| As the socialism with Chinese characteristics entering a new era,the increasing social change has brought great vitality to China’s social and economic development.With the high-quality opening-up and the vigorous implementation of the strategy of talent introduction,domestic and international exchanges are increasingly frequent,and the number of entry-exit personnel is growing.This makes it more difficult to evaluate the entry-exit behavior abnormality of entry-exit personnel,which increases the potential threat of abnormal behavior inspection of entry-exit personnel.The key to improve the efficiency of risk prevention and control work for entry and exit personnel is to accurately identify and timely warn the abnormal entry and exit behaviors,such as engaging in work that is not in accordance with their identity and age,or showing abnormal situations that are different from most normal behaviors.However,the large number of inbound and outbound people makes it almost impossible for traditional working methods,such as simple information query and statistics,to complete effective anomaly recognition.With the improvement of entry and exit management information infrastructure such as Golden Shield Project and Mesa System,a large number of entry and exit personnel information and data can be collected and stored completely.How to get valuable information from these data,and then quickly and effectively identify the inbound and outbound behavior anomalies,is a very meaningful research.This paper mainly studies the key technologies of recognizing inbound and outbound personnel’s behavior anomalies based on massive data analysis.The main research contents are as follows:Data of entry and exit behavior records of entry and exit personnel and analysis of their characteristics.The identity credentials of entry and exit personnel,each entry and exit will form entry and exit record data for easy synchronization.This data can be divided into personnel identity information data and personnel entry and exit record information data.The analysis of a large amount of data on identity information of entry and exit personnel and information on entry and exit records itself,such as the structure of entry and exit data,entry and exit frequency,stay time,and distribution of entry and exit causes,is an important basis for classifying behavior patterns.It can also greatly reduce the workload of entry and exit behavior anomaly detection of entry and exit personnel and improve the efficiency of detection.The pattern of entry and exit behavior of entry and exit personnel is unknown,so it is necessary to cluster their entry and exit record data,which is a typical time series data.In order to improve the efficiency of clustering analysis,this paper presents a frequency-based clustering analysis method,which gives different clustering algorithms for low-frequency,medium-frequency and high-frequency exit and entry data,respectively,in order to obtain an effective classification of entry and exit behavior patterns.3)Classification of entry-exit behavior patterns based on identity information of decision tree.This paper establishes a decision tree classifier based on the association between identity information features and inbound and outbound behavior patterns.When you enter the identity information of an offshore person,you will be given the most likely patterns of behavior for that person.If the corresponding behavioral patterns of their entry and exit records are not among these behavioral patterns,the overseas person can be classified as a person with abnormal behavior.The key technologies for identifying abnormal behavior of inbound and outbound personnel based on massive data analysis are applied to the risk assessment practice of inbound and outbound personnel.The results show that the method improves the accuracy of identifying abnormal behavior personnel,effectively reduces the range of people with risks that need special attention,and significantly reduces the workload and time cost. |
PDF Full Text Request