CN108876136A - Recommend the attack of terrorism methods of risk assessment of innovatory algorithm based on position - Google Patents
Recommend the attack of terrorism methods of risk assessment of innovatory algorithm based on position Download PDFInfo
- Publication number
- CN108876136A CN108876136A CN201810594064.7A CN201810594064A CN108876136A CN 108876136 A CN108876136 A CN 108876136A CN 201810594064 A CN201810594064 A CN 201810594064A CN 108876136 A CN108876136 A CN 108876136A
- Authority
- CN
- China
- Prior art keywords
- attack
- terrorism
- data
- risk assessment
- formula
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 63
- 238000012502 risk assessment Methods 0.000 title claims abstract description 46
- 238000012545 processing Methods 0.000 claims abstract description 15
- 230000008569 process Effects 0.000 claims abstract description 11
- 238000004458 analytical method Methods 0.000 claims abstract description 9
- 238000012795 verification Methods 0.000 claims abstract description 7
- 230000000694 effects Effects 0.000 claims description 16
- 238000011156 evaluation Methods 0.000 claims description 14
- 238000013210 evaluation model Methods 0.000 claims description 13
- 238000012360 testing method Methods 0.000 claims description 11
- 235000018185 Betula X alpestris Nutrition 0.000 claims description 10
- 235000018212 Betula X uliginosa Nutrition 0.000 claims description 10
- 238000013459 approach Methods 0.000 claims description 6
- 238000012935 Averaging Methods 0.000 claims description 5
- 238000002790 cross-validation Methods 0.000 claims description 5
- 239000003814 drug Substances 0.000 claims description 5
- 229940079593 drug Drugs 0.000 claims description 5
- 238000003064 k means clustering Methods 0.000 claims description 5
- 238000010606 normalization Methods 0.000 claims description 5
- 238000001556 precipitation Methods 0.000 claims description 5
- 238000013442 quality metrics Methods 0.000 claims description 5
- 238000012549 training Methods 0.000 claims description 5
- 238000010200 validation analysis Methods 0.000 claims description 5
- 235000013305 food Nutrition 0.000 claims description 3
- 230000006855 networking Effects 0.000 claims description 3
- 238000005192 partition Methods 0.000 claims description 2
- 238000003339 best practice Methods 0.000 claims 1
- 230000006870 function Effects 0.000 description 10
- 238000000354 decomposition reaction Methods 0.000 description 7
- 238000011160 research Methods 0.000 description 7
- 241001269238 Data Species 0.000 description 5
- 210000002569 neuron Anatomy 0.000 description 5
- 241000208340 Araliaceae Species 0.000 description 4
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 4
- 235000003140 Panax quinquefolius Nutrition 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 235000008434 ginseng Nutrition 0.000 description 4
- 230000004913 activation Effects 0.000 description 3
- 230000006872 improvement Effects 0.000 description 3
- 238000010801 machine learning Methods 0.000 description 3
- 210000004027 cell Anatomy 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000013480 data collection Methods 0.000 description 2
- 238000011524 similarity measure Methods 0.000 description 2
- 238000012732 spatial analysis Methods 0.000 description 2
- 206010001497 Agitation Diseases 0.000 description 1
- 230000018199 S phase Effects 0.000 description 1
- 230000002776 aggregation Effects 0.000 description 1
- 238000004220 aggregation Methods 0.000 description 1
- 230000003466 anti-cipated effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000008094 contradictory effect Effects 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000005284 excitation Effects 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000011478 gradient descent method Methods 0.000 description 1
- 238000000227 grinding Methods 0.000 description 1
- 230000005764 inhibitory process Effects 0.000 description 1
- 239000000155 melt Substances 0.000 description 1
- 238000002156 mixing Methods 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 238000013139 quantization Methods 0.000 description 1
- 230000008439 repair process Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000000547 structure data Methods 0.000 description 1
- 238000003786 synthesis reaction Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0635—Risk analysis of enterprise or organisation activities
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/26—Government or public services
- G06Q50/265—Personal security, identity or safety
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Data Mining & Analysis (AREA)
- Tourism & Hospitality (AREA)
- General Physics & Mathematics (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Educational Administration (AREA)
- Entrepreneurship & Innovation (AREA)
- Development Economics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Primary Health Care (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Health & Medical Sciences (AREA)
- Evolutionary Biology (AREA)
- General Health & Medical Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computer Security & Cryptography (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Probability & Statistics with Applications (AREA)
- Game Theory and Decision Science (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Complex Calculations (AREA)
Abstract
The invention discloses a kind of attack of terrorism methods of risk assessment for recommending innovatory algorithm based on position, including：Element processing, region division, risk assessment, model verification step；For the attack of terrorism, from grid scale by clustering algorithm in conjunction with the proposed algorithm of position, based on introducing influent factor subregion process in the proposed algorithm of position and probably attacking the space characteristics of event, space risk assessment is carried out to the attack of terrorism on the basis of comprehensive analysis attack of terrorism influent factor and has passed through the validity check of model.The method of the present invention greatly improves the accuracy and accuracy of attack of terrorism assessment.
Description
Technical field
The present invention relates to attack of terrorism appraisal procedures, and in particular to a kind of to be based on position proposed algorithm pair using improvement
The attack of terrorism carries out the high efficiency method of risk assessment, belongs to attack of terrorism risk assessment technology field.
Background technique
The attack of terrorism causes great harm to people's life and property safety, seriously affected international community stablize and
Economic development.Therefore, explore that regularity that the attack of terrorism occurs, to establish attack of terrorism risk evaluation model steady to nation's security
The fixed situation with global anti-terrorism is of great significance.
A large amount of scholar are that seek to solve the problems, such as in relation to the threat of terrorism various have made great efforts both at home and abroad, however
It is still a complexity and probabilistic problem about probably the risk assessment of event is attacked.It is current probably attack assessment be faced with it is new
Challenge, on the one hand, the presence of internet makes each corner of world, each stratum and every field connection even closer,
Sensitive variable and disturbance variable in relation to probably attacking assessment unprecedentedly increase；On the other hand, unprecedentedly pushing away due to global figure process
Into the application with various advance data acquisition means, probably attacking assessment can obtain than more polymorphic type in the past, more greatly from all angles
The related data of the scale of construction, to be also required to dexterousr, more efficient complex data processing capacity.
In many related works, leading one of Blair et al. probably attacks prediction project and uses 2008 annual datas and mind
Through Liberia's conflict in 2010 of network success prediction, and accuracy rate is between 0.65-0.74；Sivasamy et al. is proposed
A kind of new prediction technique, i.e., using mixing averaging model (MABM) zivile Opfer data caused by the attack of terrorism of South Asia into
The zivile Opfer event that row is fitted and predicts 2014；Raghavan et al. uses hidden Markov model for terroristic organization's activity
Situation establishes model, detects the emergency case of the tissue；Saha proposes a kind of method for predicting future attacks, uses a kind of function
Can powerful machine learning algorithm (referred to as integrated study (ensemble learning)) predict the weapon that may be attacked
And target.But the above-mentioned research method for the attack of terrorism seldom considers to influence the multi-source element of the attack of terrorism；And mostly it is
It is unfolded from country scale；Usually start in terms of the time series for probably attacking event generation or with regard to event itself, and
The spatial distribution characteristic for probably attacking event generation is not considered.
Risk assessment is a certain event of quantization assessment or things bring influences or the possibility degree of loss, and the attack of terrorism
Space risk assessment is to be assessed with occurrence risk from spatial analysis angle the position that the attack of terrorism occurs, including but unlimited
In assessing the risk for not occurring probably to attack event location using the position that the attack of terrorism has occurred.Existing correlation technique does not consider
The spatial distribution characteristic of event generation is probably attacked, the accuracy and accuracy of attack of terrorism assessment be not high.
Summary of the invention
Existing method there are aiming at the problem that, the present invention provide it is a kind of based on position recommend innovatory algorithm attack of terrorism thing
Part methods of risk assessment, for the attack of terrorism, from grid scale by clustering algorithm in conjunction with the proposed algorithm of position, be based on
Influent factor subregion process is introduced in the proposed algorithm of position and probably attacks the space characteristics of event, in comprehensive analysis attack of terrorism thing
Space risk assessment is carried out to the attack of terrorism on the basis of part influent factor and has passed through the validity check of model.The party
Method greatly improves the accuracy and accuracy of attack of terrorism assessment.
Technical solution provided by the invention is：
A kind of attack of terrorism methods of risk assessment for recommending innovatory algorithm based on position, for the attack of terrorism,
Attack of terrorism space risk evaluation model is established by introducing influent factor subregion process and fearing to attack the space characteristics of event；It is described
Attack of terrorism space risk evaluation model from grid scale by clustering algorithm in conjunction with the proposed algorithm of position, comprehensive analysis terror is attacked
Event influent factor is hit, is achieved in and space risk assessment is carried out to the attack of terrorism；
Establishing attack of terrorism space risk evaluation model includes：Element processing, region division, risk assessment, model verifying
Step.Detailed process includes：
1) element is handled：Choose and construct the attack of terrorism influent factor of spatialization:Comprehensively utilize society, nature, geography,
The multi-source heterogeneous data such as economy, attack of terrorism data construct attack of terrorism influent factor database；When it is implemented, probably
It is afraid of and attacks in influent factor database including multiclass attack of terrorism influent factor, wherein social economy's element includes race's multiplicity
Property, main drugs area, the density of population and night lights, lodging site, food and drink site, traffic website, religious site, political field
Institute；Natural resources element includes average precipitation, temperature on average and landform, the distance away from main navigation lake, to no ice ocean
Distance, at a distance from main navigable river；
In order to facilitate the attack of terrorism influence factor regularity of distribution of expression territory element, attack of terrorism influence factor number is realized
According to the building and expression of spatial model, reticle networking processing is carried out to it, obtains the spatial data of attack of terrorism influent factor；
It is normalized again, makes it have unified space scale；It performs the following operations：
11) attack of terrorism basic data and influent factor data are pre-processed, deletes part singular data, then
Data are screened, and carry out geocoding matching, finally rectify a deviation to have obtained the attack of terrorism and influent factor by data
Spatial data；
12) its influent factor data is normalized, unified yardstick；
Multiple influent factors are normalized especially by formula 1：
Wherein, XnormIt is the attack of terrorism influent factor value after element normalization, XminIt is in attack of terrorism influent factor
Minimum value, XmaxIt is the maximum value in attack of terrorism influent factor, n is the quantity of the element.
2) region division：Region division is carried out to space using a variety of clustering algorithms, not by clustering result quality metrics evaluation
The best proportion of each influent factor of different zones is obtained with the Clustering Effect of clustering method, and using maximum information coefficient (MIC)
Relationship；
21) region division is carried out to factor data using a variety of Spatial Clusterings；
When it is implemented, a variety of Spatial Clusterings chosen include：K-means clustering algorithm, BIRCH clustering algorithm,
DBSCAN clustering algorithm and SOM clustering algorithm.
22) clustering method different come comparison using Cluster Assessment standard obtains optimal spatial clustering algorithm, chooses optimal
The region division result that method is clustered；
23) to region division as a result, calculating different zones element to the attack of terrorism using maximum information coefficient (MIC)
Influence degree, and then obtain the best proportion relationship of each influent factor of different zones；
3) risk assessment：Density of the attack of terrorism scene in its surrounding neighbors is calculated using cuclear density analytic approach,
Attack of terrorism plot does not occur using the calculating of euclidean metric method and has occurred between attack of terrorism plot
The two is finally combined and carries out the risk assessment of attack of terrorism space by similitude；It performs the following operations：
31) density of the attack of terrorism scene in its surrounding neighbors is calculated using cuclear density analytic approach；
Cuclear density value is calculated especially by formula 5：
In formula, f (s) is that the cuclear density at the s of spatial position calculates function；H is range attenuation threshold value, i.e. bandwidth；N be and position
Element of the distance less than or equal to h for setting s is counted；xiFor each key element；L () function is kernel function；The geometry of formula 5 is anticipated
Justice is density value in each key element xiPlace is maximum, and with xiDistance continue to increase during constantly reduce, until
With key element xiDistance when reaching bandwidth h cuclear density value be reduced to 0.
32) attack of terrorism plot does not occur using the calculating of euclidean metric method and the attack of terrorism has occurred
Similitude between plot；
It is calculated especially by formula 6 and attack of terrorism plot does not occur and has occurred between attack of terrorism plot
Similarity：
In formula, X, Y indicate two sample points, xiIndicate the ith feature value of sample X, yiIndicate the ith feature of sample Y
Value, wherein the plot of the attack of terrorism has occurred for X expression, and Y indicates the plot that the attack of terrorism does not occur.N indicates sample characteristics
Number；D (X, Y) is the Euclidean distance between two sample points, indicate the plot for not occurring probably to attack and having occurred probably attack plot it
Between similitude.Attack of terrorism plot is not occurred and has been occurred according to the element factor calculation in plot by Euclid's method by the present invention
The similitude in attack of terrorism plot.
33) attack of terrorism space risk assessment：It selects highest multiple with the block similarity attack of terrorism does not occur
(such as three) region, cuclear density value and corresponding similarity figure to these regions are weighted averaging, calculate
Score attack of terrorism plot does not occur as probably a possibility that attacking event degree occurs；
It is calculated with the factor data after imparting weight, by the plot that the attack of terrorism does not occur and the attack of terrorism has occurred
Plot carry out Similarity measures, select with the block similarity highest three plot attack of terrorism does not occur, to these three ground
The cuclear density value of block and corresponding similarity figure are weighted averaging, and the score of calculating does not occur terror as and attacks
It hits event plot and a possibility that probably attacking event degree occurs.Score is higher, indicates that a possibility that attack of terrorism occurs for the plot is got over
Greatly.
4) validation verification is carried out to attack of terrorism space risk evaluation model using accurate rate, recall rate and F value.
When it is implemented, for trained and test space risk evaluation model performance, the present invention is tested using ten foldings intersection
Demonstration.Data set is divided into very, in turn will wherein 9 parts be used as training data, 1 part be used as test data, verified.Often
Sample data in a test set can all obtain the score between one 0 to 1, take threshold value 0.1 to 0.9 to be verified respectively, comment
Valence index selects accurate rate, recall rate and F value.The present invention has carried out 10 ten folding cross validations, and is averaged conduct
Finally to the estimation of model accuracy.
Compared with prior art, the beneficial effects of the invention are as follows：
The present invention provides a kind of attack of terrorism methods of risk assessment of the improvement based on position proposed algorithm, by wanting
The space risk assessment to the attack of terrorism is completed in plain processing, region division, risk assessment, model verifying.The present invention is from lattice
Net scale combines clustering algorithm with position proposed algorithm, comprehensively considers different zones multi-source elements affect degree difference and melts
The spatial distribution characteristic of the attack of terrorism is entered；Space risk assessment is carried out to the attack of terrorism on this basis and is passed through
The validity check of model.This method greatly improves the accuracy and accuracy of attack of terrorism assessment.
Detailed description of the invention
Fig. 1 is the flow diagram of attack of terrorism space provided by the invention methods of risk assessment.
Fig. 2 is the Calinski- that the specific embodiment of the invention uses K-means algorithm to be calculated under different parameters
Harabaz index value.
Fig. 3 is the Calinski- that the specific embodiment of the invention uses BIRCH algorithm to be calculated under different parameters
Harabaz index value.
Fig. 4 is the Calinski- that the specific embodiment of the invention uses DBSCAN algorithm to be calculated under different parameters
Harabaz index value.
Fig. 5 is the Calinski- that the specific embodiment of the invention uses SOM algorithm to be calculated under different parameters
Harabaz index value.
Fig. 6 is the Calinski-Harabaz index value that the specific embodiment of the invention uses four kinds of clustering algorithms to be calculated
As a result compare.
Fig. 7 is the result figure for carrying out Spacial domain decomposition in the specific embodiment of the invention using K-means algorithm.
Fig. 8 is the cuclear density analysis result figure that the specific embodiment of the invention provides.
Fig. 9 is the accurate rate comparison diagram of subregion and non-subregion under the different threshold values that the specific embodiment of the invention provides.
Figure 10 is the recall rate comparison diagram of subregion and non-subregion under the different threshold values that the specific embodiment of the invention provides.
Figure 11 is the F value comparison diagram of subregion and non-subregion under the different threshold values that the specific embodiment of the invention provides.
Figure 12 is the attack of terrorism space risk evaluation result figure that the specific embodiment of the invention provides.
Specific embodiment
With reference to the accompanying drawing, the present invention is further described by example, but do not limit the invention in any way obtains range.
The present invention provides a kind of attack of terrorism methods of risk assessment of the improvement based on position proposed algorithm, from grid
Scale combines clustering algorithm with position proposed algorithm, comprehensively considers different zones multi-source elements affect degree difference and incorporates
The spatial distribution characteristic of the attack of terrorism；By element processing, region division, risk assessment, realize to attack of terrorism thing
Part carries out space risk assessment, and has passed through the validity check of model.
The method of the present invention process is as shown in Fig. 1, and (1) is chosen and constructs the attack of terrorism influent factor of spatialization；(2) it adopts
Region division is carried out to space with clustering algorithm, by clustering result quality metrics evaluation Clustering Effect, and utilizes maximum information coefficient
(MIC) the best proportion relationship of each influent factor of different zones is obtained；(3) attack of terrorism is calculated using cuclear density analytic approach
Attack of terrorism plot does not occur and has sent out using the calculating of euclidean metric method for density of the place in its surrounding neighbors
Similitude between raw attack of terrorism plot, the two is finally combined carry out the risk assessment of attack of terrorism space；(4)
Validation verification is carried out to attack of terrorism space risk evaluation model using accurate rate, recall rate and F value.The present invention is with the southeast
Subregion is for studying area, and specific implementation comprises the following specific steps that：
A. choose and construct the attack of terrorism influent factor of spatialization
The present invention comprehensively utilizes the multi-source heterogeneous data such as society, nature, geography, economy, attack of terrorism data, building
Attack of terrorism influent factor database.Since attack of terrorism basic data and influence factor data are the knots with address information
Structure data need to carry out spatialization processing to these data for subsequent spatial modeling.Simultaneously as the attack of terrorism influence because
The multi-source of prime number evidence expresses the attack of terrorism influence factor regularity of distribution of territory element for convenience, realizes that terror is attacked
The building and expression of influence factor data space model are hit, reticle networking processing need to be carried out to it, makes it have unified sky
Between scale, and grid is exactly that the polygon (present invention select 0.1 × 0.1 degree) not overlapped is divided on geographical space-time, often
A polygon is exactly a space cell, is convenient to express each statistic unit information, the attack of terrorism of grid by grid
Influence factor data can not only it is more intuitive, be more nearly and be truly reflected reality, while also being provided for the fusion of data
Unified space reference.
The present embodiment is directed to multiple regions/country, and (such as Vietnam, Cambodia, Thailand, Burma, Malaysia, newly adds Laos
Slope, Indonesia, Brunei, Philippine, East Timor), about 4,570,000 sq-km of area.Mainly from social economy's element and
Two aspects of natural resources element have collected multiclass influent factor data.Wherein, social economy's element includes ethnic diversity, main
Want drugs area, the density of population and night lights, lodging site, food and drink site, traffic website, religious site, political place；From
Right element of resource includes average precipitation, temperature on average and landform, the distance away from main navigation lake, to no ice ocean away from
From, at a distance from main navigable river etc..
A1. normal grid spatialization processing is carried out to data
Attack of terrorism basic data and influent factor data are pre-processed first, delete part singular data, so
Data are screened afterwards, and carry out geocoding matching, finally rectify a deviation to have obtained the attack of terrorism by data and influence to want
The spatial data of element.Make all data that can be visualized and be modeled on unified scale, thus complete paired data mark
Quasi- grid spatialization processing.The present invention influences 17 classes collected in terms of social economy's element and natural resources element two
The event data of fearing to attack that factor data and 1970-2016 occur has carried out normal grid spatialization processing (0.1 × 0.1 degree),
Form 36978 standard grids.
A1.1. the information of the attack of terrorism in 1970 to 2014 is converted into raster data, selects 0.1 × 0.1
The grid of resolution ratio is spent as unit, to count the quantity and total number of casualties that determine terrorist incident.
A1.2. using ArcMap 10.3 will away from main navigation lake distance (km), the distance (km) away from main navigation river, away from
The raster data of distance (km) without ice sea, average precipitation (mm/) and temperature on average (DEG C) samples 0.1 × 0.1 degree of grid
In lattice.
A1.3. use ArcMap10.3 by ethnic diversity, main drugs region, night lights, the density of population and landform
Data sampling is into 0.1 × 0.1 degree of grid.
A1.4. Baidu map is crawled using the library requests in python3.6.For the Southeast Asia crawled
POI json formatted data, the Baidu's Mercator's coordinate transformation that carries out processing deposit table to it using python and will crawl
For WGS84 coordinate, then sampled in 0.1 × 0.1 degree of grid using ArcMap10.3.
A2. data are normalized, unified yardstick.
Attack of terrorism influent factor data have different dimension and the order of magnitude, if directly handling initial data,
May the lesser index of negligible number grade so that assessment result is not accurate enough.Since attack of terrorism influent factor is different
Unit, in order to which unified yardstick avoids the different difference present invention wanted between primitive unit cell from carrying out normalizing to multiple influent factors
Change, normalization formula is as shown in Equation 1.
Wherein, XnormIt is the value after element normalization, XminIt is the minimum value in attack of terrorism influent factor, XmaxIt is terrified
The maximum value in influent factor is attacked, n is the quantity of the element.
B. region division is carried out to space using clustering algorithm, by clustering result quality metrics evaluation Clustering Effect, and utilized
Maximum information coefficient (MIC) obtains the best proportion relationship of each influent factor of different zones.
Since the influence degree that the attack of terrorism occurs for each element in the different zones in research area is different, this
Invention considers that will study area according to factor data carries out space division, is then obtained often using maximum information coefficient (MIC) first
The best proportion relationship of each influent factor in a subregion.Spacial domain decomposition to research area is will to study area's spatially foundation
Influence factor carries out region division, and which belongs to unsupervised learning in machine learning algorithm, and clustering algorithm is typical
Unsupervised machine learning algorithm.Cluster is exactly the data set to largely unknown marks, according to the inherent similitude of data by data
Collection is divided into multiple class clusters, the entity in class cluster be it is similar, the entity of inhomogeneity cluster is dissimilar；One class cluster is test
The convergence at space midpoint, the distance between any two point of same class cluster are less than the distance between any two point of inhomogeneity cluster.
When it is implemented, choosing four kinds of classical clustering algorithms, by Experimental comparison, the calculation for being most suitable for region division of the present invention is obtained
Method.
B1. region division is carried out using K-means algorithm
K-means is the clustering algorithm based on division.K-means algorithm is calculated according to the average value of data object in cluster
Similarity regard the average value (or being mass center) of object in cluster as cluster center, and algorithm selects in n data object at random first
K object is selected, each object represents cluster average value；To remaining each object, according to it at a distance from each cluster center,
By apart from the smallest principle, nearest cluster is assigned these to；On this basis, the average value of each cluster is recalculated；So
Back and forth, until the value of error sum of squares is minimum, i.e., by formula 2：
The value being calculated is minimum, at this point, the member in cluster is no longer changed.In formula, i1It is given data object,
wjIt is the average value of cluster Cj.Research area is carried out using K-means clustering algorithm according to 17 multi-source factor datas being collected into
Spacial domain decomposition.As shown in Fig. 2, tune ginseng is carried out for K-means algorithms selection clusters number 2 to 10, finds cluster numbers
K-means Clustering Effect is best when mesh is 2.
B2. region division is carried out using BIRCH algorithm
BIRCH is a comprehensive Hierarchical clustering methods, is a kind of clustering algorithm for large-scale dataset.The algorithm
Two concepts of middle introducing：Cluster feature (Clustering Feature, CF) and clustering tree (CF-tree), by this two
A concept summarizes cluster, using the distance between each cluster, is advised using the equilibrium iteration of hierarchical method to data set
About and cluster.BIRCH method is saved memory, is calculated fastly, and only need to scan a data set can contribute, and can recognize noise spot.
But BIRCH is poly- to the cluster of non-spherical and high dimensional data Clustering Effect is bad.In addition, the sequence of data input will affect algorithm
Result.Spacial domain decomposition is carried out to research area using BIRCH clustering algorithm according to 17 multi-source factor datas being collected into.
As shown in Fig. 3, tune ginseng is carried out for BIRCH algorithms selection clusters number 2 to 10, BIRCH is clustered when discovery clusters number is 4
Effect is best.
B3. region division is carried out using DBSCAN algorithm
DBSCAN is a more representational density-based algorithms.Cluster is defined as the connected point of density by it
Maximum set, can have region division highdensity enough be cluster.The algorithm needs user to input 2 parameters：One
Parameter is radius (Eps), indicates the range of the circle shaped neighborhood region centered on set point P；Another parameter is centered on point P
The quantity (MinPts) at least put in neighborhood, the more difficult setting of the two parameters, because they need user to cluster data collection
It is rule of thumb set after having understanding substantially.According to 17 multi-source factor datas being collected into using DBSCAN clustering algorithm to grinding
Study carefully area and carries out Spacial domain decomposition.As shown in Fig. 4, for DBSCAN clustering algorithm, choosing eps, (∈-neighborhood is apart from threshold
Value) and min_samples (∈-neighborhood sample number threshold value) carry out tune ginseng, when discovery eps is 0.5 and min_samples is 8
Clustering Effect is best.
B4. region division is carried out using SOM algorithm
SOM network structure is made of input layer and competition layer (output layer).Input layer number is n, and competition layer is by m
The one-dimensional or two-dimensional planar array of a neuron composition, network connect entirely, i.e., each input node is the same as all defeated
Node is connected out.Any dimension input pattern can be mapped to one-dimensional or X-Y scheme in output layer by SOM network, and it is kept to open up
It is constant to flutter structure.Training when using " competition learning " by the way of, the sample of each input found in hidden layer one with it most
Matched node, referred to as its activation node, are also " winning neuron ".And then it is updated and is swashed with stochastic gradient descent method
The parameter of movable joint point.Meanwhile it also suitably being updated according to the distance of their distance activation nodes with the point that closes on of activation node
Parameter.Excitability side feedback is shown to the neuron of neighbour and inhibition side feedback is shown to the neuron of remote neighbour, i.e., closely
Adjacent person's phase mutual excitation, remote neighbour person mutually inhibit.According to 17 multi-source factor datas being collected into using SOM clustering algorithm to research
Area carries out Spacial domain decomposition.As shown in Fig. 5, it for SOM algorithm, chooses neuron number and carries out tune ginseng, number is worked as in discovery
Effect is best when being 2.
B5. clustering result quality metrics evaluation clustering algorithm is utilized
The present invention uses clustering result quality index Calinski-Harabaz (CH) metrics evaluation Clustering Effect.CH index is number
According to the separating degree of collection and the ratio of tightness, tightness in every class data point with represent quadratic sum at a distance from point come degree
Amount, separating degree then represent square measuring for point and data set central point distance with each.The bigger expression class itself of CH index value is more
Closely, more dispersed between class, Clustering Effect is more preferably.
In formula：K indicates clusters number, niIndicate the quantity of the data point in i-th of class, d (ci, c) and indicate i-th of class
Represent point ciWith the distance of data set center c, d (x, ci) represent in class i data point x and represent point c with itiDistance, n indicate number
According to intensive data point sum.
On this basis, cluster effect is selected using Calinski-Harabaz index according to attack of terrorism influent factor data
The best algorithm of fruit.As shown in Fig. 6, the present invention uses in the case of being directed to different clustering algorithm optimized parameters for specific implementation
Calinski-Harabaz index is compared, and finds the Calinski-Harabaz index value highest of K-means algorithm.Therefore
The present invention chooses K-means algorithm and carries out Spacial domain decomposition.
B6. the best proportion relationship of different zones element is calculated using maximum information coefficient (MIC)
Maximum information coefficient is grown up based on mutual information, is dived between variable pair suitable for seeking data set
Incidence relation, have fairness and popularity.
MIC (X, Y | D)=maxI × j < B (n){M(X,Y|D)i,j(formula 4)
In formula：X, Y indicate that variable, n indicate sample size size, and i × j < B (n) indicates the partition dimension boundary of grid G, G
Indicate variable to being divided into i × j grid, M (X, Y | D)i,jIndicate the eigenmatrix of X and Y.B (n)=n in the present invention0.6, it is clear that
0≤MIC≤1。
B7. by optimum cluster method (K-means clustering algorithm), by the attack of terrorism influent factor number of normalized
According to carrying out processing calculating, to obtain the division in region, and it is calculated in different zones respectively and is influenced using maximum information coefficient
It is as shown in Fig. 7 to obtain south east asia region division for the optimal weights of element.
C. calculate density of the attack of terrorism scene in its surrounding neighbors using cuclear density analytic approach, using Europe it is several in
It obtains measure calculating attack of terrorism plot does not occur and the similitude between attack of terrorism plot has occurred, finally
The two is combined and carries out the risk assessment of attack of terrorism space.
Using the factor data after imparting weight as the input of position proposed algorithm, calculate similar between each plot
Property, the kernel density function based on attack of terrorism severity is then constructed, finally combines similitude and cuclear density, is every
A plot that the attack of terrorism does not occur calculates the score between a 0-1, which indicates that the attack of terrorism occurs for the place
A possibility that degree.
C1. density of the attack of terrorism scene in its surrounding neighbors is calculated using cuclear density analytic approach.
Cuclear density analysis is that a kind of method of computational element density in its surrounding neighbors is used in spatial analysis, it will be every
The neighborhood of a element regards a smooth curved surface, the value highest of the element present position, with the value added with the point distance as
It is gradually reduced, is reduced to 0 reaching search radius duration.It is analyzed by cuclear density, vividly can intuitively show certain ground
Manage the hot spot region of phenomenon distribution.The calculation formula for indicating cuclear density is formula 5：
In formula, f (s) is that the cuclear density at the s of spatial position calculates function；H is range attenuation threshold value, i.e. bandwidth；N be and position
Element of the distance less than or equal to h for setting s is counted；L () function is kernel function, and kernel function of the present invention is with the works of Silverman
Described in based on secondary kernel function.The geometric meaning of this equation is density value maximum at each key element xi, and
And constantly reduced during distance xi, until cuclear density value is reduced to 0 when reaching bandwidth h at a distance from core xi.
The present invention carries out cuclear density analysis using ArcMap10.3.In the cuclear density analysis tool of ArcMap10.3,
The expression of Population field is dispersed throughout counting or quantity in the landscape for creating continuous surface, Population of the present invention
Field value is for indicating that the severity of the attack of terrorism represented by the point (combines death toll, number of injured people and wealth
Produce loss), cuclear density of each plot based on attack of terrorism severity is as shown in Fig. 8.
C2. it is calculated using euclidean metric formula and attack of terrorism plot does not occur and the attack of terrorism has occurred
Similitude between event plot.
Euclidean metric (Euclidean metric) (also referred to as Euclidean distance) is the distance definition generallyd use,
Refer to the natural length (i.e. the distance of the point to origin) in actual distance or vector in m-dimensional space between two points.Two
Euclidean distance in peacekeeping three-dimensional space is exactly the actual range between two o'clock.
In formula, X, Y indicate two sample points, wherein the plot of the attack of terrorism has occurred for X expression, and Y expression does not occur
The plot of the attack of terrorism.xiIndicate the ith feature value of sample X, yiIndicate the ith feature value of sample Y, n indicates sample characteristics
Number.Attack of terrorism plot does not occur according to the element factor calculation in plot by Euclid's method and attacks with terror has occurred by the present invention
Hit the similitude in plot.
C3. attack of terrorism space risk assessment
It is calculated with the factor data after imparting weight, by the plot that the attack of terrorism does not occur and the attack of terrorism has occurred
Plot carry out Similarity measures, then select with the block similarity highest three plot attack of terrorism does not occur, to this three
The cuclear density value in a plot and corresponding similarity figure are weighted averaging, and the score of calculating is not feared as
It is afraid of attack plot and a possibility that probably attacking event degree occurs.Score is higher, indicates that the possibility of the attack of terrorism occurs for the plot
Property is bigger.
D. validation verification is carried out to attack of terrorism space risk evaluation model using accurate rate, recall rate and F value.
It is evaluated using the combination F value of accurate rate (Precision), recall rate (Recall) and the two.It is wherein smart
True rate is used to indicate model evaluation as the ratio for actually occurring the attack of terrorism in the plot of high risk, recall rate expression reality
The ratio that model evaluation in the plot of the attack of terrorism is high risk plot occurs, F value is two kinds and comprehensively considers, and can relatively integrate
Ground reflects the assessment performance of model.
For accurate rate for prediction result, what it was indicated is that how many is really just in the sample that is positive of prediction
Sample.So prediction is positive possible with regard to there are two types of, and one kind is exactly that positive class is predicted to be positive class (TP), and another kind is exactly that negative class is pre-
Survey is positive class (FP).
Recall rate be for original sample, what it was indicated be positive example in sample how many be predicted correctly
?.Also there are two types of possibility for that, and one is original positive class prediction, at positive class (TP), another kind is exactly that original positive class is predicted
The class that is negative (FN).
In formula 9, P is accurate rate Precision；R is recall rate Recall；P and R index sometimes will appear contradictory
Situation thus needs to comprehensively consider, and F value is then the evaluation index of comprehensive both index, the finger for concentrated expression entirety
Mark.
Data set is divided into two parts by the present invention, and a part is used for Training valuation model, and another part is for testing the mould
Type.For trained and test space risk evaluation model performance, the present invention uses ten folding cross-validation methods.I.e. by data set point
At very, in turn will wherein 9 parts be used as training data, 1 part is used as test data, is verified.Sample in each test set
Data can all obtain the score between one 0 to 1, take threshold value 0.1 to 0.9 to be verified respectively, evaluation index selection accurate rate,
Recall rate and F value.The present invention has carried out 10 ten folding cross validations, and is averaged as finally estimating to model accuracy
Meter.
D1. accurate rate result is as shown in Fig. 9, as seen from the figure with the raising of threshold value, accurate rate be continuously improved and
Model accurate rate after different threshold value subregions is all much higher than non-subregion.Recall rate result is as shown in Fig. 10, as seen from the figure mould
The recall rate of type is gradually reduced with the increase of threshold value, the model recall rate and non-subregion in different threshold values, after subregion
It is not much different.The results are shown in attached figure 11 by comprehensive accurate rate and the obtained comprehensive evaluation index F of recall rate, as seen from the figure, after subregion
F value with the increase of threshold value first increase then reduce, score threshold be 0.4 when reach maximum value, illustrate mould at this point
Accurate rate and the recall rate synthesis of type are best, and the accurate rate of model is 0.88 at this time, recall rate 0.71；The F value of non-subregion becomes
It is consistent with after subregion to change situation, but in different threshold values, the F value after subregion is all higher than non-subregion, illustrate to survey region into
The method that row divides is scientific and effective.
D2. space risk evaluation result is visualized, it can be seen that attack of terrorism high risk from attached drawing 12
Area (South East Asia Mainland most southern and Philippine).In fact, these areas be the attack of terrorism in recent years high-incidencely.This
Outside the result shows that some coastal areas and national boundary area are in attack of terrorism risk area, thus it is anti-probably heavy in next step
Point should also pay close attention to these areas more.
The present invention creatively carries out region division to research area using clustering method, passes through maximum information coefficient
(MIC) each influent factor best proportion relationship in different zones is obtained, then there is aggregation using the activity in geographical location
Property this feature position proposed algorithm is improved, comprehensive attack of terrorism position element and attack of terrorism category after subregion
Property data complete the attack of terrorism space risk assessment study；The present invention is applied to position proposed algorithm probably for the first time simultaneously
It is afraid of in the risk assessment field of attack space；In terms of influent factor, in present invention specific implementation, 17 classes are had collected for the first time and are wanted
Prime number evidence, including ethnic diversity, main drugs area, the density of population and average precipitation, temperature on average, POI etc. are supported
The foundation of attack of terrorism space risk evaluation model and perfect.It can effectively solve the problem that the terror of coupling multi-source element by the model
The problems such as attacking space risk assessment, while support is provided for relevant Decision person.
It is finally noted that the purpose for publicizing and implementing example is to help to further understand the present invention, but this field
Technical staff be understood that：Without departing from the spirit and scope of the invention and the appended claims, various to replace and repair
It is all possible for changing.Therefore, the present invention should not be limited to embodiment disclosure of that, and the scope of protection of present invention is to weigh
Subject to the range that sharp claim defines.
Claims (10)
1. a kind of attack of terrorism methods of risk assessment for being recommended innovatory algorithm based on position is led to for the attack of terrorism
The space characteristics crossed influent factor subregion and probably attack event establish attack of terrorism space risk evaluation model；The attack of terrorism is empty
Between risk evaluation model from grid scale by clustering algorithm in conjunction with the proposed algorithm of position, the comprehensive analysis attack of terrorism influence
Element is achieved in and assesses the space risk of the attack of terrorism；Including：Element processing, region division, risk are commented
Estimate, model verification step；Detailed process includes：
1) element treatment process：The attack of terrorism influent factor of spatialization is chosen and constructed, attack of terrorism influent factor number is constructed
According to library, including multiclass attack of terrorism influent factor data；Data are pre-processed, at reticle networking processing and normalization
Reason, obtains normalized attack of terrorism influent factor spatial data；
2) region division process：Region division is carried out to space using a variety of clustering algorithms, not by clustering result quality metrics evaluation
The best proportion of each influent factor of different zones is obtained with the Clustering Effect of clustering method, and using maximum information coefficient (MIC)
Relationship；It performs the following operations：
21) region division is carried out to factor data using a variety of Spatial Clusterings；
22) clustering method different come comparison using Cluster Assessment standard obtains optimal spatial clustering algorithm, chooses best practice
It is clustered to obtain region division result；
23) to region division as a result, calculating different zones element to the shadow of the attack of terrorism using maximum information coefficient (MIC)
The degree of sound, and then obtain the best proportion relationship of each influent factor of different zones；
3) risk assessment processes：Density of the attack of terrorism scene in its surrounding neighbors is calculated using cuclear density analytic approach；
The similitude between each plot is obtained by Similarity Algorithm；By similitude in conjunction with cuclear density, selects and terror does not occur
The ground highest multiple regions of block similarity are attacked, cuclear density value and corresponding similarity figure to region, which are weighted, asks flat
, a possibility that attack of terrorism occurs for each plot degree is calculated；
4) using the trained performance with test space risk evaluation model of ten folding cross-validation methods, attack of terrorism space risk is commented
Estimate model and carries out validation verification；
It is achieved in the attack of terrorism risk assessment for recommending innovatory algorithm based on position.
2. recommending the attack of terrorism methods of risk assessment of innovatory algorithm based on position as described in claim 1, characterized in that
Step 1) multiclass attack of terrorism influent factor includes social economy's element and natural resources element；Social economy's element includes race
Diversity, main drugs area, the density of population and night lights, lodging site, food and drink site, traffic website, religious site, political affairs
Control place；Natural resources element includes average precipitation, temperature on average and landform, the distance away from main navigation lake, to no ice
The distance of ocean, at a distance from main navigable river.
3. recommending the attack of terrorism methods of risk assessment of innovatory algorithm based on position as described in claim 1, characterized in that
Influent factor data are normalized in step 12), and multiple influent factors are normalized especially by formula 1：
Wherein, XnormIt is the attack of terrorism influent factor value after element normalization, XminIt is the minimum in attack of terrorism influent factor
Value, XmaxIt is the maximum value in attack of terrorism influent factor, n is the quantity of the element.
4. recommending the attack of terrorism methods of risk assessment of innovatory algorithm based on position as described in claim 1, characterized in that
The a variety of Spatial Clusterings of step 21) include：K-means clustering algorithm, BIRCH clustering algorithm, DBSCAN clustering algorithm and SOM
Clustering algorithm.
5. recommending the attack of terrorism methods of risk assessment of innovatory algorithm based on position as described in claim 1, characterized in that
Step 3) risk assessment processes carry out cuclear density analysis especially by formula 5, cuclear density value are calculated：
In formula, f (s) is that the cuclear density at the s of spatial position calculates function；H is range attenuation threshold value, i.e. bandwidth；N be and position s
Distance less than or equal to h element count；xiFor each key element；L () function is kernel function；
Using euclidean metric method calculating do not occur attack of terrorism plot and occurred attack of terrorism plot it
Between similitude；It is calculated especially by formula 6 and attack of terrorism plot does not occur and attack of terrorism plot has occurred
Between similarity：
In formula, X, Y indicate two sample points, xiIndicate the ith feature value of sample X, yiIndicate the ith feature value of sample Y,
Wherein, X indicates that the plot of the attack of terrorism has occurred, and Y indicates the plot that the attack of terrorism does not occur.N indicates sample characteristics
Number；D (X, Y) is the Euclidean distance between two sample points, indicates the plot for not occurring probably to attack and has occurred probably to attack between plot
Similitude.
6. recommending the attack of terrorism methods of risk assessment of innovatory algorithm based on position as claimed in claim 5, characterized in that
Preferably, it selects with the block similarity highest 3 regions attack of terrorism does not occur, cuclear density value to region and corresponding
Similarity figure be weighted averaging.
7. recommending the attack of terrorism methods of risk assessment of innovatory algorithm based on position as described in claim 1, characterized in that
Step 22) specifically uses clustering result quality index Calinski-Harabaz metrics evaluation Clustering Effect, is expressed as formula 3：
In formula 3：CH index is the separating degree of data set and the ratio of tightness, data point and representative of the tightness in every class
The quadratic sum of the distance of point is measured, and separating degree represents square measuring for point and data set central point distance with each；K indicates poly-
Class number, niIndicate the quantity of the data point in i-th of class, d (ci, c) indicate i-th of class representative point ciWith data set center c
Distance, d (x, ci) represent in class i data point x and represent point c with itiDistance, n indicate data set in data point sum；CH refers to
Scale value is bigger, and expression class itself is closer, and more dispersed between class, Clustering Effect is more preferably.
8. recommending the attack of terrorism methods of risk assessment of innovatory algorithm based on position as described in claim 1, characterized in that
Step 23) calculates the best proportion relationship of different zones element using maximum information coefficient (MIC), is expressed as formula 4：
MIC (X, Y | D)=maxI × j < B (n){M(X,Y|D)i,j(formula 4)
In formula：X, Y indicate that variable, n indicate sample size size；The partition dimension boundary of i × j < B (n) expression grid G；G is indicated
Variable is to being divided into i × j grid；M(X,Y|D)i,jIndicate the eigenmatrix of X and Y；0≤MIC≤1.
9. recommending the attack of terrorism methods of risk assessment of innovatory algorithm based on position as described in claim 1, characterized in that
The F value that step 4) is specifically combined using accurate rate Precision, recall rate Recall and the two；
Accurate rate Precision is calculated by formula 7：
In formula 7, TP indicates the sample of class that the prediction of positive class is positive；FP indicates the sample of class that the prediction of negative class is positive；Accurate rate
Precision indicates how many ratio of real positive sample in the sample for predicting to be positive；
Recall rate Recall is calculated by formula 8：
In formula 10, TP indicates to predict original positive class at positive class；FN indicates class that original positive class prediction is negative；Recall rate
Recall indicates the ratio for the positive class being predicted correctly in sample；
F value is calculated by formula 9：
In formula 9, P is accurate rate Precision；R is recall rate Recall；The comprehensive accurate rate Precision of F value and recall rate
Recall is the overall target of attack of terrorism space risk assessment assessment models validation verification.
10. recommending the attack of terrorism methods of risk assessment of innovatory algorithm, feature based on position as described in claim 1
It is that in step 4), ten folding cross-validation methods are specifically：Data set is divided into 10 parts, in turn will wherein 9 parts be used as training data, 1
Part is used as test data, is verified；The sample data in each test set obtains the score between one 0 to 1 as a result,
Threshold value 0.1 to 0.9 is taken respectively, passes through the estimation averaged as model accuracy.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810594064.7A CN108876136A (en) | 2018-06-11 | 2018-06-11 | Recommend the attack of terrorism methods of risk assessment of innovatory algorithm based on position |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810594064.7A CN108876136A (en) | 2018-06-11 | 2018-06-11 | Recommend the attack of terrorism methods of risk assessment of innovatory algorithm based on position |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108876136A true CN108876136A (en) | 2018-11-23 |
Family
ID=64337489
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810594064.7A Pending CN108876136A (en) | 2018-06-11 | 2018-06-11 | Recommend the attack of terrorism methods of risk assessment of innovatory algorithm based on position |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108876136A (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110288140A (en) * | 2019-06-14 | 2019-09-27 | 西北大学 | A kind of opioid drug spatial prediction technique based on geo-relevance model |
| CN110490608A (en) * | 2019-07-12 | 2019-11-22 | 招联消费金融有限公司 | Methods of risk assessment, device, computer equipment and storage medium |
| CN111161443A (en) * | 2019-01-17 | 2020-05-15 | 浙江诸暨美数信息科技有限公司 | Patrol path setting method based on historical data |
| CN111881939A (en) * | 2020-06-24 | 2020-11-03 | 东南大学 | Shared single-vehicle parking area layout method based on clustering algorithm |
| CN112070315A (en) * | 2020-09-16 | 2020-12-11 | 哈尔滨工业大学 | Terrorist attack network analysis and event prediction method based on centrality measurement |
| CN112069229A (en) * | 2020-08-27 | 2020-12-11 | 贵州民族大学 | Optimal waiting point recommendation method and system for big data of moving track |
| CN113554281A (en) * | 2021-07-02 | 2021-10-26 | 北京淇瑀信息科技有限公司 | Grid-based user business risk analysis method and device and electronic equipment |
| CN116596353A (en) * | 2022-09-29 | 2023-08-15 | 中国人民解放军空军工程大学 | Quantitative analysis method for terrorist attack event record data |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20040249678A1 (en) * | 2003-06-03 | 2004-12-09 | Henderson E. Devere | Systems and methods for qualifying expected risk due to contingent destructive human activities |
| CN103559584A (en) * | 2013-11-08 | 2014-02-05 | 浙江工商大学 | Urban safety accident source tracking and risk pre-warning method based on Internet of things |
| CN105913196A (en) * | 2016-04-29 | 2016-08-31 | 武汉大学 | Navigation channel rectifying social stability risk automatically analyzing method and system |
| CN106952208A (en) * | 2017-03-17 | 2017-07-14 | 讯飞智元信息科技有限公司 | Crime automatic prediction method and system |
| US20170287032A1 (en) * | 2016-04-01 | 2017-10-05 | OneTrust, LLC | Data processing systems for modifying privacy campaign data via electronic messaging systems |
-
2018
- 2018-06-11 CN CN201810594064.7A patent/CN108876136A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20040249678A1 (en) * | 2003-06-03 | 2004-12-09 | Henderson E. Devere | Systems and methods for qualifying expected risk due to contingent destructive human activities |
| CN103559584A (en) * | 2013-11-08 | 2014-02-05 | 浙江工商大学 | Urban safety accident source tracking and risk pre-warning method based on Internet of things |
| US20170287032A1 (en) * | 2016-04-01 | 2017-10-05 | OneTrust, LLC | Data processing systems for modifying privacy campaign data via electronic messaging systems |
| CN105913196A (en) * | 2016-04-29 | 2016-08-31 | 武汉大学 | Navigation channel rectifying social stability risk automatically analyzing method and system |
| CN106952208A (en) * | 2017-03-17 | 2017-07-14 | 讯飞智元信息科技有限公司 | Crime automatic prediction method and system |
Non-Patent Citations (1)
| Title |
|---|
| FANGYU DING 等: "Understanding the dynamics of terrorism events with multiple-discipline datasets and machine learning approach", 《PLOS ONE》 * |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111161443A (en) * | 2019-01-17 | 2020-05-15 | 浙江诸暨美数信息科技有限公司 | Patrol path setting method based on historical data |
| CN110288140A (en) * | 2019-06-14 | 2019-09-27 | 西北大学 | A kind of opioid drug spatial prediction technique based on geo-relevance model |
| CN110288140B (en) * | 2019-06-14 | 2023-04-07 | 西北大学 | Opioid spatial propagation prediction method based on geographical correlation model |
| CN110490608A (en) * | 2019-07-12 | 2019-11-22 | 招联消费金融有限公司 | Methods of risk assessment, device, computer equipment and storage medium |
| CN111881939A (en) * | 2020-06-24 | 2020-11-03 | 东南大学 | Shared single-vehicle parking area layout method based on clustering algorithm |
| CN111881939B (en) * | 2020-06-24 | 2021-03-09 | 东南大学 | Shared single-vehicle parking area layout method based on clustering algorithm |
| CN112069229A (en) * | 2020-08-27 | 2020-12-11 | 贵州民族大学 | Optimal waiting point recommendation method and system for big data of moving track |
| CN112069229B (en) * | 2020-08-27 | 2024-02-27 | 贵州民族大学 | Optimal waiting point recommendation method and system for big data of moving track |
| CN112070315A (en) * | 2020-09-16 | 2020-12-11 | 哈尔滨工业大学 | Terrorist attack network analysis and event prediction method based on centrality measurement |
| CN113554281A (en) * | 2021-07-02 | 2021-10-26 | 北京淇瑀信息科技有限公司 | Grid-based user business risk analysis method and device and electronic equipment |
| CN116596353A (en) * | 2022-09-29 | 2023-08-15 | 中国人民解放军空军工程大学 | Quantitative analysis method for terrorist attack event record data |
| CN116596353B (en) * | 2022-09-29 | 2024-06-04 | 中国人民解放军空军工程大学 | Quantitative analysis method for terrorist attack event record data |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108876136A (en) | Recommend the attack of terrorism methods of risk assessment of innovatory algorithm based on position | |
| Feng et al. | Modeling urban growth with GIS based cellular automata and least squares SVM rules: a case study in Qingpu–Songjiang area of Shanghai, China | |
| Soma et al. | Optimization of causative factors using logistic regression and artificial neural network models for landslide susceptibility assessment in Ujung Loe Watershed, South Sulawesi Indonesia | |
| CN113642849B (en) | Geological disaster risk comprehensive evaluation method and device considering spatial distribution characteristics | |
| CN112819207B (en) | Geological disaster space prediction method, system and storage medium based on similarity measurement | |
| Zhao et al. | A Comparative Study of Landslide Susceptibility Mapping Using SVM and PSO‐SVM Models Based on Grid and Slope Units | |
| CN110458048A (en) | Take population distribution Spatio-temporal Evolution and the cognition of town pattern feature into account | |
| CN107292919B (en) | A kind of multiple dimensioned settlement place matching process of space-oriented subdivision | |
| CN108985327A (en) | A kind of Terrain Matching self-organizing Optimum Classification method based on factorial analysis | |
| CN106028290A (en) | WSN multidimensional vector fingerprint positioning method based on Kriging | |
| CN110031001B (en) | Adaptive area selection method for gravity-assisted inertial navigation | |
| Radhika et al. | Ensemble subspace discriminant classification of satellite images | |
| AlAfandy et al. | Artificial neural networks optimization and convolution neural networks to classifying images in remote sensing: A review | |
| CN117010274B (en) | Intelligent early warning method for harmful elements in underground water based on integrated incremental learning | |
| CN116258279B (en) | Landslide vulnerability evaluation method and device based on comprehensive weighting | |
| Celik et al. | Change detection without difference image computation based on multiobjective cost function optimization | |
| CN107679677A (en) | A kind of drought changes the preferential area's system of selection of water | |
| CN105160065B (en) | Remote sensing information method for evaluating similarity based on topological relation | |
| CN109190678A (en) | A kind of attack of terrorism type evaluation method based on MLKNN and Gravity Models | |
| CN115936773A (en) | Internet financial black product identification method and system | |
| CN113645565B (en) | Indoor positioning method based on hexagonal closest packing structure | |
| Bernat et al. | Two-stage subpixel impervious surface coverage estimation: comparing classification and regression trees and artificial neural networks | |
| Yu et al. | Remote sensing image classification based on dynamic Co-evolutionary parameter optimization of SVM | |
| CN114842328B (en) | Hyperspectral change detection method based on collaborative analysis autonomous perception network structure | |
| Usama et al. | Development of Geo-spatial building stock model for Japanese Commercial Buildings |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20181123 |