Re-crime analysis and early warning system and method for target population

Abstract

The invention provides a target population-oriented re-crime analysis and early warning system and method, and the system comprises a data collection module which is used for obtaining the related management data of a target person from a pre-integrated related data system; the data analysis module is used for processing and analyzing the related management data according to a pre-stored data processing and analysis tool to obtain a data analysis report, and adding the data analysis report into a sample database; the early warning module is used for inputting the data analysis report into a pre-constructed behavior mode and psychological tendency prediction model to obtain an early warning signal, and performing early warning according to the early warning signal; and the control module is used for inputting the early warning signal into a pre-constructed intelligent decision support model to obtain the risk level of the target personnel and corresponding management measures. According to the invention, the re-crime analysis and early warning which is high in efficiency and universal and faces the target crowd is realized without manual participation.

Description

Technical Field

[0001] The present invention relates to the technical field of reoffending warning, and in particular to a reoffending analysis and warning system and method for a target population. Background Art

[0002] At present, the early warning for people who make mistakes and are corrected mainly relies on manual monitoring and simple data analysis. For example, experienced staff make judgments and early warnings based on the offender's daily reform performance and relevant scale tests before being released. There is a lack of in-depth understanding of the behavior and psychology of people who make mistakes and are corrected, and there is a lack of early warning mechanisms. The labor cost is high and the effect depends on prior experience, which is inefficient and not universal.

[0003] The existing technology has the problems of high labor cost for early warning of personnel who make mistakes and correct them, low efficiency and non-universality. Summary of the invention

[0004] The present invention provides a reoffending analysis and early warning system and method for a target population, so as to solve the defects of the prior art such as high labor cost, low efficiency and non-universality, and realize a reoffending analysis and early warning for a target population that does not require human participation, has high efficiency and is universal.

[0005] The present invention provides a recidivism analysis and early warning system for a target population, comprising:

[0006] Data collection module, used to obtain relevant management data of target personnel from pre-integrated relevant data systems;

[0007] A data analysis module, used to process and analyze the relevant management data according to a pre-stored data processing and analysis tool, obtain a data analysis report, and add the data analysis report to a sample database; wherein the sample database includes a large amount of target personnel sample data;

[0008] An early warning module is used to input the data analysis report into a pre-built behavior pattern and psychological tendency prediction model to obtain an early warning signal, and issue an early warning according to the early warning signal; wherein the behavior pattern and psychological tendency prediction model is obtained by training the sample database after feature extraction based on a pre-selected machine learning algorithm;

[0009] A control module is used to input the warning signal into a pre-built intelligent decision support model to obtain the risk level of the target person and the corresponding management measures; wherein the intelligent decision support model is trained based on a pre-selected machine learning algorithm using the sample database after refined classification.

[0010] According to a reoffending analysis and early warning system for a target population provided by the present invention, the behavior pattern and psychological tendency prediction model is obtained by training the sample database after feature extraction based on the pre-selected machine learning algorithm, specifically including:

[0011] Performing data preprocessing on the target person sample data in the sample database; wherein the target person sample data at least includes personal information, criminal history, and reform performance of the target population; the data preprocessing at least includes data cleaning and data coding;

[0012] Dividing the sample database after the data preprocessing into a training set, a validation set and a test set according to a preset ratio;

[0013] Construct multiple base learners based on pre-selected machine learning algorithms;

[0014] Based on the base learner, the training set is used for training, a preset feature extraction algorithm is used to extract basic features from the training set, and a pre-selected deep learning model is used to extract high-level features from the training set, the basic features and the high-level features are fused, and the fusion result is fitted by the base learner to obtain a training result;

[0015] Integrate the plurality of training results using a preset integration algorithm, and perform parameter tuning to obtain a target prediction model;

[0016] The target prediction model is evaluated using the validation set and the test set, and the target prediction model is iteratively optimized according to the evaluation results until preset conditions are met, and a behavior pattern and psychological tendency prediction model is output.

[0017] According to a reoffending analysis and early warning system for a target population provided by the present invention, the intelligent decision support model includes a risk prediction sub-model and a management strategy sub-model; the intelligent decision support model is obtained by training the sample database after refined classification based on a pre-selected machine learning algorithm, specifically including:

[0018] According to the target person sample data in the sample database, the corresponding target person samples are finely classified into different risk levels to obtain classification results;

[0019] Building a basic risk prediction model based on a pre-selected first machine learning algorithm, training the basic risk prediction model using the sample database and the classification results, fitting the classification results, and obtaining a risk prediction sub-model;

[0020] Obtaining a marking result for the management measures corresponding to the risk level marking in the sample database;

[0021] A basic management strategy model is constructed based on a pre-selected second machine learning algorithm, and training is performed based on the basic management strategy model using the sample database and the marking results, and the marking results are fitted to obtain a management strategy sub-model.

[0022] According to a reoffending analysis and early warning system for a target population provided by the present invention, the corresponding target person samples are finely classified according to the target person sample data in the sample database, and are divided into different risk levels to obtain classification results, which specifically includes:

[0023] According to the target personnel sample data, the target personnel sample is graded by using a K-means or hierarchical clustering algorithm to obtain a first graded result sample;

[0024] The first division result sample is further classified using a pre-selected classification model according to the target person sample data to obtain a classification result.

[0025] According to a reoffending analysis and early warning system for a target population provided by the present invention, the acquisition of relevant management data of the target person from a pre-integrated relevant data system specifically includes:

[0026] Extracting key information of the target person from the relevant data system;

[0027] Extracting unstructured data from the key information, and performing text analysis on the unstructured data using a pre-selected natural language processing algorithm to obtain analysis results;

[0028] Relevant management data of the target personnel is obtained according to the analysis results and the structured data in the key information.

[0029] According to a reoffending analysis and early warning system for a target population provided by the present invention, the relevant management data of the target person is obtained according to the analysis result and the structured data in the key information, and then the following is further included:

[0030] Perform data cleaning and verification on the relevant management data of the target personnel.

[0031] The present invention also provides a recidivism analysis and early warning method for a target population, comprising:

[0032] Obtain relevant management data of target personnel from pre-integrated relevant data systems;

[0033] Processing and analyzing the relevant management data according to a pre-stored data processing and analysis tool to obtain a data analysis report, and adding the data analysis report to a sample database; wherein the sample database includes a large amount of target personnel sample data;

[0034] Input the data analysis report into a pre-built behavior pattern and psychological tendency prediction model to obtain an early warning signal, and issue an early warning according to the early warning signal; wherein the behavior pattern and psychological tendency prediction model is obtained by training the sample database after feature extraction based on a pre-selected machine learning algorithm;

[0035] The warning signal is input into a pre-built intelligent decision support model to obtain the risk level of the target person and the corresponding management measures; wherein the intelligent decision support model is trained based on a pre-selected machine learning algorithm using the sample database after refined classification.

[0036] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the reoffending analysis and early warning method for a target population as described in any one of the above-mentioned methods is implemented.

[0037] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the method for analyzing and warning reoffending for a target population as described in any one of the above is implemented.

[0038] The present invention also provides a computer program product, including a computer program, which, when executed by a processor, implements any of the above-mentioned reoffending analysis and early warning methods for a target population.

[0039] The reoffending analysis and early warning system and method for target population provided by the present invention comprises: a data acquisition module for acquiring relevant management data of target personnel from a pre-integrated relevant data system; a data analysis module for processing and analyzing the relevant management data according to a pre-stored data processing and analysis tool to obtain a data analysis report, and adding the data analysis report to a sample database; wherein the sample database comprises a large amount of sample data of target personnel; an early warning module for inputting the data analysis report into a pre-constructed behavior pattern and psychological tendency prediction model to obtain an early warning signal, and issuing an early warning according to the early warning signal; wherein the behavior pattern and psychological tendency prediction model is obtained by training the sample database after feature extraction based on a pre-selected machine learning algorithm; and a control module for inputting the early warning signal into a pre-constructed intelligent decision support model to obtain the risk level of the target personnel and the corresponding management measures; wherein the intelligent decision support model is obtained by training the sample database after refined classification based on a pre-selected machine learning algorithm. The present invention acquires complete and accurate relevant management data of target personnel by integrating relevant data systems, processes and analyzes the data efficiently through data processing and analysis tools, and realizes accurate analysis of target personnel through pre-trained behavior pattern and psychological tendency prediction models and intelligent decision support models, thereby realizing reoffending analysis and early warning for target populations that does not require human intervention, is highly efficient, and has universality, which helps prevent target personnel from committing crimes again. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0041] Figure 1 It is a structural schematic diagram of the reoffending analysis and early warning system for target groups provided by the present invention;

[0042] Figure 2 This is one of the flow charts of the target population reoffending analysis and early warning method provided by the present invention;

[0043] Figure 3 This is the second flow chart of the target population reoffending analysis and early warning method provided by the present invention;

[0044] Figure 4 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0045] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0046] Combine the following Figure 1-Figure 3 The present invention describes the target population-oriented reoffending analysis and early warning system and method, such as Figure 1 As shown, the system includes:

[0047] The data acquisition module 110 is used to obtain relevant management data of the target personnel from a pre-integrated relevant data system.

[0048] Specifically, the target personnel in the embodiment of the present invention refers to the personnel who have made mistakes and are being corrected, and the relevant data system refers to any system associated with the data of the personnel who have made mistakes and are being corrected. The acquired information-related management data includes but is not limited to personal information, criminal history, reform performance, etc. In terms of technical implementation, the data acquisition module 110 needs to be integrated with multiple systems to ensure the integrity and accuracy of the data. In some embodiments, the relevant data system includes at least a prison management system.

[0049] It is understood that integration with the prison management system (JMS) is an important part of data collection in the embodiment of the present invention. Through JMS, the data collection module 110 can obtain key information such as the prison record, daily performance, rewards and punishments of the offender in real time.

[0050] Further, based on the above embodiments, in some embodiments, the relevant data system includes other relevant systems of the judicial department in addition to the prison management system, such as the court case management system, the procuratorate case handling system, etc. It should be emphasized that in the case of multiple relevant data systems, in order to ensure the consistency and sharing of data, the interface design needs to be fully considered, and necessary adjustments should be made on the basis of the existing system. The present invention does not limit the specific interface design, and it can be selected according to the actual operation process.

[0051] Furthermore, in some embodiments, the target person may also be other special persons who need reoffending analysis and early warning, and can be extended to people other than those who have made mistakes and are correcting them according to actual needs. The present invention does not impose any limitation on this.

[0052] It should be emphasized that the data collection module 110 plays a vital role in the information management of the target person. The core responsibility of this module is to comprehensively and accurately collect various data related to the target person.

[0053] In addition, due to the integration of multiple systems, the key information directly obtained by the data acquisition module 110 includes structured data from different sources; it also includes unstructured data from different sources, such as visit reports, interview reports, etc. Therefore, it is necessary to integrate them.

[0054] In some embodiments, the obtaining of relevant management data of the target personnel from the pre-integrated relevant data system specifically includes:

[0055] Extracting key information of the target person from the relevant data system;

[0056] Extracting unstructured data from the key information, and performing text analysis on the unstructured data using a pre-selected natural language processing algorithm to obtain analysis results;

[0057] Relevant management data of the target personnel is obtained according to the analysis results and the structured data in the key information.

[0058] Specifically, in order to effectively integrate these data, the module needs to use pre-selected natural language processing (NLP) technology to perform text analysis, extract key information and convert it into structured data. It should be noted that the present invention does not limit the natural language processing (NLP) technology (or natural language processing algorithm), and it can be selected according to actual needs.

[0059] In some embodiments, obtaining relevant management data of the target person according to the analysis result and the structured data in the key information further includes:

[0060] Perform data cleaning and verification on the relevant management data of the target personnel.

[0061] Specifically, in order to ensure the authenticity and reliability of the data, the data needs to be cleaned and verified to remove duplicate, erroneous or false information.

[0062] In addition, security is an important factor that must be considered during the data collection process. In some embodiments, the data collection module 110 uses encryption technology to ensure the security of data transmission and storage, and sets a strict access control mechanism to prevent unauthorized access and data leakage.

[0063] In summary, the details of integration with multiple systems, application of NLP technology to process unstructured data, data cleaning and verification, and related measures to ensure data security together constitute the complete technical framework of the data collection module 110, providing a solid data foundation for the information management of personnel who correct errors.

[0064] The data analysis module 120 is used to process and analyze the relevant management data according to a pre-stored data processing and analysis tool to obtain a data analysis report, and add the data analysis report to a sample database; wherein the sample database includes a large amount of target personnel sample data.

[0065] Specifically, the data analysis module 120 is responsible for processing and analyzing the collected data, and based on the data analysis report obtained through processing and analysis, uses advanced big data and artificial intelligence technology to conduct in-depth learning and prediction of the behavior patterns and psychological tendencies of the error correction personnel. It should be emphasized that the data analysis module 120 plays a vital role in the information management of the target personnel.

[0066] Specifically, the data analysis module 120 uses pre-stored data processing and analysis tools. In some embodiments, the data processing and analysis tools include tools for data analysis and data processing, such as Hadoop distributed file system, Spark big data processing framework, etc. The above examples do not represent limitations on the embodiments of the present invention, and in actual operation, the data processing and analysis tools can be selected according to needs.

[0067] It should be pointed out that data processing and analysis tools are used to efficiently process and analyze massive amounts of relevant management data to obtain data analysis reports. These tools support the rapid reading, storage and processing of large-scale data, improving the efficiency and accuracy of data analysis.

[0068] Furthermore, the data analysis report is added to the sample database. It should be emphasized that the sample database is used for training the subsequent prediction model, and the sample database includes a large amount of target personnel sample data, which is also collected by the data collection module 110 and processed by the data analysis module 120. Furthermore, in some embodiments, the target personnel sample data may also include data from other sources.

[0069] In addition, the data analysis module 120 includes data security and privacy protection. In some embodiments, during the data transmission and storage process, the data analysis module 120 uses encryption technology and access control mechanisms to ensure data security. At the same time, through technical means such as anonymization processing, the privacy rights and interests of the person who made the error correction are protected.

[0070] Furthermore, the sample database is used to construct a behavior pattern and psychological tendency prediction model. Specifically, the behavior pattern and psychological tendency prediction model is obtained by training the sample database after feature extraction based on a pre-selected machine learning algorithm.

[0071] In some embodiments, the behavior pattern and psychological tendency prediction model is obtained by training the sample database after feature extraction based on the pre-selected machine learning algorithm, specifically including:

[0072] Performing data preprocessing on the target person sample data in the sample database; wherein the target person sample data at least includes personal information, criminal history, and reform performance of the target population; the data preprocessing at least includes data cleaning and data coding;

[0073] Dividing the sample database after the data preprocessing into a training set, a validation set and a test set according to a preset ratio;

[0074] Construct multiple base learners based on pre-selected machine learning algorithms;

[0075] Based on the base learner, the training set is used for training, a preset feature extraction algorithm is used to extract basic features from the training set, and a pre-selected deep learning model is used to extract high-level features from the training set, the basic features and the high-level features are fused, and the fusion result is fitted by the base learner to obtain a training result;

[0076] Integrate the plurality of training results using a preset integration algorithm, and perform parameter tuning to obtain a target prediction model;

[0077] The target prediction model is evaluated using the validation set and the test set, and the target prediction model is iteratively optimized according to the evaluation results until preset conditions are met, and a behavior pattern and psychological tendency prediction model is output.

[0078] Specifically, the training of the behavior pattern and psychological tendency prediction model needs to occur before the early warning module 130 applies it, so the training process can be placed in the data analysis module 120. Specifically, the behavior pattern and psychological tendency prediction model uses a pre-selected machine learning algorithm. In some embodiments, the pre-selected machine learning algorithm includes support vector machine, random forest and neural network. These algorithms can analyze and model the personal information, criminal history, reform performance and other data of the person who made the mistake, so as to predict his behavioral trend and the risk of re-offending after release.

[0079] In the prediction process, deep learning technology is also used to build high-precision and high-generalization prediction models through training and learning of a large amount of historical data. These models can fully tap the hidden information and patterns in the data of error correction personnel, providing a more accurate basis for decision-making.

[0080] In more detail, the training steps are as follows:

[0081] First, the target personnel sample data in the sample database is preprocessed. Specifically, data preprocessing includes data cleaning and data encoding.

[0082] Data cleaning includes removing duplicate, missing or outlier data records. On this basis, feature selection is also required to retain features with higher correlation. Furthermore, feature selection includes selecting features with higher correlation as input based on personal information, criminal history, reform performance, etc. in the sample data of the target person. It should be noted that in some embodiments, the selection method with higher correlation includes: calculating the correlation coefficient between personal information, criminal history, reform performance, etc. in the data and the target variable to quantify the linear relationship between them. After that, selection is performed through a set threshold, and features with correlation coefficients exceeding the set threshold are retained as features with higher correlation.

[0083] Afterwards, data encoding is performed on the obtained features. In some embodiments, the encoding is performed using one-hot encoding or label encoding.

[0084] Furthermore, the sample database after data preprocessing is divided into a training set, a validation set and a test set according to a preset ratio for subsequent training.

[0085] During the training process, feature extraction is first required, and the features to be extracted include basic features and deep features. Specifically, a preset feature extraction algorithm is used to extract basic features from the training set, and a pre-selected deep learning model is used to extract advanced features from the training set. In some embodiments, the preset feature extraction algorithm includes traditional statistical methods or domain knowledge, and the pre-selected deep learning model includes a convolutional neural network, a recurrent neural network, etc.

[0086] It is also necessary to fuse the basic features and the deep features to form a more comprehensive feature vector. The embodiment of the present invention does not limit the specific method of fusion, and the fusion method can be selected according to the needs during the actual operation.

[0087] After that, multiple base learners are constructed based on the pre-selected machine learning algorithms. In this process, multiple machine learning algorithms with good performance can be selected as base learners. The pre-selected machine learning algorithms include support vector machines, random forests, gradient boosting machines, etc. Each base learner is trained using the training set to obtain their respective training results, i.e., prediction models.

[0088] After obtaining the training results, it is also necessary to integrate the multiple training results using a preset integration algorithm to integrate the training results of multiple base learners. In some embodiments, the preset integration algorithm includes a voting method or a weighted average method.

[0089] At the same time, the parameters of the integrated model can be tuned, for example, through cross-validation or grid search methods, to tune the model parameters, improve the prediction performance of the model, and finally obtain the target prediction model.

[0090] Before outputting the final training results, the target prediction model needs to be evaluated using the validation set and the test set. The evaluation content includes evaluation indicators such as accuracy, recall rate, and F1 value. If the preset conditions are met, the final training results are output as the behavior pattern and psychological tendency prediction model. If the preset conditions are not met, the model is iteratively optimized and the fitting training steps are repeated until the actual application requirements are met. Among them, the preset conditions include the threshold of the evaluation indicator.

[0091] In summary, the complete technical framework of the data analysis module 120 is composed of efficient data processing and analysis tools, the application of machine learning algorithms, the combination of deep learning technology, and data security and privacy protection measures, which provides important decision-making support for preventing people who have made mistakes and corrected their crimes from committing crimes again.

[0092] The early warning module 130 is used to input the data analysis report into a pre-constructed behavior pattern and psychological tendency prediction model to obtain an early warning signal and issue an early warning based on the early warning signal; wherein the behavior pattern and psychological tendency prediction model is trained based on a pre-selected machine learning algorithm using the sample database after feature extraction.

[0093] It should be noted that the early warning module 130, as a core component of the target person information management system, undertakes the key task of preventing recidivism. Based on the results of data analysis, once the module detects that there is a potential risk in the behavior or psychological condition of the person who has made a mistake, it will immediately issue an early warning. The purpose of the early warning is to promptly remind relevant personnel to take measures to prevent potential criminal behavior from occurring.

[0094] Furthermore, the data analysis report is first input into the pre-built behavior pattern and psychological tendency prediction model to obtain an early warning signal, and an early warning is issued according to the early warning signal. Specifically, the behavior pattern and psychological tendency prediction model is used to evaluate the behavior and psychological condition of the person who made the error correction and obtain the early warning signal.

[0095] It should be pointed out that the core of the early warning module is the design of the early warning algorithm (i.e., the behavioral pattern and psychological tendency prediction model). In order to ensure the accuracy and timeliness of the early warning, a variety of machine learning algorithms are used, such as logistic regression, support vector machine, decision tree, random forest and neural network. These algorithms can model and predict the behavior and psychological trends of the error correction personnel, and predict their future behavior trends by analyzing the patterns and rules in the data.

[0096] In order to improve the accuracy of early warning, the early warning module also uses deep learning algorithms, such as convolutional neural networks (CNN) and recurrent neural networks (RNN). These algorithms can automatically extract features from data and make high-precision predictions on the behavior and psychology of people who make mistakes based on a large amount of historical data. Through deep learning technology, the early warning module can more accurately predict potential crime risks and provide strong support for timely intervention.

[0097] The early warning module also pays special attention to algorithm optimization and parameter adjustment. Through the combination and cross-validation of different algorithms, the model that best suits the data characteristics and prediction tasks is found. At the same time, parameter setting is also very critical. Through careful adjustment, the accuracy and timeliness of early warning can be further improved.

[0098] In addition to algorithm selection and parameter adjustment, the early warning module also has automatic learning and self-adaptation capabilities. As data continues to increase and update, the early warning module can continuously learn and improve the model to adapt to new data distribution and changes, thereby continuously improving the accuracy of early warnings.

[0099] In summary, the application of multiple machine learning algorithms, the combination of deep learning algorithms, algorithm optimization and parameter adjustment, as well as automatic learning and adaptive capabilities together constitute the complete technical framework of the early warning module 130, providing strong support for preventing people who have made corrections from committing crimes again.

[0100] In order to explain the modules 110-130 in more detail, a specific embodiment is given: Figure 2 As shown, let D be the personal information, criminal history, reform performance and other data of the person who made mistakes and corrected his mistakes, the output of the early warning module be Y, and the output of the control module be X.

[0101] Data collection module: D = {d1, d2, ..., dn}, where di represents a certain data of the i-th person who made an error and corrected it (such as age, gender, crime type, etc.).

[0102] Data analysis module: By learning and analyzing D, a prediction model F(D) of the behavior and psychological trends of error-correcting personnel is trained.

[0103] Early warning module: Based on the prediction model F(D), the behavior and psychological status of the person who made the mistake and corrected it are evaluated to obtain the early warning signal Y.

[0104] Control module: Take corresponding management measures X according to early warning signal Y.

[0105] Among them, the mathematical description of the early warning module can be as follows:

[0106] Y=F(D)

[0107] Among them, F is a function, which represents the mapping relationship from data D to warning signal Y. F can be a linear function, a nonlinear function or a neural network, etc. By adjusting the parameters and structure of F, the performance of the warning module can be optimized and the accuracy and timeliness of the warning can be improved.

[0108] The control module 140 is used to input the warning signal into a pre-built intelligent decision support model to obtain the risk level of the target person and the corresponding management measures.

[0109] It is understandable that the control module 140, as a key component of the error correction personnel information management system, undertakes the important task of implementing corresponding management measures according to the early warning information. This module ensures timely intervention and effective management of high-risk error correction personnel through close cooperation with the early warning module.

[0110] It should be noted that the control module 140 is highly flexible and adaptable, and can quickly formulate and implement personalized management measures based on early warning information. This is due to the intelligent decision support model within the module. The intelligent decision support model is based on big data analysis and artificial intelligence technology, and deeply mines and processes the data of error correction personnel to achieve accurate management decisions. Specifically, the intelligent decision support model is trained based on a pre-selected machine learning algorithm using a refined classified sample database.

[0111] In some embodiments, the intelligent decision support model includes a risk prediction sub-model and a management strategy sub-model; the intelligent decision support model is obtained by training the sample database after refined classification based on a pre-selected machine learning algorithm, specifically including:

[0112] According to the target person sample data in the sample database, the corresponding target person samples are finely classified into different risk levels to obtain classification results;

[0113] Building a basic risk prediction model based on a pre-selected first machine learning algorithm, training the basic risk prediction model using the sample database and the classification results, fitting the classification results, and obtaining a risk prediction sub-model;

[0114] Obtaining a marking result for the management measures corresponding to the risk level marking in the sample database;

[0115] A basic management strategy model is constructed based on a pre-selected second machine learning algorithm, and training is performed based on the basic management strategy model using the sample database and the marking results, and the marking results are fitted to obtain a management strategy sub-model.

[0116] Specifically, as the core of the control module, the intelligent decision support model uses a variety of advanced algorithms to achieve accurate management of offenders. In more detail, based on big data analysis and machine learning technology, the intelligent decision support model mines the historical data of offenders and combines their personal characteristics, criminal history, reform performance and other factors to predict their risk of reoffending.

[0117] In some embodiments, the first machine learning algorithm uses an integrated learning method, such as a random forest or a gradient boosting tree, to build a model that can accurately predict risks by training a large amount of historical data. Specifically, the first machine learning algorithm is used to build a basic risk prediction model, and the sample database that has been refined and classified is fitted to train a risk prediction sub-model.

[0118] In some embodiments, the method of finely classifying the corresponding target personnel samples according to the target personnel sample data in the sample database, dividing them into different risk levels, and obtaining classification results specifically includes:

[0119] According to the K-means or hierarchical clustering algorithm, the target personnel sample is graded to obtain a first graded result sample;

[0120] The first division result sample is further classified using a pre-selected classification model according to the target person sample data to obtain a classification result.

[0121] Specifically, in order to achieve refined classification of personnel who have made mistakes and are correcting their mistakes, the control module 140 adopts an algorithm based on clustering and classification. Specifically, the clustering algorithm uses methods such as K-means or hierarchical clustering to divide the target personnel sample into different risk level groups according to the multi-dimensional feature data (such as crime type, reform performance, social support, etc.) included in the target personnel sample data. Through cluster analysis, individuals with similar characteristics and behavior patterns can be identified, providing a basis for subsequent management measures.

[0122] Then, based on clustering, the pre-selected classification model is used to further use classification algorithms such as support vector machines and logistic regression to accurately classify the error-correcting personnel and obtain classification results. These algorithms learn the classification boundaries between different risk levels by training historical data with labels, thereby achieving accurate classification of new individuals.

[0123] It should be emphasized that the pre-selected classification model can use an existing model, or can be obtained by training after labeling the sample database.

[0124] The embodiment of the present invention divides people into different risk levels by analyzing multi-dimensional data such as personal characteristics, criminal history, and reform performance of people who have made mistakes and are correcting them, thereby providing a basis for subsequent management measures.

[0125] Naturally, the risk prediction sub-model obtained by training can also perform optimization steps such as integration of multiple training results, parameter tuning, and evaluation similar to the behavior pattern and psychological tendency prediction model, which will not be described in detail in the present invention. The risk prediction sub-model obtained by training can output the risk prediction result, that is, the risk level of the target person.

[0126] Afterwards, for each risk level, it is necessary to further generate a personalized management strategy based on the trained management strategy sub-model. The generated management strategy can be input as update data into the data acquisition module 110 and the data analysis module 120 for data update, such as Figure 2 shown.

[0127] Specifically, the second machine learning algorithm includes a decision tree or a deep learning model, which automatically recommends appropriate psychological counseling, vocational training or other preventive measures based on the specific circumstances of the person who made the mistake. The algorithm continuously optimizes the strategy generation logic to ensure the pertinence and effectiveness of management measures, and builds a model that can accurately predict management strategies by training a large amount of historical data. In the specific training process, the second machine learning algorithm is used to build a basic management strategy model, fit the sample database marked with management strategies, and train to obtain a management strategy sub-model. The core of this step is to use the algorithm to analyze and mine the multi-dimensional data of the person who made the mistake to discover the patterns and associations therein.

[0128] It should be emphasized that in the process of management strategy labeling, it is necessary to refer to the personal characteristics, criminal history, reform performance and other data of the target person, and combine historical data and existing knowledge for labeling. Naturally, the trained management strategy sub-model will consider the personal characteristics, criminal history, reform performance and other data of the target person, and automatically select the appropriate management strategy based on historical data and existing knowledge.

[0129] In actual application, the trained management strategy sub-model uses the prediction results output by the risk prediction sub-model as the basis for subsequent management strategy formulation and recommends management strategies.

[0130] During the automatic recommendation process, the trained management strategy sub-model will make a series of decisions and calculations. For example, for people with higher risk levels, the trained management strategy sub-model may recommend more stringent and intensive management measures, such as increasing the number of psychological counseling sessions or strengthening the intensity of vocational training; while for people with lower risk levels, more relaxed and flexible management measures may be recommended to help them better integrate into society. In addition, the trained management strategy sub-model will also take into account the personalized needs of the target person, such as personal interests and hobbies, career planning, etc., to ensure that the recommended management strategy meets both the overall management requirements and the individual's development needs.

[0131] Furthermore, in some embodiments, the generated personalized management strategy will be input as update data to the data acquisition module 110 and the data analysis module 120 for data update and further optimization. In this way, the entire system can form a closed loop, continuously improving the pertinence and effectiveness of management measures through data collection, analysis and model optimization.

[0132] It should be noted that the automatic recommendation process is not static, but needs to be constantly adjusted and optimized according to actual conditions. For example, when new data is input into the system, the algorithm needs to be retrained and adjusted to adapt to the new environment and needs. At the same time, the performance of the algorithm needs to be continuously evaluated and improved to ensure that it can continue to make automatic recommendations effectively.

[0133] It is also necessary to explain that a series of personalized management measures have been taken for high-risk error correction personnel. First of all, psychological counseling is an important part. The module cooperates with professional psychological counseling institutions to provide psychological assessment, consultation and treatment services for high-risk personnel. Through psychological counseling, they can help them adjust their mentality, solve psychological problems, and improve their ability to adapt to society.

[0134] In addition, vocational training is also an important means to reduce the risk of reoffending. By cooperating with vocational training institutions, we provide vocational skills training for people who have made mistakes according to their interests, skills and market needs. Through training, they can master a skill, increase employment opportunities and improve their ability to integrate into society.

[0135] In addition to psychological counseling and vocational training, other preventive measures can also be taken, such as regular follow-up visits, community corrections, etc. These measures are aimed at strengthening communication with people who have made mistakes and are correcting them, timely understanding their living conditions and psychological changes, and providing necessary support and help.

[0136] In summary, the application of intelligent decision support models, refined classification of target personnel, personalized psychological counseling and vocational training, and the implementation of other preventive measures together constitute the complete technical framework of the control module 140, which provides strong support for reducing the risk of reoffending by those who have made corrections.

[0137] The reoffending analysis and early warning system for target populations provided by the present invention comprises: a data acquisition module for acquiring relevant management data of target personnel from a pre-integrated relevant data system; a data analysis module for processing and analyzing the relevant management data according to a pre-stored data processing and analysis tool to obtain a data analysis report, and adding the data analysis report to a sample database; wherein the sample database comprises a large amount of sample data of target personnel; an early warning module for inputting the data analysis report into a pre-constructed behavior pattern and psychological tendency prediction model to obtain an early warning signal, and issuing an early warning according to the early warning signal; wherein the behavior pattern and psychological tendency prediction model is obtained by training the sample database after feature extraction based on a pre-selected machine learning algorithm; and a control module for inputting the early warning signal into a pre-constructed intelligent decision support model to obtain the risk level of the target personnel and the corresponding management measures; wherein the intelligent decision support model is obtained by training the sample database after refined classification based on a pre-selected machine learning algorithm. The present invention acquires complete and accurate relevant management data of target personnel by integrating relevant data systems, processes and analyzes the data efficiently through data processing and analysis tools, and realizes accurate analysis of personnel who have made corrections through pre-trained behavior patterns and psychological tendency prediction models and intelligent decision support models, thereby realizing a reoffending analysis and early warning for the target population that does not require human intervention, is highly efficient, and has universality, which helps prevent reoffending by personnel who have made corrections.

[0138] The following is a description of the target population-oriented reoffending analysis and early warning method provided by the present invention. The target population-oriented reoffending analysis and early warning method described below and the target population-oriented reoffending analysis and early warning system described above can be referenced to each other. Figure 3 This is the second flow chart of the target population reoffending analysis and early warning method provided by the present invention, such as Figure 3 As shown, the method includes:

[0139] Step: 310: Acquire relevant management data of the target personnel from the pre-integrated relevant data system;

[0140] Step 320: Process and analyze the relevant management data according to a pre-stored data processing and analysis tool to obtain a data analysis report, and add the data analysis report to a sample database; wherein the sample database includes a large amount of target personnel sample data;

[0141] Step 330: Input the data analysis report into a pre-built behavior pattern and psychological tendency prediction model to obtain an early warning signal, and issue an early warning according to the early warning signal; wherein the behavior pattern and psychological tendency prediction model is obtained by training the sample database after feature extraction based on a pre-selected machine learning algorithm;

[0142] Step 340: Input the warning signal into a pre-built intelligent decision support model to obtain the risk level of the target person and the corresponding management measures; wherein the intelligent decision support model is trained based on a pre-selected machine learning algorithm using the sample database after refined classification.

[0143] According to a reoffending analysis and early warning method for a target population provided by the present invention, the behavior pattern and psychological tendency prediction model is obtained by training the sample database after feature extraction based on a pre-selected machine learning algorithm, specifically including:

[0144] Performing data preprocessing on the target person sample data in the sample database; wherein the target person sample data at least includes personal information, criminal history, and reform performance of the target population; the data preprocessing at least includes data cleaning and data coding;

[0145] Dividing the sample database after the data preprocessing into a training set, a validation set and a test set according to a preset ratio;

[0146] Construct multiple base learners based on pre-selected machine learning algorithms;

[0147] Based on the base learner, the training set is used for training, a preset feature extraction algorithm is used to extract basic features from the training set, and a pre-selected deep learning model is used to extract high-level features from the training set, the basic features and the high-level features are fused, and the fusion result is fitted by the base learner to obtain a training result;

[0148] Integrate the plurality of training results using a preset integration algorithm, and perform parameter tuning to obtain a target prediction model;

[0149] The target prediction model is evaluated using the validation set and the test set, and the target prediction model is iteratively optimized according to the evaluation results until preset conditions are met, and a behavior pattern and psychological tendency prediction model is output.

[0150] According to a reoffending analysis and early warning method for a target population provided by the present invention, the intelligent decision support model includes a risk prediction sub-model and a management strategy sub-model; the intelligent decision support model is obtained by training the sample database after refined classification based on a pre-selected machine learning algorithm, specifically including:

[0151] According to the target person sample data in the sample database, the corresponding target person samples are finely classified into different risk levels to obtain classification results;

[0152] Building a basic risk prediction model based on a pre-selected first machine learning algorithm, training the basic risk prediction model using the sample database and the classification results, fitting the classification results, and obtaining a risk prediction sub-model;

[0153] Obtaining a marking result for the management measures corresponding to the risk level marking in the sample database;

[0154] A basic management strategy model is constructed based on a pre-selected second machine learning algorithm, and training is performed based on the basic management strategy model using the sample database and the marking results, and the marking results are fitted to obtain a management strategy sub-model.

[0155] According to a reoffending analysis and early warning method for a target population provided by the present invention, the corresponding target person samples are finely classified according to the target person sample data in the sample database, and are divided into different risk levels to obtain classification results, which specifically includes:

[0156] According to the target personnel sample data, the target personnel sample is graded by using a K-means or hierarchical clustering algorithm to obtain a first graded result sample;

[0157] The first division result sample is further classified using a pre-selected classification model according to the target person sample data to obtain a classification result.

[0158] According to a reoffending analysis and early warning method for a target population provided by the present invention, the acquisition of relevant management data of the target person from a pre-integrated relevant data system specifically includes:

[0159] Extracting key information of the target person from the relevant data system;

[0160] Extracting unstructured data from the key information, and performing text analysis on the unstructured data using a pre-selected natural language processing algorithm to obtain analysis results;

[0161] Relevant management data of the target personnel is obtained according to the analysis results and the structured data in the key information.

[0162] According to a reoffending analysis and early warning method for a target population provided by the present invention, the relevant management data of the target person is obtained according to the analysis result and the structured data in the key information, and then the method further includes:

[0163] Perform data cleaning and verification on the relevant management data of the target personnel.

[0164] The reoffending analysis and early warning method for target population provided by the present invention obtains relevant management data of target personnel from a pre-integrated relevant data system; processes and analyzes the relevant management data according to a pre-stored data processing and analysis tool to obtain a data analysis report, and adds the data analysis report to a sample database; wherein the sample database includes a large amount of sample data of target personnel; inputs the data analysis report into a pre-constructed behavior pattern and psychological tendency prediction model to obtain an early warning signal, and issues an early warning based on the early warning signal; wherein the behavior pattern and psychological tendency prediction model is trained based on a pre-selected machine learning algorithm using the sample database after feature extraction; inputs the early warning signal into a pre-constructed intelligent decision support model to obtain the risk level of the target personnel and the corresponding management measures; wherein the intelligent decision support model is trained based on a pre-selected machine learning algorithm using the sample database after refined classification. The present invention acquires complete and accurate relevant management data of target personnel by integrating relevant data systems, processes and analyzes the data efficiently through data processing and analysis tools, and realizes accurate analysis of personnel who have made corrections through pre-trained behavior patterns and psychological tendency prediction models and intelligent decision support models, thereby realizing a reoffending analysis and early warning for the target population that does not require human intervention, is highly efficient, and has universality, which helps prevent reoffending by personnel who have made corrections.

[0165] Figure 4 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 4As shown, the electronic device may include: a processor (processor) 410, a communication interface (Communications Interface) 420, a memory (memory) 430 and a communication bus 440, wherein the processor 410, the communication interface 420, and the memory 430 communicate with each other through the communication bus 440. The processor 410 can call the logic instructions in the memory 430 to execute a reoffending analysis and early warning method for a target population, the method comprising: obtaining relevant management data of the target person from a pre-integrated relevant data system; processing and analyzing the relevant management data according to a pre-stored data processing and analysis tool to obtain a data analysis report, and adding the data analysis report to a sample database; wherein the sample database includes a large amount of sample data of the target person; inputting the data analysis report into a pre-constructed behavior pattern and psychological tendency prediction model to obtain an early warning signal, and issuing an early warning based on the early warning signal; wherein the behavior pattern and psychological tendency prediction model is trained based on a pre-selected machine learning algorithm using the sample database after feature extraction; inputting the early warning signal into a pre-constructed intelligent decision support model to obtain the risk level of the target person and the corresponding management measures; wherein the intelligent decision support model is trained based on a pre-selected machine learning algorithm using the sample database after refined classification.

[0166] In addition, the logic instructions in the above-mentioned memory 430 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0167] On the other hand, the present invention also provides a computer program product, which includes a computer program, which can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the re-offending analysis and early warning method for the target population provided by the above-mentioned methods, and the method includes: obtaining relevant management data of the target person from a pre-integrated relevant data system; processing and analyzing the relevant management data according to a pre-stored data processing and analysis tool to obtain a data analysis report, and adding the data analysis report to a sample database; wherein the sample database includes a large amount of target person sample data; inputting the data analysis report into a pre-constructed behavior pattern and psychological tendency prediction model to obtain an early warning signal, and issuing an early warning based on the early warning signal; wherein the behavior pattern and psychological tendency prediction model is trained based on a pre-selected machine learning algorithm using the sample database after feature extraction; inputting the early warning signal into a pre-constructed intelligent decision support model to obtain the risk level of the target person and the corresponding management measures; wherein the intelligent decision support model is trained based on a pre-selected machine learning algorithm using the sample database after refined classification.

[0168] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the reoffending analysis and early warning method for a target population provided by the above-mentioned methods, the method comprising: obtaining relevant management data of a target person from a pre-integrated relevant data system; processing and analyzing the relevant management data according to a pre-stored data processing and analysis tool to obtain a data analysis report, and adding the data analysis report to a sample database; wherein the sample database includes a large amount of sample data of target persons; inputting the data analysis report into a pre-constructed behavior pattern and psychological tendency prediction model to obtain an early warning signal, and issuing an early warning based on the early warning signal; wherein the behavior pattern and psychological tendency prediction model is trained based on a pre-selected machine learning algorithm using the sample database after feature extraction; inputting the early warning signal into a pre-constructed intelligent decision support model to obtain the risk level of the target person and the corresponding management measures; wherein the intelligent decision support model is trained based on a pre-selected machine learning algorithm using the sample database after refined classification.

[0169] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0170] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0171] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A recidivism analysis and early warning system for target populations, characterized in that: include: Data collection module, used to obtain relevant management data of target personnel from pre-integrated relevant data systems; A data analysis module, used to process and analyze the relevant management data according to a pre-stored data processing and analysis tool, obtain a data analysis report, and add the data analysis report to a sample database; wherein the sample database includes a large amount of target personnel sample data; An early warning module is used to input the data analysis report into a pre-built behavior pattern and psychological tendency prediction model to obtain an early warning signal, and issue an early warning according to the early warning signal; wherein the behavior pattern and psychological tendency prediction model is obtained by training the sample database after feature extraction based on a pre-selected machine learning algorithm; A control module is used to input the warning signal into a pre-built intelligent decision support model to obtain the risk level of the target person and the corresponding management measures; wherein the intelligent decision support model is trained based on a pre-selected machine learning algorithm using the sample database after refined classification.

2. The target population-oriented reoffending analysis and early warning system according to claim 1 is characterized in that: The method of obtaining a behavior pattern and psychological tendency prediction model by training the sample database after feature extraction based on the pre-selected machine learning algorithm specifically includes: Performing data preprocessing on the target person sample data in the sample database; wherein the target person sample data at least includes personal information, criminal history, and reform performance of the target population; the data preprocessing at least includes data cleaning and data coding; Dividing the sample database after the data preprocessing into a training set, a validation set and a test set according to a preset ratio; Construct multiple base learners based on pre-selected machine learning algorithms; Based on the base learner, the training set is used for training, a preset feature extraction algorithm is used to extract basic features from the training set, and a pre-selected deep learning model is used to extract high-level features from the training set, the basic features and the high-level features are fused, and the fusion result is fitted by the base learner to obtain a training result; Integrate the plurality of training results using a preset integration algorithm, and perform parameter tuning to obtain a target prediction model; The target prediction model is evaluated using the validation set and the test set, and the target prediction model is iteratively optimized according to the evaluation results until preset conditions are met, and a behavior pattern and psychological tendency prediction model is output.

3. The target population-oriented reoffending analysis and early warning system according to claim 1 is characterized in that: The intelligent decision support model includes a risk prediction sub-model and a management strategy sub-model; the intelligent decision support model is obtained by training the sample database after fine classification based on a pre-selected machine learning algorithm, and specifically includes: Refined classification of corresponding target personnel samples is performed according to the target personnel sample data in the sample database, and the samples are divided into different risk levels to obtain classification results; Building a basic risk prediction model based on a pre-selected first machine learning algorithm, training the basic risk prediction model using the sample database and the classification results, fitting the classification results, and obtaining a risk prediction sub-model; Obtaining a marking result for the management measures corresponding to the risk level marking in the sample database; A basic management strategy model is constructed based on a pre-selected second machine learning algorithm, and training is performed based on the basic management strategy model using the sample database and the marking results, and the marking results are fitted to obtain a management strategy sub-model.

4. The target population-oriented reoffending analysis and early warning system according to claim 3 is characterized in that: The method of finely classifying the corresponding target personnel samples according to the target personnel sample data in the sample database, dividing them into different risk levels, and obtaining classification results specifically includes: According to the target personnel sample data, the target personnel sample is graded by using a K-means or hierarchical clustering algorithm to obtain a first graded result sample; The first division result sample is further classified using a pre-selected classification model according to the target person sample data to obtain a classification result.

5. The target population-oriented reoffending analysis and early warning system according to claim 1 is characterized in that: The step of obtaining relevant management data of the target personnel from the pre-integrated relevant data system specifically includes: Extracting key information of the target person from the relevant data system; Extracting unstructured data from the key information, and performing text analysis on the unstructured data using a pre-selected natural language processing algorithm to obtain analysis results; Relevant management data of the target personnel is obtained according to the analysis results and the structured data in the key information.

6. The target population-oriented reoffending analysis and early warning system according to claim 1 is characterized in that: The step of obtaining relevant management data of the target personnel according to the analysis results and the structured data in the key information further includes: Perform data cleaning and verification on the relevant management data of the target personnel.

7. A recidivism analysis and early warning method for a target population, characterized in that: include: Obtain relevant management data of target personnel from pre-integrated relevant data systems; Processing and analyzing the relevant management data according to a pre-stored data processing and analysis tool to obtain a data analysis report, and adding the data analysis report to a sample database; wherein the sample database includes a large amount of target personnel sample data; Input the data analysis report into a pre-built behavior pattern and psychological tendency prediction model to obtain an early warning signal, and issue an early warning according to the early warning signal; wherein the behavior pattern and psychological tendency prediction model is obtained by training the sample database after feature extraction based on a pre-selected machine learning algorithm; The warning signal is input into a pre-built intelligent decision support model to obtain the risk level of the target person and the corresponding management measures; wherein the intelligent decision support model is trained based on a pre-selected machine learning algorithm using the sample database after refined classification.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, it implements the reoffending analysis and early warning method for the target population as described in any one of claims 1 to 6.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the reoffending analysis and early warning method for a target population as described in any one of claims 1 to 6 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the reoffending analysis and early warning method for a target population as described in any one of claims 1 to 6 is implemented.

Images