CN115293477A

CN115293477A - Social security early warning system, method, equipment and storage medium based on big data

Info

Publication number: CN115293477A
Application number: CN202210316660.5A
Authority: CN
Inventors: 张博; 郭静; 杨云祥; 王金龙; 李利民; 崔遥; 陈思源; 王谋业; 张帅; 刘晨申; 王大鹏; 张红; 王文欣; 魏曦; 江逸楠
Original assignee: CETC Ocean Information Co Ltd
Current assignee: CETC Ocean Information Co Ltd
Priority date: 2022-03-29
Filing date: 2022-03-29
Publication date: 2022-11-04

Abstract

The application discloses big data-based social security early warning system, method, equipment and storage medium, and the system comprises: the system comprises an infrastructure layer, a platform software layer, a data information layer and an application software layer which are arranged from bottom to top in sequence; the infrastructure layer is used for providing basic software and hardware support based on a cloud platform and a big data platform; the platform software layer is used for providing various component supports based on a cloud platform and a big data platform, accessing various original data and providing basic data for business application; the data information layer is used for acquiring various types of original data, forming a complete track data chain of a target object and constructing service topics corresponding to different scenes; the application software layer is used for providing various service applications oriented to different scenes. By utilizing technologies such as big data, knowledge graph, visual relation mining and the like, the problem of fusion and association mining of multi-dimensional perception information of social security equipment is solved.

Description

Social security early warning system, method, equipment and storage medium based on big data

Technical Field

The present disclosure relates generally to the field of security early warning technologies, and in particular, to a social security early warning system, method, device, and storage medium based on big data.

Background

Social public security has evolved into traditional and non-traditional security threats, and social uncertainty is further increased, which requires that security systems be based on event handling and risk management, and gradually turn to risk management. Compared with the prior security technology, the method is more used as an important means for rechecking afterwards, realizes the intelligent early warning and prevention of the social security risk in advance, and has practical significance for preventing and controlling the social security risk in advance.

Disclosure of Invention

In view of the above-mentioned drawbacks and deficiencies of the prior art, it is desirable to provide a social security early warning system, method, electronic device and storage medium based on big data.

In a first aspect, a big data-based social security early warning system is provided, which includes: the system comprises an infrastructure layer, a platform software layer, a data information layer and an application software layer which are arranged from bottom to top in sequence; wherein the content of the first and second substances,

the infrastructure layer is used for providing basic software and hardware support based on a cloud platform and a big data platform;

the platform software layer is used for providing various component supports based on a cloud platform and a big data platform, accessing various original data and providing basic data for service application;

the data information layer is used for acquiring various kinds of original data, forming a complete track data chain of a target object and constructing service topics corresponding to different scenes;

the application software layer is used for providing various service applications oriented to different scenes.

In a second aspect, a big data-based social security early warning method is provided, where the big data-based social security early warning system includes:

performing deep association fusion on various sensing data acquired by the social security sensing equipment by adopting a big data technology to form fused basic data;

performing deep fusion and correlation analysis on the fused basic data to form a service subject library;

and providing a multi-dimensional stereo early warning and perception model facing different scenes based on an application program algorithm and in combination with a plurality of early warning technologies according to the service subject library.

In a third aspect, an electronic device is provided, the device comprising:

one or more processors;

a memory for storing one or more programs,

the one or more programs, when executed by the one or more processors, cause the one or more processors to perform a big data based social security pre-warning method as provided by embodiments of the present application.

In a fourth aspect, a computer-readable storage medium storing a computer program, the program being executed by a processor to perform a big data based social security early warning method as provided in various embodiments of the present application is provided.

According to the technical scheme provided by the embodiment of the application, a big data technology, a knowledge map and visual relation mining are used as supports, data collected by social security sensing equipment are used as bases, multidimensional track normalization and multidimensional relation mining analysis is carried out, and the analysis and the early warning analysis are carried out on the social security sensing data with the traditional single dimension.

Drawings

Other features, objects and advantages of the present application will become more apparent upon reading of the following detailed description of non-limiting embodiments thereof, made with reference to the accompanying drawings in which:

fig. 1 is an exemplary overall framework diagram of a big data-based social security early warning system provided by an embodiment of the present application;

FIG. 2 is a functional framework diagram of the application software layers of FIG. 1;

fig. 3 is a technical architecture diagram of a big data based social security early warning system according to an embodiment of the present disclosure;

fig. 4 is a working schematic diagram of a big data-based social security early warning system according to an embodiment of the present disclosure;

fig. 5 is a schematic structural diagram of an electronic device according to an embodiment of the present application.

Detailed Description

The present application will be described in further detail with reference to the following drawings and examples. It is to be understood that the specific embodiments described herein are merely illustrative of the relevant invention and not restrictive of the invention. It should be noted that, for convenience of description, only the portions related to the present invention are shown in the drawings.

It should be noted that, in the present application, the embodiments and features of the embodiments may be combined with each other without conflict. The present application will be described in detail below with reference to the embodiments with reference to the attached drawings.

The existing various security technologies are more taken as means for after-the-fact rechecking, and the risk of various social illegal behaviors and the social security trend cannot be seamlessly monitored by a single security means. The method has the advantages that a big data technology is urgently needed, various security and protection means are combined, deep fusion association is carried out, various illegal behavior characteristics are combined to construct an intelligent early warning model, advance intelligent early warning is carried out on various illegal behavior risks, and the social security risk prevention and control capability is comprehensively improved.

Compared with the prior security and protection technology, the method is more used as an important means for after-the-fact review, and in recent years, target detection, face recognition, target tracking, target classification, target retrieval and behavior analysis in videos are realized through artificial intelligent frontier technologies such as deep learning. Taking the human body of the monitoring scene as an example, the physiological attributes of the pedestrian are identified. By analyzing the body structure of the pedestrian, various physiological characteristics of the person in the video, such as gender, age group, posture and the like, can be accurately identified. In the aspect of pedestrian and vehicle identification, the pedestrian detection algorithm based on deep learning can accurately find out the position of a pedestrian under various shielding conditions, can further analyze the posture and the action of the pedestrian, and can be applied to traffic monitoring, auxiliary driving, unmanned driving and the like. The method can detect vehicles at different angles in a driving scene, a traffic monitoring scene and a checkpoint scene, and simultaneously give physical characteristics such as license plate numbers, automobile brands, models and colors. In addition, crowd analysis is achieved. In high-density public places such as subways and squares, the number and the density of people are estimated, and various abnormal phenomena such as crowd density, abnormal gathering, detention, retrograde motion, confusion and the like are detected. However, only through the technical means of video monitoring and analysis, a large amount of high-definition video monitoring needs to be covered, and the effect is greatly reduced or even directly disabled when the blind area or the cross-region association is performed.

In order to solve the above technical problems, please refer to fig. 1, which illustrates that a social security early warning system based on big data is provided according to an embodiment of the present application, and for various sensing data acquired by a conventional social security sensing device, a big data technology is used to perform deep association fusion, and then an intelligent early warning model is constructed by combining various illegal behavior means, characteristics, backgrounds and other factors, so that the prior social security risk intelligent early warning and prevention are realized, the prior social security risk prevention and control has practical significance, and the value of the conventional security means combining big data and an experience model is highlighted.

As shown in fig. 1-2, in this embodiment, the social security early warning system based on big data provided by the present invention includes: the system comprises an infrastructure layer, a platform software layer, a data information layer and an application software layer which are arranged from bottom to top in sequence; wherein the content of the first and second substances,

the platform software layer is used for providing various component supports based on a cloud platform and a big data platform, accessing various original data and providing basic data for business application;

the data information layer is used for acquiring various types of original data, forming a complete track data chain of a target object and constructing service topics corresponding to different scenes;

In one embodiment, the infrastructure layer includes at least one of: the system comprises a virtual machine, a big data assembly, a cloud service module, a network module and a storage module.

Specifically, as shown in fig. 1, an infrastructure as a service (IaaS) layer is mainly deployed based on basic software and hardware supports provided by a cloud platform and a big data platform, and includes resources such as a virtual machine, a big data component, a cloud service, a network, and storage.

In one embodiment, the platform software layer includes at least one of: the system comprises a population data interface, a map data interface, a vehicle card port data interface, an electronic fence data interface, a WiFi data interface and a face recognition data interface.

Specifically, as shown in fig. 1, a platform software layer (PaaS layer) is mainly based on various component supports provided by a cloud platform and a big data platform, and is used for docking various data resources including a population data interface, a map data interface, a vehicle access data interface, an electronic fence data interface, a WiFi data interface, a face recognition data interface and the like, so that unified access, fusion and normalization management and the like of various basic data and track data are realized, and business applications are supported.

In one embodiment, the data information layer comprises a service subject library and a superior data interface, wherein the service subject library is used for constructing the incidence relation of various types of data, the incidence relation of various types of tracks and service knowledge; and the superior data interface is used for being connected with various superior track data interfaces, supplementing various local track data and forming a complete track data chain of the target object.

Specifically, as shown in fig. 1, the data information layer (DaaS layer) includes a service item library and an upper data interface. The business special topic library module constructs the incidence relation of various data, the incidence relation of various tracks, business knowledge and the like based on business rules and visual angles. And (3) supplementing various local track data through butting various upper-level track data interfaces to form a complete track data chain of the target object.

In one embodiment, the business topic library includes at least one of: the system comprises a human subject database, a multi-dimensional track database, a business relation database, an association track database and a business knowledge database;

the superior data interface includes at least one of: personnel data interface, electricity enclose data interface, wiFi data interface, people's face data interface, bayonet socket data interface.

In one embodiment, the application software layer includes at least one of: the system comprises a figure multidimensional analysis module, a social relationship network evolution module, an individual early warning module, a group early warning module, a social security risk evolution module and an early warning model management module.

Specifically, as shown in fig. 1-2, the application software layer (SaaS layer) mainly includes the above six modules, and the six modules can implement the intelligent pre-warning and prevention of social security risks in advance. The character multidimensional analysis module abstracts the multidimensional relation of characters through various social relations of the characters, conducts multidimensional correlation mining through data information acquired by the characters on different social security sensing devices in combination with the multidimensional relation of the characters, provides character multidimensional relation search, character multidimensional track search, dangerous entity portrait display and the like, and provides support for classification prevention and control and the like.

The social relation network evolution module tracks, excavates and analyzes the group relation network evolution and the organization relation network evolution, excavates and expands the relation of each dimension of the group and the organization to form an evolution model of the group relation network and the organization relation network.

An individual early warning module: the method comprises a first individual intelligent early warning, a second individual intelligent early warning, a third individual intelligent early warning, a fourth individual intelligent early warning and a fifth individual intelligent early warning; and combining the multidimensional track information, background information, key illegal behavior factors and the like of the individual, and combining the type of the individual behavior case to carry out intelligent early warning.

A group early warning module: the method comprises a first group early warning, a second group early warning, a third group early warning and a fourth group early warning. After group information is formed based on the individual illegal behavior early warning information and by combining multidimensional relation mining among individuals, various group early warning models are established for the group illegal behaviors, and intelligent early warning for the group illegal behaviors is achieved.

The social security risk evolution module comprises large-scale activity security risk, public health security risk, financial security risk and key area security risk. The social security risk evolution model is constructed by combing large-scale activities, public health events, financial asset conditions and key area conditions in the jurisdiction, combining people, organizations, group events and the like associated with related areas and combining security risk factors with different dimensions, so that intelligent early warning of social security risks is realized.

In one embodiment, the person multidimensional analysis module comprises a multidimensional relation searching unit, a multidimensional track searching unit and a dangerous entity portrait unit; wherein the content of the first and second substances,

the multi-dimensional relation searching unit is used for abstracting the multi-dimensional relation of the character through various social relations of the character;

the multi-dimensional track searching unit is used for performing multi-dimensional association mining according to data information of people collected on different social security sensing devices and the multi-dimensional relation of the people abstracted by the multi-dimensional relation searching unit, and providing multi-dimensional track searching of the people;

the dangerous entity portrait unit is used for extracting the characteristics of dangerous personnel and forming portrait of the dangerous personnel.

Specifically, as shown in fig. 1-2, the character multidimensional analysis module comprises a multidimensional relation search unit, a multidimensional track search unit and a dangerous entity representation unit. Data information collected on different social security sensing devices based on characters comprises face monitoring collection information, vehicle access collection information, electronic fence collection information, hotel internet bar collection information and the like, data extraction is carried out through an ETL tool, identity normalization fusion processing is carried out based on the identity card number of the characters, the associated license plate number, the mobile phone IMEI/IMSI and the like, different track information of the characters is fused and associated to the characters, association mining is carried out by combining the multidimensional relation of the characters, indexes are built by taking the characters as cores, and a data searching function is provided.

Wherein the multidimensional relation searching unit: through a big data distributed search engine technology, in addition to traditional searching of relationships such as relatives, classmates, colleagues and villages, deep mining is performed on figure information such as QQ numbers, mobile phone numbers and micro signals, and more association relationships are obtained.

A multi-dimensional track search unit: the journey information extracted from QQ, weChat and telephone is analyzed by NLP technology, and is integrated with the actual track of people, network ticket buying, camera capturing and other track information, so that the movement of important attention people is sensed in advance.

A dangerous entity image unit: constructing dangerous entity portraits, and forming a holographic file of dangerous persons, wherein the holographic file comprises basic information (age, height, sex, weight and the like) of persons, portrait portraits, vehicle information, student status information, work information, marital information, residence information, medical information, case/event information, virtual space information (all the associated virtual space information can be used for checking corresponding call records, transfer records, chat records, purchase records and the like), a relationship graph corresponding to the physical space and the virtual space of the persons and the analysis of the activity travel tracks of the persons within thirty days.

The method comprises the steps that data extraction is carried out through an ETL tool on the basis of face monitoring information, vehicle access port acquisition information, video monitoring, electronic fence acquisition information, hotel internet bar acquisition information and the like acquired by social security sensing equipment; and then, based on the identity card number of the figure, the associated license plate number, the IMEI/IMSI of the mobile phone, the real-name virtual identity ID and the like, the figure identity normalization fusion processing is carried out, so that different track information of the figure is fused and associated to a single figure, a complete track chain of a figure target is formed, the organic fusion of multi-dimensional data is realized, and high-quality basic data is provided for high-end application.

In one embodiment, the multi-dimensional relationship of the character includes at least one of: the relationship between persons, the relationship between persons and places, the relationship between persons and vehicles, the relationship between persons and terminal devices, the relationship between persons and cases, the relationship between persons and information, and the relationship between persons and organizations.

Specifically, the character multidimensional analysis module: the system comprises a multi-dimensional relation searching unit, a multi-dimensional track searching unit and a dangerous entity portrait unit. By the aid of various social relations of people, relations between people and places, relations between people and vehicles, relations between people and terminal equipment, relations between people and cases, relations between people and information, relations between people and organizations and the like are abstracted, data information collected by people on different social security sensing devices comprises face monitoring collected information, vehicle bayonet collected information, electronic fence collected information, hotel internet bar collected information and the like, multidimensional association mining is carried out by combining the multidimensional relations of people, people multidimensional relation searching, people multidimensional track searching, dangerous entity portrait displaying and the like are provided, and support is provided for hierarchical classification prevention and control, arrangement pursuit and escape and the like.

In one embodiment, the social relationship network evolution module comprises at least one of: a group relation network evolution unit and an organization relation network evolution unit.

Specifically, as shown in fig. 1-2, the social relationship network evolution model module includes: a group relation network evolution unit and an organization relation network evolution unit. The method mainly aims at tracking and mining analysis of group relationship network evolution and organization relationship network evolution conditions, and performs periodic relationship path calculation, relationship strength calculation, relationship updating calculation and the like according to various data sensed and collected by various social security equipment on the basis of relationships between people and relationships between people and organizations in a jurisdiction, so that the group relationship network evolution conditions and the organization relationship network evolution conditions are completely drawn, and support is provided for prevention and control of organization risks. The functional roles of the population relationship network evolution unit and the organization relationship network evolution unit are exemplified by the following examples:

(1) A population relationship network evolution unit: group relationships are exemplified by gangs: 11/1/2020, (each letter represents a person, replaced by a name in the relationship graph) ABC, CDE, and ABF act together as a group to form a network graph, and 12/1/DEF act together. All the people are temporarily gathered together, only the foreign number and the name are known but not known, and the people are temporarily gathered together only when moving. Without explicit context, it may be that AB is often associated with brother, and before often associated with C, but some time C is caught in action with others, and F is found, and the connections between people may be tight, loose, some often associated with action, and some temporarily grouped together. Therefore, the relation of the people needs to be found out, which is troublesome in reality and increases the difficulty for detecting the case. Therefore, the doubtful objects are deeply mined, analyzed and judged through a group relation network evolution module to form a group evolution process.

The method and the device realize the mining of the group relation network, and form the logic division of the group by mining analysis of individual multidimensional tracks, fusion analysis of multidimensional relations and combination of the strength and weakness of the relations, thereby combining with the formation of different risk group targets.

(2) An organizational relationship network evolution unit: the organization relationship evolution takes the travel industry as an example: 2010. in the year, life a is not ideal, and 10 people such as DEFG are collected, and frequently go to a hotel and a tour guide, and the hotel and the tour guide are required to go to the store for consumption. In 2013, A recruits his brother H to start multiple hotels and disturb other hotels to avoid normal operation, thus monopolizing the local tourism industry. Basically, a hotel organization is formed by ABCH, DEFGXYZ and the like. And performing deep relation mining analysis and judgment on the social relation network of the hotel organization to form an evolution process and members of the hotel organization.

The one-dimensional element abnormity early warning with wider application at present is that monitoring and risk identification are carried out only aiming at certain social safety related data, and when a certain monitoring data has larger deviation relative to the historical data or standard data, alarm information is generated. The one-dimensional element abnormity early warning structure is simple and easy to realize, but the defect is that only simple abnormity study and judgment can be carried out, the study and judgment threshold value is difficult to set, the checking work is heavy due to the fact that the threshold value is too low and the false alarm rate is too high, the risk of false alarm exists due to too low threshold value, and accurate abnormity alarm information cannot be generated for complex social security events.

The method adopts a multi-dimensional element abnormity early warning mode, establishes individual and group early warning models by combining individual and group characteristics according to individual illegal behavior characteristics, activity tracks and group association information, realizes individual early warning and group (group) early warning, and can generate accurate abnormity warning information for complex social security events.

Specifically, as shown in fig. 1-2, the individual early warning module mainly builds intelligent early warning of various illegal behaviors based on various data sensed and collected by various social security devices and by combining multidimensional relation mining of individuals, context information association of individuals, experience knowledge, key factors and the like. The individual warning units of the individual warning modules are illustrated by the following examples:

(1) The first individual early warning unit: through deep learning of the theft history case, the extracted suspect object features, such as the suspect object is a non-industrial/unemployed person, a farmer's house, an external population, low cultural degree and the like. Through deep learning of historical cases, extracted abnormal behavior characteristics, such as that theft often occurs in districts lacking security, villages in cities, streets or areas with dense pedestrian flows, and the like, suspects that objects often wander in high-incidence areas of such cases; the camera shoots that the suspected object is concealed and has suspicious behaviors of eastern and western views; wandering and wandering in the midnight with frequent cases; purchase of tools such as unlocking/control tools; resale of suspected stolen valuables in second-hand markets, etc.

(2) Second individual early warning unit: through deep learning of the robbery history case, the extracted suspect object characteristics, such as the suspect object is a no-business/no-business person, the farmer family, the external population, the cultural degree is low, and the like. Through deep learning of a robbery history case, extracted abnormal behavior characteristics, such as the fact that the robbery often occurs in Internet bars and surroundings, near entertainment places, urban and rural combination areas, rental houses, factory concentration areas, remote streets and the like, and suspected objects often wander in such case high-incidence areas; the camera shoots that suspected objects may abnormally follow some independent people in the middle of the night; wandering and wandering in the midnight with frequent cases; purchase of tools such as unlocking/control tools; resale valuable items suspected of being stolen in the second-hand market, etc.

(3) A third individual pre-warning unit: and extracting the characteristics of the suspicious object through deep learning of the historical case, wherein the suspicious object is a special crowd, a military referee and the like. Through deep learning of historical cases, extracted abnormal features such as lingering or loitering in key areas and sensitive parts, recent contact with dangerous explosives, recent purchase of controlled items, and the like. Through deep learning of the historical cases, the extracted abnormal speech characteristics of the virtual space are analyzed through NLP (Natural Language Processing) semantics, and the fact that the suspected object has a Language used for speaking related to the alarm for many times in the speech of the virtual space is found.

(4) A fourth body early warning unit: through deep learning of the historical case, the extracted suspicious object features, such as the fact that the suspicious object is a dishonest record, etc. Through deep learning of fraud history cases, abnormal features of fund flow are extracted, such as abnormal fund inflow, unrelated people who have fund flow through association analysis, and the like. Through deep learning of the historical cases, extracted abnormal speech characteristics of the virtual space are found through NLP semantic analysis, and keywords such as transfer, money borrowing, fund, investment and the like which are related to the fund appear for many times in chatting between the suspected object and the fund-going person.

(5) A fifth body alarm unit: through deep learning of historical cases, the extracted suspected object features, such as the suspected object is an emotional instability or a drug abuse. Through deep learning of historical cases, the extracted background features of the suspicious objects have similar information about family conditions, disputes, cases and the like or mediation information and the like. Through deep learning of historical cases, the extracted relevant information of the suspect target spouse has similar family emotion dispute report information, suspected injured medical record records and the like. Through deep learning of a historical case, the extracted abnormal speech characteristics of the virtual space are discovered through NLP semantic analysis, and the words and expressions related to physical and mental injuries in the family emotion such as beating me, kicking me, fanning me, insulting and the like of the suspected object spouse which occur in the virtual space for many times are discovered.

Specifically, as shown in fig. 1-2, the group early warning module mainly combines a social relationship network evolution model and an individual early warning module, performs deep association mining on a calculation result of a group relationship network and individual early warning information, mines a group associated with an individual by using the individual as a core, and performs mining analysis on the group by combining the strength of the individual relationship and the relationship evolution condition. The following examples illustrate the individual warning units of the group warning module:

(1) A first population early warning unit: through machine learning, it is known that the period of time for which a theft case occurs somewhere is mostly 0 to 4. Through camera data, collision theft group early warning label model, discover someone or several people, hit one of them label (unusual motion in unconventional time quantum) according to this route of this time of real-time orbit collision, many people hit a plurality of labels, according to the early warning model of backstage, conjecture out the probability of committing a crime again. And the early warning can be output when the stealing cases appear in the evening before a certain place and the stealing groups can still exert the same skills in the next day.

(2) The second group early warning unit: early warning factors are discovered through channels of monitoring markets to discover unknown source frozen products, people reporting and the like, and a plurality of early warning factors are integrated to perform correlation analysis mining and group mining. And (4) supplementing a group list, excavating the back groups and related shops and enterprises, and predicting the occurrence probability and early warning possible back groups by using a collision background model.

(3) A third group early warning unit: and clicking the left early warning list, and checking the details of the early warning factors. And then, a plurality of early warning factors are selected, and a problem shop and personnel behind the problem enterprise are found through multi-dimensional correlation analysis and mining of the known early warning factors and are supplemented to form a group member list. And then a multi-dimensional track, a fund flow and a network speech collision group early warning model are used. When N persons in the group hit the N labels, the probability of crime of the group is deduced according to historical data, and more suspicion objects are early warned through multidimensional correlation analysis.

(4) A fourth group pre-warning unit: the early warning factors of the group are gathered by collecting data of network public opinion data monitoring, reporting by various modes of the group and reporting. And integrating the subject objects related to the early warning factors, and supplementing a group list through three-layer incidence relation mining, fund flow mining and virtual account number mining. And combing the information of the early warning factor and the supplemented information according to the background early warning model, and combing the information of the group. And according to the capital flow direction, a large back group is excavated and early warned, and illegal evidence is collected.

In one embodiment, the social security risk evolution module includes at least one of: a large-scale activity safety risk unit, a public health safety risk unit, a financial safety risk unit and a key area safety risk unit.

Specifically, as shown in fig. 1-2, the social security risk evolution module performs social security comprehensive risk assessment based on individual early warning and group early warning by combining multidimensional fusion track information and multidimensional fusion relationship information, and realizes quantitative perception and auxiliary decision for risk evolution of different dimensions and regions of the social security risk. The social security risk evolution module comprises: large-scale activity safety risk, public health safety risk, financial safety risk, key area safety risk and the like. The method mainly aims at core social security risk points of large-scale activities, public health security, financial security, security of key areas and the like in jurisdictions, combines people, social relations, individual and group illegal behavior information and the like associated with the social security risk points to perform deep fusion and association mining, combines various information acquired by social security sensing equipment in real time to perform real-time risk assessment and calculation, and realizes sensing and prevention and control of the security risk of the large-scale activities, the public health security risk, the financial security risk and the security risk of the key areas. The functional role of the social security risk evolution module is illustrated by the following example:

(1) Large-scale activity safety risk unit: through monitoring network public sentiment, the activity of a certain spontaneous organization is known. Through the analysis to regional population density (through operator data) on the same day, the comparison is more than usual (when flat, weekend, the population density of this period of holiday), know again that nearby subway goes out the number of people of entering and leaving, the number of people of getting on or off the bus is all more than usual, nearby parking area is also more than usual full, through the analysis of many characteristics, compare present security strength and the condition contrast of this region full load again, it lacks extremely greatly to obtain the security strength in this region at present, for preventing that the accident from taking place, at this moment, need increase security strength, dredge the crowd, prevent that the accident from taking place.

(2) Public health safety risk unit: the data are analyzed by monitoring the electronic medical record system of the hospital and the record of asking for medical information on the internet, so that the attributes of the disease such as infectivity, fatality and concealment can be known, and the origin can be obtained by analyzing the track of the disease. The extent of the effect of this disease can be generally understood in conjunction with past public health safety incidents.

(3) Financial security risk unit: by monitoring the behaviors of enterprises (such as recruitment and fund exchange) and finding that an enterprise is suspected to induce public investment, promise high return, change and absorb public deposit, unauthorized and private financial service development of the enterprise, over-payment to persons without capability of returning money and the like, the actions may cause financial security events and need to be investigated and disposed by related departments.

(4) Security risk unit of key area: and (4) aiming at all key areas in the district, defining key area boundary lines by taking the key areas as units. According to the multi-dimensional information of people, vehicles, IDs, code addresses, organizations, groups and the like entering and exiting from the key areas, a key area safety risk monitoring model is built, key area safety risk indexes are comprehensively evaluated, key area safety risk classification management is carried out, and the risk condition of the key areas is quickly warned and classified and controlled.

In one embodiment, the early warning model management module comprises at least one of: the system comprises a label definition management unit, an early warning model training unit and an early warning model application unit.

Specifically, as shown in fig. 1-2, the early warning model management module includes: and the label definition management unit, the early warning model training unit, the early warning model application unit and the like. After different risk early warning labels are defined based on various accessed data, early warning model training and application are carried out, and intelligent early warning of various illegal behavior risks is supported.

The architecture of the big data-based social security early warning system is shown in fig. 3, and the system adopts a mainstream mature technical architecture and comprises a hardware platform, an operating system, a data access processing method, a data storage method, a data calculation method, an interface docking method and an application program. Wherein, the hardware platform comprises X86, X86_64, IA64, loongson, feiteng, shenwei and the like; operating systems include Windows, linux, etc.; the data access processing method comprises ETL, JDBC, ODBC, REST, niFi and the like; the data storage method comprises HDFS, hbase, MPP, elasticisarch, postgreSQL and the like; the data calculation method comprises Spark, flink, mapReduce, storm and the like; the interface docking method comprises JDBC, ODBC, REST, webservice and the like; the front-end technology adopted by the application program comprises Spring, struts, javaScript, CSS3, ajax, freeMarker, HTML5, jquery, node. Js, groovy and the like, and the adopted algorithm model comprises GBDT, C4.5, K-means, CART, SVM and the like. In the system architecture, a big data technology component is adopted in the aspects of data storage and calculation, high-concurrency calculation of mass data is achieved, and mining analysis is carried out on various collected and gathered data by combining a machine learning algorithm and a data mining algorithm.

As shown in fig. 4, the design principle of the big data-based social security early warning system of the present application is that face monitoring, video monitoring, electronic fence, wiFi probe, vehicle access, background information, historical case data, and other more than ten kinds of structured and unstructured data are extracted and managed in the manner of ETL extraction, interface docking, and data management, and distributed storage management is adopted to improve the efficiency of data reading and writing and calculation. After basic data are formed through data standardization treatment, data cleaning, data association and the like, a virtual and real space information association mapping and subject identity technology, a fragmentation behavior data real-time integration activity track mapping technology and a dangerous entity portrait construction and relation map generation technology are innovatively invented, and a subject database, a multi-dimensional track database, an association relation database, an association track database, a business knowledge base and the like are formed. Based on algorithms such as GBDT and K-means, the social relationship network evolution model construction and analysis technology, the individual illegal behavior intelligent early warning technology based on deep learning, the group illegal behavior intelligent early warning technology based on social relationship network evolution, the social security risk evolution simulation and critical early warning technology and the like are invented in combination with business experience, the applications such as character multidimensional analysis, social relationship network evolution analysis, individual early warning, group early warning and social security risk evolution are provided for different application scenes, and meanwhile, the user is supported to construct personalized applications by himself, and the application requirements of actual combat businesses are met.

In a second aspect, as shown in fig. 4, a big data-based social security early warning method is provided, where the big data-based social security early warning system includes:

s10: performing deep association fusion on various sensing data acquired by the social security sensing equipment by adopting a big data technology to form fused basic data;

s20: performing deep fusion and correlation analysis on the fused basic data to form a service subject library;

s30: and providing a multi-dimensional three-dimensional early warning and perception model facing different scenes according to the service subject library, based on an application program algorithm and in combination with a plurality of early warning technologies.

Specifically, in step S10, the various sensing data collected by the social security sensing device include: the system comprises face monitoring, video monitoring, an electronic fence, a WiFi probe, a vehicle access, background information, civil aviation ticket booking, train ticket booking, hotel accommodation, passenger ticket booking, ferry ticket booking and the like. Before deep association fusion is carried out on various sensing data acquired by the social security sensing equipment by adopting a big data technology, operations such as data standardization, data cleaning and the like can be carried out on the various sensing data acquired by the social security sensing equipment.

In one embodiment, step S20 specifically includes: and analyzing the fused basic data by adopting a virtual-real space information association mapping and main body same technology, a fragmentation behavior data real-time integration activity track mapping technology, a dangerous entity portrait construction and relational map generation technology to form a subject database, a multi-dimensional track database, an association relation database, an association track database and a business knowledge database.

Specifically, the deep fusion and the correlation analysis are performed on the fused basic data through a multidimensional social security information deep fusion and correlation analysis technology to form a business theme library, which specifically comprises the following steps:

(1) The virtual-real space information association mapping and the main body are the same technology: at present, almost all social software is registered based on micro-signals, mobile phone numbers and the like, and the micro-signal mobile phone number must be registered based on the identity card number, so that a plurality of social software numbers are associated with people through the two-layer and three-layer association analysis. Through big data, in addition to traditional entity relations such as relatives, classmates, colleagues and villages, figure information such as QQ numbers, mobile phone numbers and micro signals is deeply mined to obtain more association relations. The virtual space information and the entity space information are combined, and the information of the person is mastered in multiple dimensions.

In the prior art and the existing built-up system, the problems that the virtual information and the entity information are not fused and the like still exist. Virtual information is difficult to obtain and is distributed among multiple systems. The entity information of the suspected object is scattered in each system and cannot be uniformly integrated into one system. In order to investigate a suspicion object, a worker needs to spend a lot of time and effort. And face: each system does not necessarily allow login; the system has insufficient authority after logging in, clearly has information and cannot see the information at that time; and the system is too many, so that key information is easy to miss, and the like. The method and the device solve the problem that virtual information and entity information are not fused by adopting the same technology of virtual-real space information association mapping and a main body.

(2) The fragmented behavior data is integrated in the activity track mapping technology in real time: the portrait track + the associated entity track + the virtual space track = the multidimensional track, and the person is combined with the associated item and the virtual account, for example, the vehicle is associated with the person, and the track of the vehicle represents the track of the person, and so on. Various license numbers and virtual account numbers of the personnel are associated with the personnel by combining a multi-dimensional relation search function, and the track of the associated account numbers also represents the track of the personnel, so that the multi-dimensional track search is realized. Travel information extracted from QQ, weChat and telephone is analyzed, and the travel information is integrated together by combining the actual track of people, network ticket buying, camera capturing and other track information, so that the movement of important attention personnel is sensed in advance.

(3) Dangerous entity portrait construction and relationship map generation technology: on the basis of the functions of 1.1 and 1.2, dangerous person characteristics are extracted through a dangerous entity portrait model to form a dangerous person portrait, all ubiquitous space information and ubiquitous space tracks of dangerous persons are integrated, and a dangerous person ubiquitous space holographic file is constructed and comprises person basic information, person portrait, vehicle information, student status information, work information, marital information, residence information, medical information, case/event information and virtual space information (all virtual space information related to 1.1 can be checked at the same time, corresponding call records, transfer records, chat records, purchase records and the like), a relation graph corresponding to a person entity space and a virtual space and the analysis of the movement track of the persons within thirty days.

Step S30 specifically includes: based on algorithms such as GBDT, C4.5, K-means, CART, SVM and the like, in combination with business experience, the invention provides a multi-dimensional early warning and perception model such as character multi-dimensional analysis, social relationship network evolution analysis, individual violation early warning, group violation early warning and social security risk evolution and the like for different application scenes by combining a social relationship network evolution model construction and analysis technology, an individual violation behavior intelligent early warning technology based on deep learning, a group violation behavior intelligent early warning technology based on social relationship network evolution, a social security risk evolution simulation and critical early warning technology and the like, and the social security risk intelligent prediction early warning and trend evolution analysis are as follows:

(1) The social relationship network evolution model construction and analysis technology comprises the following steps: the method comprises the steps of constructing a person relation knowledge graph for suspected objects, excavating detailed incidence relation of the person relation knowledge graph, combining a social relation network evolution model, carrying out 'gene' extraction, relation cognition, graph reconstruction, graph prediction and the like on a relation network of the suspected objects, extracting the most key social relation network (such as organization relation, group relation and the like) of the suspected objects, finding out the graph of each key event node, and visually displaying the evolution of the social relation network.

(2) The intelligent early warning technology for the individual behaviors based on deep learning comprises the following steps: various individual historical case information including attribute information of dangerous personnel, historical behavior information, reality and virtual space information and the like is analyzed through technologies such as artificial intelligence and deep learning, expert knowledge or information which has a large influence on illegal behaviors is extracted at the same time, the related behavior information of various individuals is learned, modeled and trained, the characteristics are extracted, label sets of various individual illegal behaviors are formed, and various individual intelligent early warning models are formed. The model is applied to the analysis of dangerous personnel, the abnormal behavior of the personnel is identified, and after the personnel hit a plurality of model labels, the early warning occurrence probability (early warning score) of the personnel is calculated through an early warning model algorithm, so that the aim of early warning is fulfilled.

(3) The group behavior intelligent early warning technology based on social relationship network evolution comprises the following steps: through the machine learning of historical group early warning information, the NLP technology is used for extracting the labels of historical case events, new case events are continuously input into the system, the NLP is continuously extracted, the latest case event behavior characteristics are continuously extracted, and the purpose of machine learning is achieved.

(4) The social security risk evolution simulation and critical early warning technology comprises the following steps: the method comprises the steps that historical social security events are learned through a machine, the labels of the historical events are extracted through an NLP technology, new events are continuously input into a system, the NLP is continuously extracted, and the latest event characteristics are continuously extracted, so that an early warning model is constructed.

The individual risk overview and the group/group evolution route are predicted by analyzing and processing various social security perception data by adopting technologies such as Hadoop, spark, storm and the like, and the data are analyzed by depending on the existing mature and improved GBDT algorithm, data mining algorithm and various data analysis models, so that the data opening capability is provided for upper-layer application.

According to the social security prevention and control system based on big data depth correlation and intelligent early warning, the outstanding problems of weak fusion, difficult correlation and incomplete early warning of various social security means data are solved by means of big data technology, knowledge map technology and visual analysis technology through information extraction and visual correlation analysis, the fundamental targets of deep fusion correlation of various perception data, intelligent early warning model construction and early warning and prejudgment on social security risks are achieved, and intelligent early warning prediction, prejudgment on social security risks, auxiliary decision making and the like of cases are provided for business lines of public security, criminal investigation, information, command and the like. The intelligent support is provided for important security, daily patrol prevention and control, command decision, prevention and the like; the method creates a new pattern with big data enabling safety, comprehensively improves the supporting force of scientific and technical information on social security risk prevention and control, and continuously promotes the development of social security information work.

It should be noted that while the operations of the method of the present invention are depicted in the drawings in a particular order, this does not require or imply that the operations must be performed in this particular order, or that all of the illustrated operations must be performed, to achieve desirable results. Rather, the steps depicted in the flowcharts may change the order of execution.

As shown in fig. 5, as another aspect, the present application also provides an electronic device 300 including one or more Central Processing Units (CPUs) 301 that can perform various appropriate actions and processes according to a program stored in a Read Only Memory (ROM) 302 or a program loaded from a storage section 308 into a Random Access Memory (RAM) 303. In the RAM 303, various programs and data necessary for the operation of the system 300 are also stored. The CPU 301, ROM 302, and RAM 303 are connected to each other via a bus 304. An input/output (I/O) interface 305 is also connected to bus 304.

The following components are connected to the I/O interface 305: an input portion 306 including a keyboard, a mouse, and the like; an output section 307 including a display such as a Cathode Ray Tube (CRT), a Liquid Crystal Display (LCD), and the like, and a speaker; a storage section 308 including a hard disk and the like; and a communication section 309 including a network interface card such as a LAN card, a modem, or the like. The communication section 309 performs communication processing via a network such as the internet. A drive 310 is also connected to the I/O interface 305 as needed. A removable medium 311 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, or the like is mounted on the drive 310 as necessary, so that the computer program read out therefrom is mounted into the storage section 308 as necessary.

In particular, the process described above with reference to fig. 4 may be implemented as a computer software program, according to an embodiment of the present disclosure. For example, embodiments of the present disclosure include a computer program product comprising a computer program tangibly embodied on a machine-readable medium, the computer program comprising program code for performing a big-data based social security alert method. In such an embodiment, the computer program may be downloaded and installed from a network through the communication section 309, and/or installed from the removable medium 311.

The flowchart and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present invention. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems that perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.

As yet another aspect, the present application also provides a computer-readable storage medium, which may be the computer-readable storage medium included in the apparatus in the above embodiments; or it may be a separate computer readable storage medium not incorporated into the device. The computer readable storage medium stores one or more programs for use by one or more processors in performing the big data based social security alert method described in the present application.

The units or modules described in the embodiments of the present application may be implemented by software or hardware. The described units or modules may also be provided in a processor, for example, each of the described units may be a software program provided in a computer or a mobile intelligent device, or may be a separately configured hardware device. Wherein the designation of a unit or module does not in some way constitute a limitation of the unit or module itself.

The above description is only a preferred embodiment of the application and is illustrative of the principles of the technology employed. It will be appreciated by a person skilled in the art that the scope of the invention according to the present application is not limited to the specific combination of the above-mentioned features, but also covers other embodiments where any combination of the above-mentioned features or their equivalents is made without departing from the inventive concept. For example, the above features may be replaced with (but not limited to) features having similar functions disclosed in the present application.

Claims

1. A big data-based social security early warning system is characterized by comprising: the system comprises an infrastructure layer, a platform software layer, a data information layer and an application software layer which are arranged from bottom to top in sequence; wherein the content of the first and second substances,

2. The big data-based social security pre-warning system according to claim 1, wherein the application software layer comprises at least one of: the system comprises a figure multidimensional analysis module, a social relationship network evolution module, an individual early warning module, a group early warning module, a social security risk evolution module and an early warning model management module.

3. The big data-based social security early warning system according to claim 2, wherein the people multidimensional analysis module comprises a multidimensional relation search unit, a multidimensional track search unit and a dangerous entity portrait unit; wherein the content of the first and second substances,

the dangerous entity portrait unit is used for extracting the characteristics of dangerous personnel and forming a portrait of the dangerous personnel.

4. The big-data based social security pre-warning system of claim 2, wherein the social relationship network evolution module comprises at least one of: a group relation network evolution unit and an organization relation network evolution unit.

5. The big data based social security pre-warning system of claim 2, wherein the individual pre-warning module comprises at least one of: the first individual early warning unit, the second individual early warning unit, the third individual early warning unit, the fourth individual early warning unit and the fifth individual early warning unit.

6. The big data based social security pre-warning system of claim 2, wherein the group pre-warning module comprises at least one of: the early warning system comprises a first group early warning unit, a second group early warning unit, a third group early warning unit and a fourth group early warning unit.

7. The big data-based social security pre-warning system according to claim 2, wherein the social security risk evolving module comprises at least one of: a large-scale activity safety risk unit, a public health safety risk unit, a financial safety risk unit and a key area safety risk unit.

8. The big data-based social security early warning system according to claim 1, wherein the data information layer comprises a service subject library and a superior data interface, and the service subject library is used for constructing incidence relations of various types of data, incidence relations of various types of tracks and service knowledge; and the superior data interface is used for butting various superior track data interfaces, supplementing various local track data and forming a complete track data chain of the target object.

9. The big data-based social security pre-warning system according to claim 8, wherein the business topic library comprises at least one of: the system comprises a human subject database, a multi-dimensional track database, a business relation database, an association track database and a business knowledge database;

10. The big data-based social security pre-warning system of claim 1, wherein the platform software layer comprises at least one of: the system comprises a population data interface, a map data interface, a vehicle card port data interface, an electronic fence data interface, a WiFi data interface and a face recognition data interface.

11. A social security early warning method based on big data, the social security early warning system based on big data of any claim 1-10, comprising:

12. An electronic device, comprising:

one or more processors;

a memory for storing one or more programs,

the one or more programs, when executed by the one or more processors, cause the one or more processors to perform the big data based social security pre-warning method of claim 11.

13. A computer-readable storage medium storing a computer program, wherein the program, when executed by a processor, implements the big-data based social security early warning method as claimed in claim 11.