Method and device for predicting violation personnel

Abstract

The invention relates to a violation person prediction method and device. The method comprises the following steps: in response to a violation prediction request for a target person, obtaining attribute characteristics and behavior characteristics of the target person; processing the attribute features and the behavior features by using a preset violation prediction model to obtain a violation prediction result of the target person; the violation prediction model is obtained by training based on attribute tags and behavior tags of a plurality of sample violation persons and attribute tags and behavior tags of a plurality of sample non-violation persons. Attribute tags and behavior tags of a plurality of sample violation persons and attribute tags and behavior tags of a plurality of sample non-violation persons are adopted in advance to carry out model training to obtain a violation prediction model, so that when violation prediction needs to be carried out on a target person, the violation prediction efficiency is improved. The feature data of the target person can be processed according to the violation prediction model to obtain the violation prediction result of the target person, and violation prediction can be effectively performed on the target person.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to a method and device for predicting violators. Background Art

[0002] Violators are individuals who violate relevant regulations, charters, systems or laws in the process of performing their duties or executing tasks. For example, people in some institutions, enterprises, organizations and other systems who have abnormal behavior, illegal behavior or non-compliant operations are called violators.

[0003] In order to ensure the normal operation of the system, timely prediction of violators is a very necessary means. Through prediction, potential non-compliant behaviors can be identified and prevented in advance, thereby effectively managing and controlling risks.

[0004] Therefore, how to predict violators is a problem that needs to be solved urgently. Summary of the invention

[0005] Based on this, it is necessary to provide a method and device for predicting violators in response to the above-mentioned technical problems, which can effectively predict violators, identify and prevent potential non-compliant behaviors in advance, and thus effectively manage and control risks.

[0006] In a first aspect, the present application provides a method for predicting violators, comprising:

[0007] In response to a violation prediction request for a target person, obtaining attribute characteristics and behavior characteristics of the target person;

[0008] The preset violation prediction model is used to process the attribute characteristics and behavior characteristics to obtain the violation prediction results of the target person; the violation prediction model is trained based on the attribute labels and behavior labels of multiple sample violators and the attribute labels and behavior labels of multiple sample non-violators.

[0009] In the method for predicting illegal personnel provided in the embodiment of the present application, in response to a request for predicting illegal personnel for a target person, the attribute characteristics and behavior characteristics of the target person are obtained, and then the attribute characteristics and behavior characteristics are processed using a preset illegal prediction model to obtain the illegal prediction result of the target person, and the illegal prediction model is trained based on the attribute labels and behavior labels of multiple sample illegal personnel, as well as the attribute labels and behavior labels of multiple sample non-illegal personnel. In this method, the illegal prediction model is obtained by pre-training the model using the attribute labels and behavior labels of multiple sample illegal personnel, as well as the attribute labels and behavior labels of multiple sample non-illegal personnel. In this way, when it is necessary to predict illegal personnel for the target person, the characteristic data of the target person can be processed according to the illegal prediction model to obtain the illegal prediction result of the target person, that is, the illegal prediction of the target person can be effectively performed to identify and prevent potential non-compliant behaviors in advance, thereby effectively managing and controlling risks.

[0010] In one embodiment, a preset violation prediction model is used to process attribute features and behavior features to obtain violation prediction results of the target person, including:

[0011] Determine the target attribute label of the target person according to the attribute characteristics and the preset label rules; and determine the target behavior label of the target person according to the behavior characteristics and the label rules;

[0012] The target attribute label and target behavior label are input into the violation prediction model to obtain the violation prediction result of the target person.

[0013] In the violation prediction method provided in the embodiment of the present application, the target attribute label of the target person is determined according to the attribute characteristics and the preset label rules, and the target behavior label of the target person is determined according to the behavior characteristics and the label rules, and then the target attribute label and the target behavior label are input into the violation prediction model to obtain the violation prediction result of the target person. When the violation prediction model is used to determine the violation prediction result of the target person, the attribute characteristics and the behavior characteristics of the target person are first labeled to obtain the same label data as the input data when training the violation prediction model, and then the attribute label and the behavior label of the target person are input into the violation prediction model, so that the violation prediction result of the target person can be quickly obtained, and the accuracy of the violation prediction result is improved.

[0014] In one embodiment, the training process of the violation prediction model includes:

[0015] According to the attribute labels and behavior labels of each sample violator, and the attribute labels and behavior labels of each sample non-violation person, a training data set and a test data set are obtained;

[0016] According to the training data set, the preset neural network is trained to obtain an initial model;

[0017] The initial model is tested based on the test data set to obtain the violation prediction model.

[0018] In the method for predicting violators provided in the embodiment of the present application, a training data set and a test data set are obtained based on the attribute labels and behavior labels of each sample violator, and the attribute labels and behavior labels of each sample non-violation personnel, and then a preset neural network is trained according to the training data set to obtain an initial model, and the initial model is tested according to the test data set to obtain a violation prediction model. In this method, an optional way to train a violation prediction model is provided by first dividing a training data set and a test data set according to sample data, and then using the training data set for model training to obtain an initial model, and then using the test data set to test the initial model to obtain a violation prediction model, which can help to find problems in a timely manner and make corresponding adjustments, thereby improving the final performance of the model.

[0019] In one embodiment, before testing the initial model according to the test data set to obtain the violation prediction model, the method further includes:

[0020] Obtain a validation data set based on the attribute labels and behavior labels of each sample violator and the attribute labels and behavior labels of each sample non-violation person;

[0021] According to the validation data set, the model parameters of the initial model are adjusted to obtain an adjusted initial model.

[0022] In the violator prediction method provided in the embodiment of the present application, a validation data set is obtained based on the attribute labels and behavior labels of each sample violator and the attribute labels and behavior labels of each sample non-violation person, and then the model parameters of the initial model are adjusted based on the validation data set to obtain an adjusted initial model. In this method, before the initial model is tested, the initial model is verified using a divided validation data set, so that problems with the model or model parameters can be discovered in a timely manner. In this way, the parameters can be readjusted or the model can be adjusted in a timely manner to ensure that the model can perform well on unseen data.

[0023] In one embodiment, a preset neural network is trained according to a training data set to obtain an initial model, including:

[0024] According to the training data set, a positive sample data set and a negative sample data set are obtained; the positive sample data set is determined according to the attribute labels and behavior labels of the non-violated persons in each sample; the negative sample data set is determined according to the attribute labels and behavior labels of the violating persons in each sample;

[0025] The positive sample data set and the negative sample data set are input into the neural network for model training to obtain the initial model.

[0026] In the violator prediction method provided in the embodiment of the present application, a positive sample data set and a negative sample data set are obtained according to the training data set, the positive sample data set is determined according to the attribute labels and behavior labels of the non-violated persons in each sample, and the negative sample data set is determined according to the attribute labels and behavior labels of the violators in each sample, and then the positive sample data set and the negative sample data set are input into the neural network for model training to obtain an initial model. In this method, by obtaining the positive sample data set and the negative sample data set from the training data set, the positive sample data set and the negative sample data set are input into the neural network, and the neural network can process the positive sample data set and the negative sample data set to obtain the output result of the neural network, and the neural network can be iteratively trained according to the output result to obtain the initial model, which provides an optional way to quickly determine the initial model.

[0027] In one embodiment, the positive sample data set and the negative sample data set are input into a neural network for model training to obtain an initial model, including:

[0028] Input the positive sample data set and the negative sample data set into the neural network to obtain the positive sample prediction results and the negative sample prediction results;

[0029] According to the positive sample prediction results, the negative sample prediction results and the preset loss function, the prediction loss value of the neural network is obtained;

[0030] Based on the predicted loss value, the model parameters of the neural network are adjusted until the training is completed and the initial model is obtained.

[0031] In the illegal personnel prediction method provided by the embodiment of the present application, the positive sample data set and the negative sample data set are input into the neural network to obtain the positive sample prediction result and the negative sample prediction result, and then the prediction loss value of the neural network is obtained according to the positive sample prediction result, the negative sample prediction result and the preset loss function, and then the model parameters of the neural network are adjusted based on the predicted loss value until the training is completed to obtain the initial model. In this method, an optional way to quickly determine the initial model is provided; by inputting the positive sample data set and the negative sample data set into the neural network, the prediction result of the neural network for the positive sample data set and the prediction result for the negative sample data set can be obtained, and then the loss value is calculated in combination with the loss function, and the model parameters are continuously adjusted based on the loss value until the training result, the initial model can be obtained, so that by continuously adjusting the model parameters, the prediction ability of the model can be improved and the performance of the initial model can be improved.

[0032] In one embodiment, obtaining a positive sample data set and a negative sample data set according to a training data set includes:

[0033] Perform feature processing on the attribute labels and behavior labels of each sample non-violation personnel to obtain the label values ​​corresponding to the attribute labels and behavior labels of each sample non-violation personnel, and determine the label values ​​corresponding to the attribute labels and behavior labels of each sample non-violation personnel as the positive sample data set;

[0034] The attribute labels and behavior labels of each sample violator are feature processed to obtain label values ​​corresponding to the attribute labels and behavior labels of each sample violator, and the label values ​​corresponding to the attribute labels and behavior labels of each sample violator are determined as negative sample data sets.

[0035] In the method for predicting violators provided in the embodiment of the present application, by performing feature processing on the attribute labels and behavior labels of each sample of non-violation personnel, the label values ​​corresponding to the attribute labels and behavior labels of each sample of non-violation personnel are obtained, and the label values ​​corresponding to the attribute labels and behavior labels of each sample of non-violation personnel are determined as positive sample data sets, by performing feature processing on the attribute labels and behavior labels of each sample of violation personnel, the label values ​​corresponding to the attribute labels and behavior labels of each sample of violation personnel are obtained, and the label values ​​corresponding to the attribute labels and behavior labels of each sample of violation personnel are determined as negative sample data sets. In this method, an optional way to quickly obtain positive sample data sets and negative sample data sets is provided; by performing feature processing on the attribute labels and behavior labels of each sample of non-violation personnel, a positive sample data set is obtained, and by performing feature processing on the attribute labels and behavior labels of each sample of violation personnel, a negative sample data set is obtained, which provides data support for the subsequent training of violation prediction models.

[0036] In one embodiment, the method further comprises:

[0037] Obtain the identity information of non-violation personnel in each sample;

[0038] Using each identity information as the merge primary key, the attribute labels and behavior labels of each sample non-violation personnel are associated with each identity information to obtain multiple mapping relationships;

[0039] According to multiple mapping relationships, a personnel label database is generated.

[0040] In the method for predicting illegal personnel provided in the embodiment of the present application, by obtaining the identity information of the non-illegal personnel in each sample, and then using each identity information as the merge primary key, the attribute label and behavior label of each non-illegal personnel in each sample are associated with each identity information to obtain multiple mapping relationships, and then a personnel label database is generated based on the multiple mapping relationships. In this method, by associating the attribute label and behavior label of each non-illegal personnel in each sample with the identity information of each non-illegal personnel in each sample, multiple mapping relationships can be obtained, and then the multiple mapping relationships are stored in the database, and a personnel label database can be obtained, which facilitates the label query of the subsequent non-illegal personnel in the sample.

[0041] In one embodiment, the method further comprises:

[0042] In response to a tag query request of a person to be queried, obtaining identity information of the person to be queried;

[0043] According to the identity information of the person to be queried, the attribute labels and behavior labels corresponding to the identity information are filtered from the person label database;

[0044] The filtered attribute labels and behavior labels corresponding to the identity information are determined as labels of the person to be queried.

[0045] In the method for predicting illegal persons provided in the embodiment of the present application, the identity information of the person to be queried is obtained by responding to the label query request of the person to be queried, and then the attribute label and behavior label corresponding to the identity information are screened from the personnel label database according to the identity information of the person to be queried, and then the screened attribute label and behavior label corresponding to the identity information are determined as the label of the person to be queried. In this method, when it is necessary to query the label of the person to be queried, the attribute label and behavior label corresponding to the identity information can be quickly matched from the personnel label database based on the identity information of the person to be queried. The matched attribute label and behavior label are the label of the person to be queried, and the efficiency of label query is improved by quickly matching the label with the identity information.

[0046] In a second aspect, the present application also provides a device for predicting illegal personnel, including:

[0047] A feature acquisition module, used to obtain the attribute characteristics and behavior characteristics of the target person in response to a violation prediction request for the target person;

[0048] The result determination module is used to process the attribute characteristics and behavior characteristics using a preset violation prediction model to obtain the violation prediction results of the target person; the violation prediction model is trained based on the attribute labels and behavior labels of multiple sample violators and the attribute labels and behavior labels of multiple sample non-violators.

[0049] In a third aspect, the present application also provides a computer device, which includes a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps in any one of the embodiments of the first aspect when executing the computer program.

[0050] In a fourth aspect, the embodiments of the present application further provide a computer-readable storage medium, wherein a computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, the steps in any one of the embodiments of the first aspect are implemented.

[0051] In a fifth aspect, the embodiments of the present application further provide a computer program product, which includes a computer program, and when the computer program is executed by a processor, the steps in any one of the embodiments in the first aspect are implemented.

[0052] The above-mentioned method and device for predicting illegal personnel, by responding to the illegal prediction request for the target personnel, obtains the attribute characteristics and behavior characteristics of the target personnel, and then uses the preset illegal prediction model to process the attribute characteristics and behavior characteristics to obtain the illegal prediction result of the target personnel, and the illegal prediction model is trained based on the attribute labels and behavior labels of multiple sample illegal personnel, as well as the attribute labels and behavior labels of multiple sample non-illegal personnel. In this method, the illegal prediction model is obtained by pre-training the model with the attribute labels and behavior labels of multiple sample illegal personnel, as well as the attribute labels and behavior labels of multiple sample non-illegal personnel. In this way, when it is necessary to predict illegal activities of the target personnel, the characteristic data of the target personnel can be processed according to the illegal prediction model to obtain the illegal prediction result of the target personnel, that is, it is possible to effectively predict illegal activities of the target personnel, so as to identify and prevent potential non-compliant behaviors in advance, thereby effectively managing and controlling risks. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related technologies, the drawings required for use in the embodiments or the related technical descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0054] Figure 1 is an internal structure diagram of a computer device in one embodiment;

[0055] Figure 2 A schematic diagram of a flow chart of a method for predicting violators in one embodiment;

[0056] Figure 3 A schematic diagram of a process for determining a violation prediction result in one embodiment;

[0057] Figure 4 A schematic diagram of a process for training a violation prediction model in one embodiment;

[0058] Figure 5 A schematic diagram of a flow chart of a verification model in an embodiment;

[0059] Figure 6 A schematic diagram of a process for determining an initial model in one embodiment;

[0060] Figure 7 A schematic diagram of a process for determining an initial model in another embodiment;

[0061] Figure 8 A schematic diagram of a process for obtaining a positive sample data set and a negative sample data set in one embodiment;

[0062] Fig. 9 A schematic diagram of a process for generating a personnel tag database in one embodiment;

[0063] Fig.10 A schematic diagram of a process of querying a tag in an embodiment;

[0064] Fig.11 A schematic diagram of a flow chart of a method for predicting violators in another embodiment;

[0065] Fig.12 Schematic diagram of the structure of a device for predicting illegal personnel in one embodiment. DETAILED DESCRIPTION

[0066] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0067] The illegal personnel prediction method provided in the embodiment of the present application can be applied to a computer device. The computer device can be a server, and its internal structure diagram can be as follows: Figure 1As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, memory and input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The database of the computer device is used to store data. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a method for predicting violators is implemented. Those skilled in the art can understand that Figure 1 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0068] Violators are individuals who violate relevant regulations, charters, systems or laws in the process of performing their duties or executing tasks. For example, people in some institutions, enterprises, organizations and other systems who have abnormal behavior, illegal behavior or non-compliant operations are called violators.

[0069] In order to ensure the normal operation of the system, timely prediction of violators is a very necessary means. Through prediction, potential non-compliant behaviors can be identified and prevented in advance, thereby effectively managing and controlling risks.

[0070] Therefore, how to predict violators is a problem that needs to be solved urgently.

[0071] Based on this, the present application provides a method for predicting violators, by pre-training the model using the attribute labels and behavior labels of multiple sample violators and the attribute labels and behavior labels of multiple sample non-violators. In this way, when it is necessary to predict violations for the target person, the characteristic data of the target person can be processed according to the violation prediction model to obtain the violation prediction result of the target person, that is, the violation prediction of the target person can be effectively performed, so as to identify and prevent potential non-compliant behaviors in advance, thereby effectively managing and controlling risks.

[0072] It should be noted that the beneficial effects brought about by the embodiments of the present application or the technical problems solved are not limited to this one, but may also include other implicit or related problems. For details, please refer to the description of the following embodiments.

[0073] The technical solution of the present application and how the technical solution of the present application solves the above-mentioned technical problems are described in detail below with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present application will be described below in conjunction with the accompanying drawings.

[0074] In an exemplary embodiment, Figure 2 As shown, a method for predicting illegal personnel is provided, and the method is applied to Figure 1 The computer device in the example is used to illustrate, including the following steps 201 to 202. Among them:

[0075] S201, in response to a violation prediction request for a target person, obtaining attribute characteristics and behavior characteristics of the target person.

[0076] Among them, attribute characteristics include gender, age, education level, income and debt situation, work situation, personal violations, violations of immediate family members, personal personality type, etc.; behavioral characteristics include communication characteristics, accommodation behavior characteristics, online behavior characteristics, personal account characteristics, personal consumption characteristics, entry and exit characteristics, etc.

[0077] The violation prediction request for the target person refers to a request that is triggered when a violation prediction is performed on the target person in the violation prediction application of the computer device. Naturally, after receiving the violation prediction request, the computer device needs the target person to perform a violation prediction to determine whether the target person is a violator. Taking an enterprise as an example, violators may include those who do not follow enterprise regulations, take enterprise items privately, damage equipment and tools, forge documents, or are involved in economic crimes.

[0078] In one embodiment, the request is triggered by a user in a violation prediction application through a computer device interface; wherein the triggering method includes but is not limited to external input device click, voice control, touch screen trigger, etc., and the present application embodiment does not limit the triggering method. The user here may refer to relevant personnel of the enterprise management department, etc.

[0079] In another embodiment, the request is triggered according to a preset automatic program; for example, if the computer device receives the personnel information of the target person (such as name, work number, etc.), it automatically triggers the violation prediction request for the target person according to the preset program; or, after the computer device receives the personnel information of the target person, when the application is in an idle state, it automatically triggers the violation prediction request for the target person according to the preset program. The embodiment of the present application does not limit the method of triggering the violation prediction request for the target person.

[0080] Exemplarily, when a computer device receives a violation prediction request for a target person, it can send a data acquisition request to multiple third-party systems, wherein the data acquisition request includes identification information of the target person, such as name, age, etc. After receiving the data acquisition request sent by the computer device, each third-party system filters the data corresponding to the identification information from the database according to the identification information carried in the data acquisition request, and sends the filtered data to the computer device, so that the computer device can receive the attribute characteristics and behavioral characteristics of the target person. It should be noted that the personnel information, attribute characteristics, behavioral characteristics and other data of the target person in the embodiments of the present application are all authorized or fully authorized by all parties, and the collection, use and processing of relevant data need to comply with relevant regulations.

[0081] S202, using a preset violation prediction model to process the attribute characteristics and behavior characteristics to obtain a violation prediction result of the target person; the violation prediction model is trained based on the attribute labels and behavior labels of multiple sample violators and the attribute labels and behavior labels of multiple sample non-violators.

[0082] After acquiring the attribute characteristics and behavioral characteristics of the target person, the computer device can use a preset violation prediction model to process the attribute characteristics and behavioral characteristics of the target person to obtain the output result of the violation prediction model, that is, obtain the violation prediction result of the target person.

[0083] Among them, the violation prediction model is a model that is pre-trained using the attribute labels and behavior labels of multiple sample violators and the attribute labels and behavior labels of multiple sample non-violation persons.

[0084] The attribute labels and behavior labels of multiple sample violators can be obtained by labeling the attribute characteristics and behavior characteristics of multiple sample violators; similarly, the attribute labels and behavior labels of multiple sample non-violators can also be obtained by labeling the attribute characteristics and behavior characteristics of multiple sample non-violators.

[0085] Among them, the attribute characteristics and behavioral characteristics of multiple sample violators, as well as the attribute characteristics and behavioral characteristics of multiple sample non-violators, can also be obtained from multiple third-party systems, and the acquisition method is the same as the method of obtaining the attribute characteristics and behavioral characteristics of the target person, which will not be repeated here. It should be noted that in the embodiments of the present application, the personnel information and attribute characteristics, behavioral characteristics and other data of multiple sample violators, as well as the personnel information and attribute characteristics, behavioral characteristics and other data of multiple sample non-violators are all authorized or fully authorized by all parties. The collection, use and processing of relevant data need to comply with relevant regulations.

[0086] When labeling the attribute characteristics and behavioral characteristics of multiple sample violators, or the attribute characteristics and behavioral characteristics of multiple sample non-violation personnel, the preset labeling rules can be used for processing, and different labeling rules are used for different characteristics. The labeling rules for each of the attribute characteristics and behavioral characteristics mentioned above can be shown as the following rules (the following is an example of multiple sample violators).

[0087] For gender, the gender of each sample violator, male or female, is the gender label of each sample violator.

[0088] For age, the sample age distribution is extracted and the Gaussian distribution equation is established to facilitate the formulation of age labels, such as "minor" [12-18 years old], "youth stage one" [19-23 years old], "youth stage two" [24-28 years old], "youth stage three" [29-33 years old], "youth stage four" [34-38 years old], "youth stage five" [39-45 years old], "middle-aged stage one" [45-50 years old], "middle-aged and elderly stage" [over 50 years old], etc.

[0089] For academic qualifications, the sample student information is extracted and labels are divided according to the highest academic qualification level. For example, the label for those who drop out or fail to graduate from ordinary primary school or adult primary school is "primary education stage not completed"; the label for those who graduate from ordinary primary school or adult primary school is "primary education stage completed"; the label for those who drop out or fail to graduate from junior high school or high school is "secondary education stage not completed"; the label for the graduation mark of junior high school or high school is "secondary education stage completed"; the label for those who drop out or fail to graduate from ordinary or adult undergraduate or junior college is "higher education stage one not completed"; the label for those who graduate from ordinary or adult undergraduate or junior college is "ordinary or adult undergraduate or junior college graduated"; the label for those who fail to graduate or drop out of postgraduate school is "higher education stage two not completed", and the label for postgraduate graduation is "higher education stage two completed", etc.

[0090] For income and debt, analyze the family income and debt to assess the comprehensive debt situation. For example, calculate the ratio of the monthly loan repayment to the monthly income of the individual and spouse, denoted as D, and the labels are shown in Table 1 below.

[0091] Table 1

[0092]

[0093] Regarding employment status, based on social security data, those who have paid social security for six consecutive months are marked as “stable employment”, and those on the unemployment registration list are marked as “unemployed”.

[0094] For personal violations, if there is a history of violations, the label will be "person with historical violations"; if there is no history of violations, the label will be "person with no historical violations".

[0095] For violations by direct relatives, the actual situation of the violator and the direct relative is taken into consideration. If there is a violation, the label is "Father-Violation"; if there is no violation, the label is "Father-No Violation".

[0096] For individual personality types, taking the MBTI (Myers-Briggs Type Indicator) type as an example, labels can be added, including personality types such as "analyst", "architect", "commander", and "debater". Furthermore, the MBTI types of people in the sample are extracted to construct a Markov prediction model to predict the approximate distribution of personality types of individuals who will violate regulations in the future.

[0097] For example, the personality of "Analyst" is marked as node ①, the personality of "Architect" is marked as node ②, and the personality of "Commander" is marked as node ③. The ratio of node ① to the total sample is 11 , the ratio of the number of people in node ② to the total sample is P 12 , the ratio of the number of people in node ③ to the total samples P 13 , then: P 11 + P 12 + P 13 =1. Get the samples of a certain stage, and the corresponding ratios of nodes ①, ②, and ③ are P 11 , P 12 , P 13 ; The corresponding ratios of nodes ①, ②, and ③ in the next year are P 21 , P 22 , P 23 ; The initial transition probability matrix of the Markov model is , where the first row is the node value of the current stage, and the second row is the node value of the next stage. Each time the Markov model iterates, the iteration result is retained as the initial value of the next iteration. Assuming that the predicted sample stage has undergone n iterations compared to the current stage, the n-step transfer matrix corresponding to the final transfer state probability is , P n1 Sample assigned to the "Analyst" personality type, P n2 Sample assigned the personality type of "Architect", P n3 Sample assignment of the "Commander" personality type.

[0098] For the communication features, extract the call records of each violator during the violation period, set the main label and sub-label according to the call distance, the communication object feature, the communication time feature, the communication duration feature, and the communication object are the main labels, and the other features are sub-labels. For example, the communication time between 22:00 and 6:00 is marked as "night call", and the other time periods are marked as "normal time call"; the communication objects can be classified into "violators", "dishonest natural persons", "large debtors", "unemployed persons" and so on according to specific labels. The call duration is greater than 1 hour and marked as "too long call", and less than 15 seconds is marked as "too short call"; for example, XX is a violator, who called XX at 12 o'clock in the middle of the night for three consecutive days in a week. The first call time is two hours, and the remaining calls are no more than 10 seconds. Then the violator can be marked with the main label "Communication with violator XX", and the sub-labels are "night communication", "abnormal communication duration", "continuous communication date", etc.

[0099] For accommodation behavior characteristics, analyze the hotel accommodation patterns of the offenders during the period of violation, including abnormal accommodation time periods and abnormal accommodation frequencies. Take check-in and check-out as the standard, and establish a gradient sequence by time period particles (-1 night, 0 morning, 1 afternoon, 2 nights). Set the accommodation frequency coefficient S1, which is positively correlated with the accommodation frequency.

[0100] For online behavior characteristics, the online behavior characteristics of violators during the violation period are extracted, mainly mobile phone number registration and VPN registration behaviors, and the frequency coefficient S2 is set, and S2 is positively correlated with the registration frequency.

[0101] Personal account characteristics include account opening and account transaction characteristics. Set the frequency coefficient S3, which is positively correlated with the frequency of bank account opening. Associate the account IDs of the violators with the combined account amounts, analyze the transaction patterns of UnionPay accounts, and label them as "large amount outgoing transfers", "large amount incoming transfers", "frequent outgoing transfers", "frequent incoming transfers", etc.

[0102] For personal consumption characteristics, we extract consumption records of violators and their relatives, analyze consumption patterns, and determine the nature of consumption, and label them as "large-scale consumption", "luxury consumption", "frequent large-scale consumption", etc.

[0103] For entry-exit characteristics, we extract entry-exit records and dates to identify individual entry-exit behaviors. Furthermore, we analyze the temporal and spatial laws of multiple atomic behavioral characteristics and diverge to form derived behavioral characteristics, such as "leaving the country one week after a large transfer in" and "large transfers in through UnionPay after frequent registration of mobile phone numbers."

[0104] Exemplarily, after obtaining the attribute characteristics and behavioral characteristics of multiple sample violators and the attribute characteristics and behavioral characteristics of multiple non-violators, the computer device labels the multiple sample violators according to the above-mentioned labeling rules to obtain the attribute labels and behavioral labels of multiple sample violators, and labels the multiple sample non-violators to obtain the attribute labels and behavioral labels of multiple sample non-violators; then the computer device inputs the attribute labels and behavioral labels of the multiple sample violators and the attribute labels and behavioral labels of the multiple sample non-violators into a preset neural network for model training to obtain a violation prediction model.

[0105] After the violation prediction model is trained, the violation prediction model can be used to predict violations of the target person. The computer device can input the attribute characteristics and behavior characteristics of the target person into the violation prediction model, and the violation prediction model analyzes and processes the attribute characteristics and behavior characteristics of the target person to obtain the output result of the violation prediction model, that is, the violation prediction result of the target person.

[0106] In the method for predicting illegal personnel provided in the embodiment of the present application, in response to a request for predicting illegal personnel for a target person, the attribute characteristics and behavior characteristics of the target person are obtained, and then the attribute characteristics and behavior characteristics are processed using a preset illegal prediction model to obtain the illegal prediction result of the target person, and the illegal prediction model is trained based on the attribute labels and behavior labels of multiple sample illegal personnel, as well as the attribute labels and behavior labels of multiple sample non-illegal personnel. In this method, the illegal prediction model is obtained by pre-training the model using the attribute labels and behavior labels of multiple sample illegal personnel, as well as the attribute labels and behavior labels of multiple sample non-illegal personnel. In this way, when it is necessary to predict illegal personnel for the target person, the characteristic data of the target person can be processed according to the illegal prediction model to obtain the illegal prediction result of the target person, that is, the illegal prediction of the target person can be effectively performed to identify and prevent potential non-compliant behaviors in advance, thereby effectively managing and controlling risks.

[0107] Based on the above embodiment, an embodiment is provided to illustrate the process of using the preset violation prediction model to process the attribute characteristics and behavior characteristics to obtain the violation prediction results of the target person.

[0108] In an exemplary embodiment, Figure 3As shown, the preset violation prediction model is used to process the attribute characteristics and behavior characteristics to obtain the violation prediction results of the target person, including:

[0109] S301, determining a target attribute label of a target person according to attribute characteristics and preset label rules; and determining a target behavior label of the target person according to behavior characteristics and label rules.

[0110] After acquiring the attribute characteristics and behavioral characteristics of the target person, the computer device can label the attribute characteristics of the target person according to the preset label rules to obtain the target attribute label of the target person, and label the behavioral characteristics of the target person according to the preset label rules to obtain the target behavior label of the target person.

[0111] Among them, the preset labeling rules are consistent with the labeling rules used for labeling multiple sample violators and multiple sample non-violators mentioned above, and will not be repeated here.

[0112] S302, input the target attribute label and the target behavior label into the violation prediction model to obtain the violation prediction result of the target person.

[0113] After obtaining the target attribute label and target behavior label of the target person, the computer device can input the target attribute label and target behavior label of the target person into the trained violation prediction model to obtain the result output by the violation prediction model, which is the violation prediction result of the target person.

[0114] In the violation prediction method provided in the embodiment of the present application, the target attribute label of the target person is determined according to the attribute characteristics and the preset label rules, and the target behavior label of the target person is determined according to the behavior characteristics and the label rules, and then the target attribute label and the target behavior label are input into the violation prediction model to obtain the violation prediction result of the target person. When the violation prediction model is used to determine the violation prediction result of the target person, the attribute characteristics and the behavior characteristics of the target person are first labeled to obtain the same label data as the input data when training the violation prediction model, and then the attribute label and the behavior label of the target person are input into the violation prediction model, so that the violation prediction result of the target person can be quickly obtained, and the accuracy of the violation prediction result is improved.

[0115] Based on any of the above embodiments, an embodiment is provided to illustrate the training process of the above violation prediction model.

[0116] In an exemplary embodiment, Figure 4 As shown in Figure 1, the training process of the violation prediction model includes:

[0117] S401, obtaining a training data set and a test data set according to the attribute labels and behavior labels of each sample violator and the attribute labels and behavior labels of each sample non-violator.

[0118] Before training the violation prediction model, the attribute labels and behavior labels of each sample violator and the attribute labels and behavior labels of each sample non-violation person are first divided into a training data set and a test data set.

[0119] The computer device can use the attribute labels and behavior labels of the non-violated persons in each sample as positive samples, and the attribute labels and behavior labels of the sample violating persons as negative samples, and then divide the attribute labels and behavior labels of the non-violated persons in each sample and the attribute labels and behavior labels of the sample violating persons into 70% training data sets and 30% test data sets. The training data sets and the test data sets both include positive samples and negative samples.

[0120] S402, performing model training on a preset neural network according to a training data set to obtain an initial model.

[0121] After the training data set and the test data set are divided, the computer device uses the training data set to train the model.

[0122] Exemplarily, the computer device can input the training data set into a preset neural network, and then obtain the positive sample prediction results and negative sample prediction results output by the neural network. Thereafter, the model parameters of the neural network are adjusted according to the positive sample prediction results and the negative sample prediction results until the number of iterations reaches a threshold, thereby obtaining an initial model.

[0123] S403, testing the initial model according to the test data set to obtain a violation prediction model.

[0124] After the neural network is trained using the training data set to obtain an initial model, the computer device may test the initial model using the test data set to obtain a violation prediction model.

[0125] For example, the computer device can select appropriate evaluation indicators according to the specific task in advance, input the test data set into the initial model, obtain the prediction results output by the initial model, and then calculate the previously selected evaluation indicators according to the prediction results and the target results to evaluate the performance of the model. If the initial model performs poorly in some predictions, the model needs to be further adjusted until the initial model passes the test and obtains the violation prediction model; if the initial model performs well, the current initial model is determined as the violation prediction model.

[0126] In the method for predicting violators provided in the embodiment of the present application, a training data set and a test data set are obtained based on the attribute labels and behavior labels of each sample violator, and the attribute labels and behavior labels of each sample non-violation personnel, and then a preset neural network is trained according to the training data set to obtain an initial model, and the initial model is tested according to the test data set to obtain a violation prediction model. In this method, an optional way to train a violation prediction model is provided by first dividing a training data set and a test data set according to sample data, and then using the training data set for model training to obtain an initial model, and then using the test data set to test the initial model to obtain a violation prediction model, which can help to find problems in a timely manner and make corresponding adjustments, thereby improving the final performance of the model.

[0127] Based on the above embodiment, an embodiment is provided to illustrate the process before the initial model is tested according to the test data set to obtain the violation prediction model.

[0128] In an exemplary embodiment, Figure 5 As shown, before testing the initial model according to the test data set to obtain the violation prediction model, the method also includes:

[0129] S501, obtaining a verification data set based on the attribute labels and behavior labels of each sample violator and the attribute labels and behavior labels of each sample non-violation person.

[0130] When dividing the data set according to the attribute labels and behavior labels of each sample violator and the attribute labels and behavior labels of each sample non-violation person, in addition to dividing the training data set and the test data set, a part of the verification data set can also be divided to verify the initial model.

[0131] The computer device can divide the attribute labels and behavior labels of the non-violated persons in each sample, and the attribute labels and behavior labels of the violating persons in each sample into 60% training data set, 30% test data set and 10% verification set. Among them, the verification data set also includes positive samples and negative samples. The positive samples are the attribute labels and behavior labels of the non-violated persons in each sample, and the negative samples are the attribute labels and behavior labels of the violating persons in each sample.

[0132] S502: Adjust the model parameters of the initial model according to the verification data set to obtain an adjusted initial model.

[0133] After the neural network is trained using the training data set to obtain an initial model, the computer device may first use the verification data set to adjust the model parameters of the initial model, and then test the initial model.

[0134] Exemplarily, the computer device can input the verification data set into the initial model to obtain the prediction results output by the initial model, and then calculate the loss value based on the prediction results, and then adjust the model parameters of the initial model according to the loss value until the loss value is less than a preset loss threshold, thereby obtaining the adjusted initial model.

[0135] After obtaining the adjusted initial model, the computer device may input a test data set into the adjusted initial model and test the adjusted initial model to obtain a violation prediction model.

[0136] In the violator prediction method provided in the embodiment of the present application, a validation data set is obtained based on the attribute labels and behavior labels of each sample violator and the attribute labels and behavior labels of each sample non-violation person, and then the model parameters of the initial model are adjusted based on the validation data set to obtain an adjusted initial model. In this method, before the initial model is tested, the initial model is verified using a divided validation data set, so that problems with the model or model parameters can be discovered in a timely manner. In this way, the parameters can be readjusted or the model can be adjusted in a timely manner to ensure that the model can perform well on unseen data.

[0137] Based on the above embodiment, an embodiment is provided to illustrate the process of performing model training on a preset neural network according to the training data set to obtain an initial model.

[0138] In an exemplary embodiment, Figure 6 As shown, according to the training data set, the preset neural network is trained to obtain an initial model, including:

[0139] S601, according to the training data set, obtain a positive sample data set and a negative sample data set; the positive sample data set is determined according to the attribute labels and behavior labels of the non-violated persons in each sample; the negative sample data set is determined according to the attribute labels and behavior labels of the violating persons in each sample.

[0140] As mentioned above, the training data set includes positive samples and negative samples. When dividing the training data set, the training data set already includes positive samples and negative samples. The computer device can separate the positive samples and negative samples from the divided training data set to obtain a positive sample data set and a negative sample data set.

[0141] In one embodiment, the computer device can obtain the positive sample data set and the negative sample data set from the training data set according to the identification information of the positive sample and the negative sample. For example, the positive sample is composed of the labels of the non-violated persons in each sample, and its identification information can be 1, and the negative sample is composed of the labels of the violating persons in each sample, and its identification information can be 0. In this way, the computer device can filter out the positive sample and the negative sample from the training data set according to the identification information of 1 and 0, and then determine the data corresponding to the positive sample as the positive sample data set, and determine the data corresponding to the negative sample as the negative sample data set.

[0142] In another embodiment, the computer device can filter out positive samples and negative samples from the training data set according to the identification information of the positive samples and negative samples, and then perform data processing on the data in the positive samples and negative samples to obtain the positive sample data set and the negative sample data set. The data processing method may include data normalization, data dimension reduction, one-hot encoding, etc.

[0143] S602: Input the positive sample data set and the negative sample data set into the neural network for model training to obtain an initial model.

[0144] After obtaining the positive sample data set and the negative sample data set, the positive sample data set and the negative sample data set can be input into a neural network for model training to obtain an initial model.

[0145] For example, a computer device can input a positive sample data set and a negative sample data set into a neural network to obtain positive sample prediction results and negative sample prediction results output by the neural network, and then calculate a loss value based on the positive sample prediction results and the negative sample prediction results, and then continuously adjust the model parameters of the neural network based on the loss value until the training result is obtained to obtain an initial model.

[0146] In the violator prediction method provided in the embodiment of the present application, a positive sample data set and a negative sample data set are obtained according to the training data set, the positive sample data set is determined according to the attribute labels and behavior labels of the non-violated persons in each sample, and the negative sample data set is determined according to the attribute labels and behavior labels of the violators in each sample, and then the positive sample data set and the negative sample data set are input into the neural network for model training to obtain an initial model. In this method, by obtaining the positive sample data set and the negative sample data set from the training data set, the positive sample data set and the negative sample data set are input into the neural network, and the neural network can process the positive sample data set and the negative sample data set to obtain the output result of the neural network, and the neural network can be iteratively trained according to the output result to obtain the initial model, which provides an optional way to quickly determine the initial model.

[0147] Based on the above embodiment, an embodiment is provided to illustrate the process of inputting the positive sample data set and the negative sample data set into the neural network for model training to obtain the initial model.

[0148] In an exemplary embodiment, Figure 7 As shown, the positive sample data set and the negative sample data set are input into the neural network for model training to obtain the initial model, including:

[0149] S701, inputting the positive sample data set and the negative sample data set into the neural network to obtain the positive sample prediction results and the negative sample prediction results.

[0150] The computer device can input the positive sample data set and the negative sample data set into the neural network, and then the neural network processes the positive sample data set to obtain the prediction result of the neural network for the positive sample data set, and outputs the positive sample prediction result, and processes the negative sample data set to obtain the prediction result of the neural network for the negative sample data set, and outputs the negative sample prediction result.

[0151] S702, obtaining a prediction loss value of a neural network according to the positive sample prediction result, the negative sample prediction result and a preset loss function.

[0152] After obtaining the positive sample prediction results and negative sample prediction results output by the neural network, the computer device can combine the preset loss function, input the positive sample prediction results and negative sample prediction results into the loss function, and obtain the prediction loss value of the neural network.

[0153] S703, based on the predicted loss value, adjusting the model parameters of the neural network until the training is completed to obtain an initial model.

[0154] After obtaining the predicted loss value, the computer device can adjust the model parameters of the neural network according to the predicted loss value until the predicted loss value is less than a preset loss threshold, stop adjusting the model parameters, and determine the model with current parameters as the initial model.

[0155] In the illegal personnel prediction method provided by the embodiment of the present application, the positive sample data set and the negative sample data set are input into the neural network to obtain the positive sample prediction result and the negative sample prediction result, and then the prediction loss value of the neural network is obtained according to the positive sample prediction result, the negative sample prediction result and the preset loss function, and then the model parameters of the neural network are adjusted based on the predicted loss value until the training is completed to obtain the initial model. In this method, an optional way to quickly determine the initial model is provided; by inputting the positive sample data set and the negative sample data set into the neural network, the prediction result of the neural network for the positive sample data set and the prediction result for the negative sample data set can be obtained, and then the loss value is calculated in combination with the loss function, and the model parameters are continuously adjusted based on the loss value until the training result, the initial model can be obtained, so that by continuously adjusting the model parameters, the prediction ability of the model can be improved and the performance of the initial model can be improved.

[0156] Based on the above embodiment, an embodiment is provided to illustrate the process of obtaining a positive sample data set and a negative sample data set based on a training data set.

[0157] In an exemplary embodiment, Figure 8 As shown, according to the training data set, the positive sample data set and the negative sample data set are obtained, including:

[0158] S801, feature processing is performed on the attribute labels and behavior labels of each sample non-violation personnel to obtain label values ​​corresponding to the attribute labels and behavior labels of each sample non-violation personnel, and the label values ​​corresponding to the attribute labels and behavior labels of each sample non-violation personnel are determined as the positive sample data set.

[0159] In the embodiment of the present application, the positive samples are composed of the attribute labels and behavior labels of the non-violated persons in each sample, and the positive sample data set is obtained by performing feature processing on the attribute labels and behavior labels of the non-violated persons in each sample.

[0160] The computer device can perform feature processing on the attribute labels and behavior labels of each sample non-violation person respectively, obtain the label values ​​corresponding to the attribute labels of each sample non-violation person and the label values ​​corresponding to the behavior labels, and then determine the label values ​​corresponding to the attribute labels of each sample non-violation person and the label values ​​corresponding to the behavior labels as the positive sample data set.

[0161] Among them, feature processing methods can include unique hot encoding, data normalization, categorical-quantitative conversion, data dimensionality reduction and other feature processing.

[0162] For example, labels such as personality, personality type, income and debt status can be one-hot encoded. Taking gender as an example, the gender label is converted into a binary code. If the data contains n items, the converted binary code has n bits, such as male gender is converted to '01' and female gender is converted to '10'.

[0163] Data normalization can also be performed on labels such as age and education level. Taking age as an example, the age label is scaled to a reasonable data domain range, such as scaling the age label to the range of (0, 1). The transcoding is shown in Table 2 below.

[0164] Table 2

[0165]

[0166] It is also possible to perform categorical-quantitative conversion on entry-exit characteristics, personal consumption characteristics, personal account characteristics, and communication characteristics, for example, by establishing feature transcoding for such discrete characteristics.

[0167] S802, feature processing is performed on the attribute labels and behavior labels of each sample violator to obtain label values ​​corresponding to the attribute labels and behavior labels of each sample violator, and the label values ​​corresponding to the attribute labels and behavior labels of each sample violator are determined as negative sample data sets.

[0168] Similarly, the computer device can perform feature processing on the attribute labels and behavior labels of each sample violator, obtain the label value corresponding to the attribute label of each sample violator and the label value corresponding to the behavior label, and then determine the label value corresponding to the attribute label of each sample violator and the label value corresponding to the behavior label as a negative sample data set. Among them, the feature processing method is consistent with the feature processing method corresponding to each sample non-violation person, and will not be repeated here.

[0169] In the method for predicting violators provided in the embodiment of the present application, by performing feature processing on the attribute labels and behavior labels of each sample of non-violation personnel, the label values ​​corresponding to the attribute labels and behavior labels of each sample of non-violation personnel are obtained, and the label values ​​corresponding to the attribute labels and behavior labels of each sample of non-violation personnel are determined as positive sample data sets, by performing feature processing on the attribute labels and behavior labels of each sample of violation personnel, the label values ​​corresponding to the attribute labels and behavior labels of each sample of violation personnel are obtained, and the label values ​​corresponding to the attribute labels and behavior labels of each sample of violation personnel are determined as negative sample data sets. In this method, an optional way to quickly obtain positive sample data sets and negative sample data sets is provided; by performing feature processing on the attribute labels and behavior labels of each sample of non-violation personnel, a positive sample data set is obtained, and by performing feature processing on the attribute labels and behavior labels of each sample of violation personnel, a negative sample data set is obtained, which provides data support for the subsequent training of violation prediction models.

[0170] After obtaining the attribute labels and behavior labels of each sample non-violation personnel, the attribute labels and behavior labels of each sample non-violation personnel can also be stored in a database to provide a label query function. Based on this, the following embodiment describes the method of generating a label database.

[0171] In an exemplary embodiment, Fig. 9 As shown, the method also includes:

[0172] S901, obtaining the identity information of the non-violation personnel in each sample.

[0173] In the embodiment of the present application, the identity information of the non-violation persons in each sample may include the name, ID number, home address, contact information, etc. of the non-violation persons in each sample.

[0174] S902, using each identity information as the merge primary key, associating the attribute label and behavior label of each sample non-violation person with each identity information to obtain multiple mapping relationships.

[0175] When constructing a label database, the computer device uses the identity information of the non-violation personnel in each sample as the merge primary key, associates the attribute labels and behavior labels of the non-violation personnel in each sample with the identity information of the non-violation personnel in each sample, and obtains multiple mapping relationships, so that when querying the labels of the non-violation personnel in each sample, the labels of the non-violation personnel in each sample can be quickly located based on the associated identity information.

[0176] S903: Generate a personnel tag database according to multiple mapping relationships.

[0177] After obtaining a plurality of mapping relationships, the plurality of mapping relationships are stored in a database to obtain a personnel label database corresponding to the non-violation personnel in the sample.

[0178] It should be noted that the same method can be used to generate the corresponding personnel label database for multiple sample violators. Specifically, the identity information of multiple sample violators is obtained, and the identity information of each sample violator is used as the merge primary key. The attribute label and behavior label of each sample violator are associated with the identity information of each sample violator to obtain multiple mapping relationships, and then the multiple mapping relationships are stored in the database to obtain the personnel label database corresponding to each sample violator.

[0179] In the method for predicting illegal personnel provided in the embodiment of the present application, by obtaining the identity information of the non-illegal personnel in each sample, and then using each identity information as the merge primary key, the attribute label and behavior label of each non-illegal personnel in each sample are associated with each identity information to obtain multiple mapping relationships, and then a personnel label database is generated based on the multiple mapping relationships. In this method, by associating the attribute label and behavior label of each non-illegal personnel in each sample with the identity information of each non-illegal personnel in each sample, multiple mapping relationships can be obtained, and then the multiple mapping relationships are stored in the database, and a personnel label database can be obtained, which facilitates the label query of the subsequent non-illegal personnel in the sample.

[0180] Based on the above embodiment, an embodiment is provided to illustrate the application process after the above-mentioned generation of the personnel label database.

[0181] In an exemplary embodiment, Fig.10 As shown, the method also includes:

[0182] S1001, in response to a tag query request for a person to be queried, obtaining identity information of the person to be queried.

[0183] In an embodiment of the present application, when it is necessary to query a person's label, a label query request can be triggered based on the identity information of the person to be queried. In this way, the label query request will carry the identity information of the person to be queried.

[0184] In response to a tag query request for a person to be queried, the computer device may obtain the identity information of the person to be queried according to the information carried in the tag query request.

[0185] S1002, based on the identity information of the person to be queried, filter the attribute tags and behavior tags corresponding to the identity information from the person tag database.

[0186] After obtaining the identity information of the person to be queried, the computer device screens the attribute tags and behavior tags that match the identity information in the person tag database according to the association relationship between the identity information and the attribute tags and behavior tags of the person.

[0187] S1003: Determine the filtered attribute tags and behavior tags corresponding to the identity information as tags of the person to be queried.

[0188] After the computer device screens out the attribute tags and behavior tags that match the identity information of the person to be queried from the person tag database, it directly determines the screened out attribute tags and behavior tags corresponding to the identity information as the tags of the person to be queried.

[0189] In the method for predicting illegal persons provided in the embodiment of the present application, the identity information of the person to be queried is obtained by responding to the label query request of the person to be queried, and then the attribute label and behavior label corresponding to the identity information are screened from the personnel label database according to the identity information of the person to be queried, and then the screened attribute label and behavior label corresponding to the identity information are determined as the label of the person to be queried. In this method, when it is necessary to query the label of the person to be queried, the attribute label and behavior label corresponding to the identity information can be quickly matched from the personnel label database based on the identity information of the person to be queried. The matched attribute label and behavior label are the label of the person to be queried, and the efficiency of label query is improved by quickly matching the label with the identity information.

[0190] In addition, in an exemplary embodiment, the present application also provides an optional example of a method for predicting illegal personnel, such as Fig.11 As shown, the following steps may be included:

[0191] S1101, obtaining a training data set, a validation data set, and a test data set based on the attribute labels and behavior labels of each sample violator and the attribute labels and behavior labels of each sample non-violator.

[0192] S1102, obtaining a positive sample data set and a negative sample data set according to the training data set.

[0193] Among them, the positive sample data set is determined based on the attribute labels and behavior labels of the non-violated persons in each sample; the negative sample data set is determined based on the attribute labels and behavior labels of the violating persons in each sample.

[0194] S1103, inputting the positive sample data set and the negative sample data set into the neural network to obtain the positive sample prediction results and the negative sample prediction results.

[0195] S1104, obtaining a prediction loss value of the neural network according to the positive sample prediction result, the negative sample prediction result and a preset loss function.

[0196] S1105, based on the predicted loss value, adjust the model parameters of the neural network until the training is completed to obtain an initial model.

[0197] S1106, adjusting the model parameters of the initial model according to the verification data set to obtain an adjusted initial model.

[0198] S1107, testing the adjusted initial model according to the test data set to obtain a violation prediction model.

[0199] S1108, in response to a violation prediction request for a target person, obtaining attribute characteristics and behavior characteristics of the target person.

[0200] S1109, determining the target attribute label of the target person according to the attribute characteristics and the preset label rules; and determining the target behavior label of the target person according to the behavior characteristics and the label rules.

[0201] S1110, input the target attribute label and the target behavior label into the violation prediction model to obtain the violation prediction result of the target person.

[0202] The above processes of S1101 - S1110 can refer to the description of the above method embodiment, and the implementation principles and technical effects are similar, which will not be repeated here.

[0203] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0204] Based on the same inventive concept, the embodiment of the present application also provides a device for predicting illegal personnel for implementing the method for predicting illegal personnel involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more embodiments of the device for predicting illegal personnel provided below can refer to the limitations of the method for predicting illegal personnel above, and will not be repeated here.

[0205] In an exemplary embodiment, Fig.12 As shown, a violation person prediction device 1 is provided, comprising: a feature acquisition module 10 and a result determination module 20, wherein:

[0206] A feature acquisition module 10 is used to obtain the attribute characteristics and behavior characteristics of the target person in response to a violation prediction request for the target person;

[0207] The result determination module 20 is used to process the attribute characteristics and behavior characteristics using a preset violation prediction model to obtain a violation prediction result of the target person; the violation prediction model is trained based on the attribute labels and behavior labels of multiple sample violators and the attribute labels and behavior labels of multiple sample non-violators.

[0208] In one embodiment, the result determination module 20 is further configured to:

[0209] Determine the target attribute label of the target person according to the attribute characteristics and the preset label rules; and determine the target behavior label of the target person according to the behavior characteristics and the label rules;

[0210] The target attribute label and target behavior label are input into the violation prediction model to obtain the violation prediction result of the target person.

[0211] In one embodiment, the above-mentioned violator prediction device 1 further includes:

[0212] A training data acquisition module is used to acquire a training data set and a test data set based on the attribute labels and behavior labels of each sample violator and the attribute labels and behavior labels of each sample non-violation person;

[0213] The model training module is used to perform model training on the preset neural network according to the training data set to obtain an initial model;

[0214] The model testing module is used to test the initial model based on the test data set to obtain the violation prediction model.

[0215] In one embodiment, the above-mentioned violator prediction device 1 further includes:

[0216] The verification data set acquisition module is used to acquire the verification data set based on the attribute labels and behavior labels of each sample violator and the attribute labels and behavior labels of each sample non-violation person.

[0217] The model adjustment module is used to adjust the model parameters of the initial model according to the verification data set to obtain the adjusted initial model.

[0218] In one embodiment, the model training module is further used to:

[0219] According to the training data set, a positive sample data set and a negative sample data set are obtained; the positive sample data set is determined according to the attribute labels and behavior labels of the non-violated persons in each sample; the negative sample data set is determined according to the attribute labels and behavior labels of the violating persons in each sample; the positive sample data set and the negative sample data set are input into the neural network for model training to obtain an initial model.

[0220] In one embodiment, the model training module is further used to:

[0221] The positive sample data set and the negative sample data set are input into the neural network to obtain the positive sample prediction results and the negative sample prediction results; according to the positive sample prediction results, the negative sample prediction results and the preset loss function, the prediction loss value of the neural network is obtained; based on the prediction loss value, the model parameters of the neural network are adjusted until the training is completed to obtain the initial model.

[0222] In one embodiment, the model training module is further used to:

[0223] Perform feature processing on the attribute labels and behavior labels of each sample non-violation personnel to obtain the label values ​​corresponding to the attribute labels and behavior labels of each sample non-violation personnel, and determine the label values ​​corresponding to the attribute labels and behavior labels of each sample non-violation personnel as the positive sample data set;

[0224] The attribute labels and behavior labels of each sample violator are feature processed to obtain label values ​​corresponding to the attribute labels and behavior labels of each sample violator, and the label values ​​corresponding to the attribute labels and behavior labels of each sample violator are determined as negative sample data sets.

[0225] In one embodiment, the above-mentioned violator prediction device 1 further includes:

[0226] The first acquisition module is used to obtain the identity information of the non-violation personnel in each sample;

[0227] The relationship determination module is used to associate the attribute labels and behavior labels of the non-violation personnel in each sample with each identity information with each identity information, thereby obtaining multiple mapping relationships;

[0228] The database determination module is used to generate a personnel label database according to multiple mapping relationships.

[0229] In one embodiment, the above-mentioned violator prediction device 1 further includes:

[0230] A second acquisition module, configured to obtain identity information of the person to be queried in response to a tag query request of the person to be queried;

[0231] The tag filtering module is used to filter the attribute tags and behavior tags corresponding to the identity information from the personnel tag database according to the identity information of the person to be queried;

[0232] The label determination module is used to determine the filtered attribute labels and behavior labels corresponding to the identity information as the labels of the person to be queried.

[0233] Each module in the above-mentioned illegal personnel prediction device can be implemented in whole or in part by software, hardware and a combination thereof. Each of the above-mentioned modules can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute the operations corresponding to each of the above modules.

[0234] In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented:

[0235] In response to a violation prediction request for a target person, obtaining attribute characteristics and behavior characteristics of the target person;

[0236] The preset violation prediction model is used to process the attribute characteristics and behavioral characteristics to obtain the violation prediction results of the target person; the violation prediction model is trained based on the attribute labels and behavioral labels of multiple sample violators, as well as the attribute labels and behavioral labels of multiple sample non-violators.

[0237] The implementation principles and technical effects of each step implemented by the processor in the embodiment of the present application are similar to the principles of the above-mentioned violator prediction method, and will not be repeated here.

[0238] In one embodiment, a computer readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:

[0239] In response to a violation prediction request for a target person, obtaining attribute characteristics and behavior characteristics of the target person;

[0240] The preset violation prediction model is used to process the attribute characteristics and behavioral characteristics to obtain the violation prediction results of the target person; the violation prediction model is trained based on the attribute labels and behavioral labels of multiple sample violators, as well as the attribute labels and behavioral labels of multiple sample non-violators.

[0241] The implementation principles and technical effects of the various steps implemented when the computer program in the embodiment of the present application is executed by the processor are similar to the principles of the above-mentioned violator prediction method, and will not be repeated here.

[0242] In one embodiment, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the following steps:

[0243] In response to a violation prediction request for a target person, obtaining attribute characteristics and behavior characteristics of the target person;

[0244] The preset violation prediction model is used to process the attribute characteristics and behavioral characteristics to obtain the violation prediction results of the target person; the violation prediction model is trained based on the attribute labels and behavioral labels of multiple sample violators, as well as the attribute labels and behavioral labels of multiple sample non-violators.

[0245] The implementation principles and technical effects of the various steps implemented when the computer program in the embodiment of the present application is executed by the processor are similar to the principles of the above-mentioned violator prediction method, and will not be repeated here.

[0246] It should be noted that the personnel information (including but not limited to personnel equipment information, personnel personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the personnel or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0247] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.

[0248] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0249] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A method for predicting illegal personnel, characterized in that: The method comprises: In response to a violation prediction request for a target person, obtaining attribute characteristics and behavior characteristics of the target person; The attribute characteristics and the behavior characteristics are processed using a preset violation prediction model to obtain a violation prediction result of the target person; the violation prediction model is trained based on the attribute labels and behavior labels of multiple sample violators and the attribute labels and behavior labels of multiple sample non-violators.

2. The method according to claim 1, characterized in that The method of using a preset violation prediction model to process the attribute characteristics and the behavior characteristics to obtain a violation prediction result of the target person includes: Determine the target attribute label of the target person according to the attribute characteristics and the preset labeling rules; and determine the target behavior label of the target person according to the behavior characteristics and the labeling rules; The target attribute label and the target behavior label are input into the violation prediction model to obtain the violation prediction result of the target person.

3. The method according to claim 1 or 2, characterized in that: The training process of the violation prediction model includes: Obtaining a training data set and a test data set based on the attribute labels and behavior labels of each of the sample violators and the attribute labels and behavior labels of each of the sample non-violation persons; According to the training data set, a preset neural network is trained to obtain an initial model; The initial model is tested according to the test data set to obtain the violation prediction model.

4. The method according to claim 3, characterized in that Before testing the initial model according to the test data set to obtain the violation prediction model, the method further includes: Acquire a verification data set based on the attribute labels and behavior labels of each of the sample violators and the attribute labels and behavior labels of each of the sample non-violation persons; According to the verification data set, the model parameters of the initial model are adjusted to obtain an adjusted initial model.

5. The method according to claim 3, characterized in that: The step of performing model training on a preset neural network according to the training data set to obtain an initial model includes: According to the training data set, a positive sample data set and a negative sample data set are obtained; the positive sample data set is determined according to the attribute labels and behavior labels of each sample non-violation person; the negative sample data set is determined according to the attribute labels and behavior labels of each sample violation person; The positive sample data set and the negative sample data set are input into the neural network for model training to obtain the initial model.

6. The method according to claim 5, characterized in that The step of inputting the positive sample data set and the negative sample data set into the neural network for model training to obtain the initial model includes: Inputting the positive sample data set and the negative sample data set into the neural network to obtain positive sample prediction results and negative sample prediction results; Obtaining a prediction loss value of the neural network according to the positive sample prediction result, the negative sample prediction result and a preset loss function; Based on the predicted loss value, the model parameters of the neural network are adjusted until the training is completed to obtain the initial model.

7. The method according to claim 5, characterized in that The step of obtaining a positive sample data set and a negative sample data set according to the training data set includes: Performing feature processing on the attribute labels and behavior labels of each of the sample non-violation personnel to obtain label values ​​corresponding to the attribute labels and behavior labels of each of the sample non-violation personnel, and determining the label values ​​corresponding to the attribute labels and behavior labels of each of the sample non-violation personnel as the positive sample data set; Feature processing is performed on the attribute labels and behavior labels of each of the sample violators to obtain label values ​​corresponding to the attribute labels and behavior labels of each of the sample violators, and the label values ​​corresponding to the attribute labels and behavior labels of each of the sample violators are determined as the negative sample data set.

8. The method according to claim 7, characterized in that The method further comprises: Obtaining the identity information of the non-violated persons in each sample; Using each identity information as the merge primary key, the attribute label and behavior label of each sample non-violation person are associated with each identity information to obtain multiple mapping relationships; A personnel label database is generated according to the multiple mapping relationships.

9. The method according to claim 8, characterized in that The method further comprises: In response to a tag query request of a person to be queried, obtaining identity information of the person to be queried; According to the identity information of the person to be queried, the attribute label and the behavior label corresponding to the identity information are screened from the person label database; The screened attribute labels and behavior labels corresponding to the identity information are determined as labels of the person to be queried.

10. A device for predicting illegal personnel, characterized in that: The device comprises: A feature acquisition module, used to obtain attribute characteristics and behavior characteristics of a target person in response to a violation prediction request for the target person; The result determination module is used to process the attribute characteristics and the behavior characteristics using a preset violation prediction model to obtain a violation prediction result of the target person; the violation prediction model is trained based on the attribute labels and behavior labels of multiple sample violators and the attribute labels and behavior labels of multiple sample non-violators.

Images