Based on the above embodiments, Figure 5 A flowchart of a training method for a decision tree model provided in one embodiment of the present application is shown in FIG. Figure 5 As shown, the method of the embodiment of the present application may include:

[0119] S501. Obtain a training data set, where the training data set includes positive samples and negative samples. Positive samples are samples that do not contain blacklisted samples, and negative samples are blacklisted samples.

[0120] In this step, the positive samples included in the training data set are samples that do not contain blacklist items, that is, the positive samples are samples whose risk levels are to be determined.

[0121] S502: Generate a sample relationship graph based on the positive samples and negative samples in the training data set.

[0122] In this step, after obtaining the training dataset, a sample relationship graph can be generated based on the positive and negative samples in the training dataset. For example, each positive or negative sample in the training dataset is treated as a node in the relationship graph; the associations between samples in the training dataset are used as the inter-node associations in the sample relationship graph. The inter-node associations in the sample relationship graph may be, for example, kinship, calling, or subordinate relationships, which are not specifically limited in this application; based on the negative samples in the training dataset, the nodes corresponding to the negative samples are marked as blacklisted nodes; and the sample relationship graph is generated based on the nodes and the inter-node associations.

[0123] Optionally, after obtaining the training data set, the training method of the decision tree model may further include: preprocessing the positive samples in the training data set to obtain preprocessed positive samples, wherein the preprocessing is credibility filtering of the source channels of the positive samples and data cleaning of the positive samples.

[0124] For example, referring to step S402 above, if the source of the positive sample is, for example, from a carrier, credibility filtering can be performed based on the source of the positive sample. Data cleaning can be performed based on whether the positive sample information is empty, thereby filtering out credible data, thus obtaining preprocessed positive samples. A sample relationship graph can then be generated based on the preprocessed positive and negative samples.

[0125] S503: Perform clustering processing on the nodes in the sample relationship graph to obtain sample clustering results corresponding to the nodes in the sample relationship graph.

[0126] Exemplarily, referring to the above step S404, for example, the nodes in the sample relationship graph may be clustered using the Louvain clustering algorithm to obtain sample clustering results corresponding to the nodes in the sample relationship graph.

[0127] S504: Obtain the sample features of each node in the sample relationship graph and the feature values ​​corresponding to the sample features according to the sample clustering results and the preset sample features. The sample features of each node correspond one-to-one to the preset sample features.

[0128] In this step, the preset sample features include at least one of the above-mentioned in-group features and out-of-group features. For example, referring to the first group in the above-mentioned S405 step example, for each node in the group, the sample features of each node and the feature values ​​corresponding to the sample features can be obtained as follows: (1) In-group features: the number of people in the group to which the node belongs is 10; the number of people on the blacklist of the group to which the node belongs is 8; the blacklist ratio of the group to which the node belongs is 80%; based on the number of first-degree related people of each node in the first group, the number of first-degree related people in the group to which the node belongs can be obtained; based on the number of blacklisted people among the first-degree related people of each node in the first group, the number of first-degree related blacklisted people in the group to which the node belongs can be obtained; (2) Out-of-group features: based on the number of first-degree related people of each node in the first group with nodes in other groups, the number of all first-degree related people of each node in the first group can be obtained; based on the number of blacklisted people among the first-degree related people of each node in the first group with nodes in other groups, the number of all first-degree related blacklisted people of each node in the first group can be obtained; through the PageRank algorithm, the node PageRank value can be obtained; based on the node PageRank value, the PageRank value ranking of the node in the group to which the node belongs can be obtained.

[0129] S505: Determine the information value corresponding to the preset sample feature according to the feature value corresponding to the sample feature of each node.

[0130] Exemplarily, after obtaining the sample features of each node in the sample relationship graph and the eigenvalues ​​corresponding to the sample features, the information value corresponding to the preset sample features can be determined based on the eigenvalues ​​corresponding to the sample features of each node. Specifically, when determining the information value corresponding to each preset sample feature, the eigenvalues ​​corresponding to the sample features of each node in the sample relationship graph are used as input features of the preset IV value algorithm, and whether the node is a blacklist node is used as a label of the preset IV value algorithm. The IV value corresponding to each preset sample feature is calculated by the preset IV value algorithm to evaluate the importance of the preset sample feature. Among them, the specific preset IV value algorithm can refer to the current relevant technology and will not be repeated here.

[0131] S506: Determine a target feature from the preset sample features according to the information value corresponding to the preset sample feature.

[0132] In this step, after obtaining the information value corresponding to each preset sample feature, the preset sample features with high IV values ​​and strong predictive ability can be screened out based on the information value as target features in the decision tree model. For example, for example, preset sample features with IV values ​​greater than 1 are screened out as target features in the decision tree model. This application does not limit the specific screening method.

[0133] S507. Train the initial decision tree model according to the training data set and the target feature to obtain a decision tree model, wherein the decision tree model includes the target feature and the cutoff value corresponding to the target feature.

[0134] Exemplarily, after determining the target features in the decision tree model, an initial decision tree model can be established based on the training data set and the screened target features, and the initial decision tree model can be trained. During the training process, the parameters of the initial decision tree model are adjusted. When it is determined that the blacklist ratio corresponding to at least one leaf node in the leaf nodes contained in the initial decision tree model is greater than or equal to the ratio threshold, the training is stopped to obtain the decision tree model. Among them, the ratio threshold is, for example, 90%; the parameters of the decision tree model include at least one of the segmentation algorithm parameters, the segmentation method (such as a random method or an optimal method), the maximum depth of the decision tree model, and the maximum number of separated features. The decision tree model contains the target features and the cutoff value corresponding to the target features. On this basis, three risk rules can be established according to the size of the blacklist ratio corresponding to the leaf nodes in the decision results output by the decision tree model, namely, a high-risk rule, a medium-risk rule, and a low-risk rule.

[0135] It should be noted that in the traditional machine learning modeling process, after the decision tree model is used for modeling, the resulting decision tree model is directly used to calculate the results, that is, the corresponding results are either yes or no. However, this application does not directly use the calculation results as the prediction results, but uses the decision tree model to calculate the cutoff values ​​corresponding to each target feature. The output decision result includes the blacklist ratio. The reason is: considering the accuracy of the blacklist samples in the training data set, for example, some blacklist samples may be used as positive samples. In the decision tree model training process, since some blacklist samples are regarded as positive samples, the trained decision tree model may have certain errors. The blacklist samples used as positive samples can only express the possibility of risk in the form of probability. Specifically, the blacklist ratio is used to express the probability of risk. Among the leaf nodes of the decision tree model, the nodes with a high blacklist ratio are very likely to be blacklisted.

[0136] The training method of the decision tree model provided in the embodiment of the present application obtains the sample clustering results and preset sample features obtained by clustering the nodes in the sample relationship graph, and obtains the sample features of each node in the sample relationship graph and the feature values ​​corresponding to the sample features; determines the information value corresponding to the preset sample features according to the feature values ​​corresponding to the sample features of each node; determines the target features from the preset sample features according to the information values ​​corresponding to the preset sample features; iteratively trains the initial decision tree model according to the training data set and the target features to obtain a decision tree model. The result output by the decision tree model includes the blacklist ratio corresponding to the node in the sample relationship graph, which is used to determine the risk level of the object in the positive sample corresponding to the node, that is, to express the risk possibility in the form of probability, rather than directly using the result calculated by the decision tree model. Therefore, the obtained decision tree model can be accurately used for the automatic screening of suspicious persons.

[0137] The following are device embodiments of the present application, which can be used to implement the method embodiments of the present application. For details not disclosed in the device embodiments of the present application, please refer to the method embodiments of the present application.

[0138] Figure 6 This is a structural diagram of an information processing device provided in one embodiment of the present application, as shown in FIG. Figure 6 As shown, the information processing device 600 of the embodiment of the present application includes: a first acquisition module 601, a generation module 602, a second acquisition module 603, a processing module 604 and a determination module 605. Among them:

[0139] The first acquisition module 601 is used to acquire a data set to be processed, where the data set includes a plurality of objects to be processed.

[0140] The generating module 602 is used to generate a relationship graph according to the data set, where the relationship graph includes nodes and the association relationships between nodes, wherein the nodes correspond one-to-one to the objects in the data set.

[0141] The second acquisition module 603 is used to obtain the characteristics of each node in the relationship graph and the characteristic value corresponding to the characteristics of each node, wherein the characteristics of each node correspond to the target characteristics in the pre-trained decision tree model, and the decision tree model includes multiple tree nodes, and the tree nodes correspond to the target characteristics one by one.

[0142] The processing module 604 is used to input the feature value corresponding to the feature of each node into a pre-trained decision tree model for decision processing to obtain the blacklist ratio corresponding to each node.

[0143] The determination module 605 is configured to determine the risk level of the object corresponding to each node according to the blacklist ratio corresponding to each node.

[0144] In some embodiments, the processing module 604 can be specifically used to: input the feature value corresponding to the feature of each node into a pre-trained decision tree model, compare the feature value corresponding to the feature of each node with the cutoff value corresponding to the target feature in the pre-trained decision tree model, and determine the decision path corresponding to each node; according to the blacklist object corresponding to the leaf node in the decision path, obtain the blacklist ratio corresponding to each node.

[0145] Optionally, the second acquisition module 603 can be specifically used to: perform clustering processing on each node in the relationship graph to obtain a clustering result corresponding to each node in the relationship graph; based on the clustering result and the target feature, obtain the feature of each node in the relationship graph and the feature value corresponding to the feature of each node.

[0146] Optionally, the determination module 605 can be specifically used to: if the blacklist ratio is greater than or equal to a first threshold, determine the risk level as a high risk level; if the blacklist ratio is less than the first threshold and greater than or equal to a second threshold, determine the risk level as a medium risk level, and the second threshold is less than the first threshold; if the blacklist ratio is less than the second threshold, determine the risk level as a low risk level.

[0147] The device of the embodiment of the present application can be used to execute the information processing method scheme in any of the above method embodiments. Its implementation principle and technical effects are similar and will not be repeated here.

[0148] Figure 7 A structural diagram of a training device for a decision tree model provided in an embodiment of the present application is shown in FIG. Figure 7 As shown, the training device 700 of the decision tree model of the embodiment of the present application includes: a first acquisition module 701, a generation module 702, a second acquisition module 703, a third acquisition module 704, a first determination module 705, a second determination module 706 and a training module 707. Among them:

[0149] The first acquisition module 701 is used to acquire a training data set, where the training data set includes positive samples and negative samples. Positive samples are samples that do not contain blacklisted samples, and negative samples are blacklisted samples.

[0150] The generating module 702 is used to generate a sample relationship graph based on the positive samples and negative samples in the training data set.

[0151] The second acquisition module 703 is used to perform clustering processing on the nodes in the sample relationship graph to obtain sample clustering results corresponding to the nodes in the sample relationship graph.

[0152] The third acquisition module 704 is used to obtain the sample feature of each node in the sample relationship graph and the feature value corresponding to the sample feature according to the sample clustering result and the preset sample feature. The sample feature of each node corresponds to the preset sample feature one by one.

[0153] The first determining module 705 is configured to determine an information value corresponding to a preset sample feature according to a feature value corresponding to the sample feature of each node.

[0154] The second determining module 706 is configured to determine a target feature from the preset sample features according to information values ​​corresponding to the preset sample features.

[0155] The training module 707 is used to iteratively train the initial decision tree model according to the training data set and the target feature to obtain a decision tree model, wherein the decision tree model includes the target feature and the cutoff value corresponding to the target feature.

[0156] Optionally, after acquiring the training data set, the first acquisition module 701 can also be used to: preprocess the positive samples in the training data set to obtain preprocessed positive samples, wherein the preprocessing is to perform credibility filtering on the source channels of the positive samples and to perform data cleaning on the positive samples.

[0157] The device of the embodiment of the present application can be used to execute the solution of the training method of the decision tree model in any of the above method embodiments. Its implementation principle and technical effects are similar and will not be repeated here.

[0158] Figure 8 This is a schematic diagram of the structure of an electronic device provided in one embodiment of the present application. For example, the electronic device can be provided as a server or a computer. Figure 8 The electronic device 800 includes a processing component 801, which further includes one or more processors, and a memory resource represented by a memory 802 for storing instructions executable by the processing component 801, such as an application. The application stored in the memory 802 may include one or more modules, each corresponding to a set of instructions. In addition, the processing component 801 is configured to execute the instructions to perform any of the above-described method embodiments.

[0159] The electronic device 800 may further include a power supply component 803 configured to perform power management of the electronic device 800, a wired or wireless network interface 804 configured to connect the electronic device 800 to a network, and an input / output (I / O) interface 805. The electronic device 800 may operate based on an operating system stored in the memory 802, such as Windows Server™, Mac OS X™, Unix™, Linux™, FreeBSD™, or the like.

[0160] The present application also provides a computer-readable storage medium, which stores computer-executable instructions. When a processor executes the computer-executable instructions, the above-mentioned information processing method and decision tree model training method are implemented.

[0161] The present application also provides a computer program product, including a computer program, which, when executed by a processor, implements the above-mentioned information processing method and decision tree model training method.

[0162] The computer-readable storage medium mentioned above can be implemented by any type of volatile or non-volatile memory device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk, or optical disk. The computer-readable storage medium can be any available medium that can be accessed by a general-purpose or special-purpose computer.

[0163] An exemplary readable storage medium is coupled to a processor so that the processor can read information from the readable storage medium and write information to the readable storage medium. Of course, the readable storage medium can also be an integral part of the processor. The processor and the readable storage medium can be located in an application specific integrated circuit (ASIC). Of course, the processor and the readable storage medium can also exist as discrete components in an information processing device or a training device for a decision tree model.

[0164] Those skilled in the art will appreciate that all or part of the steps in the above-described method embodiments can be implemented using hardware associated with program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0165] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some or all of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present application.

Claims

1. An information processing method, characterized in that: include: Acquire a data set to be processed, wherein the data set includes a plurality of objects to be processed; The subjects to be processed include any of the blacklisted persons, suspicious persons or compliance personnel; generating a relationship graph based on the data set, wherein the relationship graph includes nodes and association relationships between nodes, wherein the nodes correspond one-to-one to the objects in the data set; Obtain the characteristics of each node in the relationship graph and the characteristic values ​​corresponding to the characteristics of each node, wherein the characteristics of each node correspond to the target characteristics in a pre-trained decision tree model, and the decision tree model includes multiple tree nodes, and the tree nodes correspond to the target characteristics one by one; the decision paths composed of different tree nodes correspond to different blacklist ratios; the target characteristics are determined from the preset sample characteristics according to the information value corresponding to the preset sample characteristics during the training stage of the decision tree model, and the information value is determined according to the characteristic value corresponding to the sample characteristics of each node, and the characteristic value corresponding to the sample characteristics of each node is determined based on the sample clustering result and the preset sample characteristics; the target characteristics include in-group characteristics or out-of-group characteristics; the in-group characteristics include at least one of the number of people in the group to which the node belongs, the number of people on the key list of the group to which the node belongs, the proportion of the key list of the group to which the node belongs, the number of people who are first-degree associated with the group to which the node belongs, or the number of people on the key list of the group to which the node belongs; the out-of-group characteristics include at least one of the number of all first-degree associated people of the node, the number of all first-degree associated people on the key list of the node, the webpage ranking value of the node, or the ranking of the webpage ranking values ​​of the nodes in the group to which the node belongs; Inputting the feature value corresponding to the feature of each node into the pre-trained decision tree model for decision processing to obtain the blacklist ratio corresponding to each node; The risk level of the object corresponding to each node is determined according to the blacklist ratio corresponding to each node.

2. The information processing method according to claim 1, wherein: The step of inputting the feature value corresponding to the feature of each node into the pre-trained decision tree model for decision processing to obtain the blacklist ratio corresponding to each node includes: Inputting the feature value corresponding to the feature of each node into the pre-trained decision tree model, comparing the feature value corresponding to the feature of each node with the cutoff value corresponding to the target feature in the pre-trained decision tree model, and determining the decision path corresponding to each node; According to the objects in the blacklist corresponding to the leaf nodes in the decision path, the blacklist ratio corresponding to each node is obtained.

3. The information processing method according to claim 1 or 2, characterized in that: The acquiring the feature of each node in the relationship graph and the feature value corresponding to the feature of each node includes: Performing clustering processing on each node in the relationship graph to obtain a clustering result corresponding to each node in the relationship graph; Based on the clustering result and the target feature, a feature of each node in the relationship graph and a feature value corresponding to the feature of each node are obtained.

4. The information processing method according to claim 1 or 2, characterized in that: The determining, based on the blacklist ratio corresponding to each node, the risk level of the object corresponding to each node includes: If the blacklist ratio is greater than or equal to a first threshold, determining that the risk level is a high risk level; If the blacklist ratio is less than the first threshold and greater than or equal to a second threshold, the risk level is determined to be a medium risk level, and the second threshold is less than the first threshold; If the blacklist ratio is less than the second threshold, the risk level is determined to be a low risk level.

5. A training method for a decision tree model, characterized in that: include: Obtain a training data set, wherein the training data set includes positive samples and negative samples, wherein the positive samples are samples that do not include blacklisted individuals and the positive samples are compliant personnel; The negative sample is a blacklist sample, and the negative sample is any one of a blacklist person and a suspicious person; Generating a sample relationship graph based on the positive samples and negative samples in the training data set; Performing clustering processing on the nodes in the sample relationship graph to obtain sample clustering results corresponding to the nodes in the sample relationship graph; According to the sample clustering result and the preset sample features, a sample feature of each node in the sample relationship graph and a feature value corresponding to the sample feature are obtained, where the sample feature of each node corresponds one-to-one to the preset sample feature; Determining the information value corresponding to the preset sample feature according to the feature value corresponding to the sample feature of each node; Determining a target feature from the preset sample features according to the information value corresponding to the preset sample feature; the target feature includes an in-cluster feature or an out-cluster feature; The intra-group features include at least one of the number of people in the group to which the node belongs, the number of people on the key list of the group to which the node belongs, the proportion of the key list of the group to which the node belongs, the number of people who are first-degree related to the group to which the node belongs, or the number of people on the key list of the group to which the node belongs; the extra-group features include at least one of the number of people who are first-degree related to the node, the number of people on the key list of the group to which the node belongs, the node's webpage ranking value, or the ranking of the node's webpage values ​​within the group to which the node belongs; An initial decision tree model is iteratively trained according to the training data set and the target feature to obtain the decision tree model, wherein the decision tree model includes the target feature and the cutoff value corresponding to the target feature.

6. The training method of the decision tree model according to claim 5, characterized in that: After obtaining the training data set, the method further includes: The positive samples in the training data set are preprocessed to obtain preprocessed positive samples, wherein the preprocessing is to perform credibility filtering on the source channels of the positive samples and to perform data cleaning on the positive samples.

7. An information processing device, characterized in that include: A first acquisition module is used to acquire a data set to be processed, wherein the data set includes a plurality of objects to be processed; The subjects to be processed include any of the blacklisted persons, suspicious persons or compliance personnel; A generating module, configured to generate a relationship graph based on the data set, wherein the relationship graph includes nodes and association relationships between nodes, wherein the nodes correspond one-to-one to the objects in the data set; A second acquisition module is used to obtain the characteristics of each node in the relationship graph and the characteristic values ​​corresponding to the characteristics of each node, wherein the characteristics of each node correspond to the target characteristics in a pre-trained decision tree model, and the decision tree model includes multiple tree nodes, and the tree nodes correspond to the target characteristics one by one; the decision paths composed of different tree nodes correspond to different blacklist ratios; the target characteristics are determined from the preset sample characteristics according to the information value corresponding to the preset sample characteristics during the training stage of the decision tree model, and the information value is determined according to the characteristic value corresponding to the sample characteristics of each node, and the characteristic value corresponding to the sample characteristics of each node is determined based on the sample clustering result and the preset sample characteristics; the target characteristics include in-group characteristics or out-of-group characteristics; the in-group characteristics include at least one of the number of people in the group to which the node belongs, the number of people on the key list of the group to which the node belongs, the proportion of the key list of the group to which the node belongs, the number of people who are first-degree associated with the group to which the node belongs, or the number of people on the key list of the group to which the node belongs; the out-of-group characteristics include at least one of the number of all first-degree associated people of the node, the number of all first-degree associated people on the key list of the node, the webpage ranking value of the node, or the ranking of the webpage ranking values ​​of the nodes in the group to which the node belongs; A processing module, configured to input the feature value corresponding to the feature of each node into the pre-trained decision tree model for decision processing to obtain the blacklist ratio corresponding to each node; The determination module is used to determine the risk level of the object corresponding to each node according to the blacklist ratio corresponding to each node.

8. An electronic device, characterized in that: include: a processor, and a memory communicatively connected to the processor; The memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory to implement the method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer program instructions, and when the computer program instructions are executed by a processor, the method according to any one of claims 1 to 6 is implemented.

10. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images