CN112052337A - Target relation detection method, system and storage medium based on time-space correlation - Google Patents
Target relation detection method, system and storage medium based on time-space correlation Download PDFInfo
- Publication number
- CN112052337A CN112052337A CN202010720486.1A CN202010720486A CN112052337A CN 112052337 A CN112052337 A CN 112052337A CN 202010720486 A CN202010720486 A CN 202010720486A CN 112052337 A CN112052337 A CN 112052337A
- Authority
- CN
- China
- Prior art keywords
- target
- node
- similarity
- information
- preset
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/367—Ontology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9537—Spatial or temporal dependent retrieval, e.g. spatiotemporal queries
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Animal Behavior & Ethology (AREA)
- Computational Linguistics (AREA)
- Artificial Intelligence (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a method, a system and a storage medium for detecting a target relation based on space-time association, which relate to the technical field of data processing and aim to solve the problem of more accurately and effectively detecting the relation of a target, wherein the method for detecting the target relation based on the space-time association comprises the following steps: constructing a knowledge graph based on the information acquired in the acquisition region; inputting known characteristics of the target based on the knowledge graph, and determining a node representing the target; screening out a candidate node set related to the target based on the nodes representing the target; calculating the similarity between each node in the candidate node set and the node representing the target; based on the similarity, a population strongly correlated with the target is determined.
Description
Technical Field
The invention relates to the technical field of data processing, in particular to a method, a system and a storage medium for detecting a target relation based on space-time correlation.
Background
With the development of the internet, the content of the network data presents an explosive growth situation. Many government departments or enterprises gather a large amount of internal and external data resources, and the data resources are often extracted in a data governance mode, but most of the data are static data, so that the real-time performance and the effectiveness of the data are difficult to ensure after extraction, and in addition, a strong recessive relation is hidden behind the mass data. In real life, the dynamic data with implicit relationship better meets the requirement that the current state of the art is known through data. Of course, the difficulty of extracting and analyzing such data is relatively large, so that it is necessary to use a wider data collection source and a wider data analysis and mining means, so that such data can be extracted and utilized.
Disclosure of Invention
The invention mainly aims to provide a method, a system and a storage medium for detecting a target relation based on space-time association, and aims to solve the problem of how to more accurately and effectively detect the relation of a target.
In order to achieve the above object, the present invention provides a method, a system and a storage medium for detecting a target relationship based on spatio-temporal correlation, which comprises the following steps:
constructing a knowledge graph based on the information acquired in the acquisition region;
inputting known characteristics of the target based on the knowledge graph, and determining a node representing the target;
screening out a candidate node set related to the target based on the nodes representing the target;
calculating the similarity between each node in the candidate node set and the node representing the target;
based on the similarity, a population strongly correlated with the target is determined.
In an embodiment of the present application, constructing the knowledge-graph based on the information collected in the collection area further comprises the steps of:
calling information acquired by acquisition equipment preset in the acquisition area;
generating subgraphs of corresponding layers based on the information acquired by the acquisition equipment;
and associating the information of the subgraph to obtain a corresponding knowledge graph.
In an embodiment of the application, the information related to the object comprises at least location information and time information of the object's presence.
In an embodiment of the present application, screening out a candidate node set related to a target based on the information related to the target, further includes the following steps:
screening out nodes which meet the condition that the nodes appear in a preset range of the place where the target appears in a preset time range based on the place information and the time information of the target;
and classifying the screened nodes into a candidate node set, calling other information of the screened nodes and performing associated storage.
In an embodiment of the present application, calculating a similarity between each node in the candidate node set and the node representing the target further includes:
vectorizing each Node in the candidate Node set through a Node2vec algorithm to obtain a corresponding Node vector;
and calculating the similarity of the node vectors by using a cosine similarity algorithm based on the information of the node vectors.
In an embodiment of the present application, after calculating the similarity of the node vectors by using a cosine similarity algorithm based on the information of the node vectors, the method further includes the following steps:
dividing node vectors meeting the preset same similarity range into the same cluster;
calculating the modularity of each cluster;
and comparing the modularity of each cluster with a preset modularity threshold, when the modularity of each cluster does not reach the preset modularity threshold, adjusting the node vector and repeating the steps, otherwise, ending the current step.
In an embodiment of the application, determining a population strongly related to the target based on the similarity further includes:
comparing the final similarity of any two adjacent node vectors with a preset similarity threshold, judging that the node is not related to the target when the similarity of the node vectors is lower than the preset similarity threshold, and deleting the node and the edge connected with the node; otherwise, judging that the node has no relation with the target, and reserving the node and the edge connected with the node, wherein the reserved node is a group strongly related to the target.
The invention also provides a system for realizing the target relation detection method based on the space-time association, which is characterized by comprising the following steps:
a construction module for constructing the knowledge-graph based on the information acquired by the acquisition device;
an acquisition module for acquiring information related to the target based on the known characteristics of the target;
the screening module is used for screening out a candidate node set related to the target based on the information related to the target;
a calculation module, configured to calculate a similarity between each node in the candidate node set and the node representing the target;
and the determining module is used for determining the group which is strongly related to the target based on the similarity calculated by the calculating module.
The invention also provides a storage medium having stored thereon a plurality of instructions adapted to be loaded by a processor for performing a method according to any one of the preceding claims.
The invention has the following beneficial effects: according to the target relation detection method based on the time-space association, nodes appearing in the preset range of the place where the target appears in the preset time range are classified into the candidate node set, namely the nodes are screened according to the time and place conditions, and the screening accuracy is improved. The similarity between the nodes in the candidate node set and the nodes representing the targets is calculated to determine the group strongly related to the targets, so that the screening accuracy is further improved. The node vectors are adjusted through calculating the modularity, the calculation effectiveness of the similarity is improved, and the accuracy of screening the group strongly related to the target is further ensured.
Drawings
In order to more clearly illustrate embodiments of the present invention or technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only exemplary embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the structures shown in the drawings without inventive effort, wherein:
FIG. 1 is a schematic flow chart of a method for detecting a target relationship based on spatiotemporal correlation according to an embodiment of the present invention;
fig. 2 is a schematic structural diagram of a target relation detection system based on spatiotemporal correlation according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings, and it is obvious that the described embodiments are only exemplary embodiments of the present invention, and not exclusive embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
As shown in fig. 1, the method, system and storage medium for detecting a target relationship based on spatiotemporal correlation provided by the present invention includes the following steps:
s10 constructs a knowledge graph based on the information collected at the collection area.
S11 retrieves information acquired by an acquisition device previously set in the acquisition region.
Specifically, the acquisition device is deployed in the acquisition area and acquires corresponding information by using the acquisition device, the acquisition device may be a camera, a public security card port, a base station, or the like in the prior art, the information acquired by the acquisition device may be face information, license plate information, and positioning information, and the information acquired by the acquisition device may also be information of the acquisition device and time information corresponding to the information acquired by the acquisition device. When the face information or the license plate information is recorded through the camera or the public security card port, the position information corresponding to the main body acquired by the acquisition equipment can be known by recording the position of the acquisition equipment.
S12, generating a subgraph of the corresponding layer based on the information collected by the collecting device.
Specifically, subgraphs of a plurality of corresponding layers are generated based on information acquired by the acquisition equipment, and the subgraphs of the corresponding layers can be a plurality of base station probe subgraphs, face probe subgraphs, license plate probe subgraphs and the like, and the subgraphs of the corresponding layers at least comprise nodes representing the acquisition equipment, nodes representing a main body acquired by the acquisition equipment, namely a face, a license plate, a positioning address and the like, and edges representing connection relations of the nodes.
S13, the information of the subgraph is correlated, and then a corresponding knowledge graph is obtained.
Specifically, the subgraphs of each layer are correlated, that is, the same or similar nodes and/or edges in the subgraphs of the same layer are combined based on the information of the nodes and edges of the subgraphs of each layer, and the same or similar nodes and/or edges in the subgraphs of different layers are correlated by mining the implicit relationship of the nodes in the subgraphs of different layers, so as to generate a knowledge graph capable of displaying the information of the main body acquired by the acquisition device and the connection relationship thereof.
S20 determines a node representing the target based on the knowledge-graph, inputting known characteristics of the target.
Specifically, known features of the target, such as face information, license plate information, location addresses and the like, are input based on the knowledge graph, and are used for calling nodes which represent the features in accordance with the knowledge graph so as to determine the target. After the nodes meeting the characteristics are called, information corresponding to the nodes is obtained, namely information related to the target is obtained, the information related to the target at least comprises place information and time information of the target, and the information related to the target can also be information of other nodes connected with the nodes of the target and information of connection relations between the nodes of the target and the other nodes. Of course, when the input target has less known features, a large number of nodes conforming to the known features may appear, and the nodes representing the target may be determined by screening through human intervention, or by performing similarity calculation on the nodes conforming to the known features.
S30 screens out a set of candidate nodes relevant to the target based on the nodes representing the target.
S31 screens out nodes that meet a preset range of the point where the object appears within a preset time range, based on the point information where the object appears and the time information.
Specifically, the time difference and the space difference are combined to serve as a screening condition, so that the nodes which are in relation with the target can be screened effectively and accurately, the screened nodes are not limited to other nodes which are in connection with the nodes which represent the target at the same side, and the relation of the target in real life is not only the dominant relation which is in connection with the side in the knowledge graph, for example, when a police handles a complex case, the police often needs to search for personnel who are in relation with the target by analyzing a daily scene, and further judge the related group of the target or the track of the target, and based on the dominant relation which is in connection with the side in the graph, the information that the target arrives at the same place at the same time or has an obvious interactive relation can only be obtained. However, since the target and its associated group may arrive at a certain place at a later time to realize interaction, the accuracy of judgment only through the explicit relationship is low. By adopting the technical scheme, the nodes which meet the condition that the nodes appear in the preset range of the place where the target appears in the preset time range are screened out, so that the population which has a recessive relation with the target is excavated through the knowledge graph, and the accuracy of screening the related population of the target can be improved to a certain extent. The preset time range and the preset range of the place where the target appears are set according to specific conditions, for example, the preset time range is N hours, the preset range of the place where the target appears is within a circle with the place where the target appears as a center and the radius of the circle being M kilometers, and the preset range of the place where the target appears can also be that when the target appears in the third floor of a certain market, the preset range of the place where the target appears is the whole area of the certain market.
S32, classifying the screened nodes into a candidate node set, calling other information of the screened nodes and performing associated storage.
Specifically, nodes within a preset range satisfying a position where the target appears within a preset time range are classified into the candidate node set, and other information of the subject represented by the nodes is called from a database and/or an external source system, wherein the database and/or the external source system can be a public security system, a communication system, a payment system and the like. And correspondingly storing the information of the candidate node set in an associated manner, that is, storing the behavior of the subject corresponding to each node at the corresponding time point and the corresponding place, for example, calling the subject represented by a certain node from a payment system for n hours after the time when the subject is far from the target, and having a payment record at the place where the target appears. After the candidate node set is determined, useful information is extracted and supplemented from the knowledge graph based on node information in the candidate node set and information of corresponding edges, the useful information comprises information of a main body corresponding to the nodes and relevant tracks of the main body, and useless information is eliminated so as to improve the efficiency of subsequent calculation.
S40 calculates the similarity between each node in the candidate node set and the node representing the target.
S41, vectorizing each Node in the candidate Node set through the Node2vec algorithm to obtain a corresponding Node vector.
Specifically, each Node in the candidate Node set is vectorized through a Node2vec algorithm in the prior art to further improve the relationship information between the nodes.
S42 calculates the similarity of the node vectors using a cosine similarity algorithm based on the information of the node vectors.
Specifically, the similarity of the node vectors is calculated by a cosine similarity algorithm, and the nodes are further screened according to the similarity so as to improve the screening precision.
S43 divides the node vectors satisfying the preset same similarity degree range into the same cluster.
Specifically, the node vectors satisfying the same preset similarity range are divided into the same cluster, so that the corresponding node vectors are calculated and compared. The similarity range is set according to actual conditions, the similarity range is divided into a plurality of sub-ranges, and the node vectors meeting the same sub-range are divided into the same cluster.
S44 calculates the modularity of each cluster.
Specifically, the modularity is a measurement parameter used for measuring the partition quality of each cluster in the prior art, and the closer the value of the modularity is to 1, the more reasonable the partition of each cluster is, the better the partition quality is. And (3) by calculating the modularity of each cluster, judging whether the node vector information is reasonable or not in an auxiliary manner, and further improving the precision of the screened associated group.
S45, comparing the modularity of each cluster with the preset threshold value of the modularity, when the modularity of each cluster does not reach the preset threshold value of the modularity, adjusting the node vector and repeating the steps, otherwise, ending the step S40.
Specifically, the modularity of each cluster is compared with a preset threshold of the modularity, when the modularity of each cluster does not reach the preset threshold of the modularity, it indicates that the node vector information is not reasonable, the node vector needs to be adjusted, and steps S41-S45 are repeated, and when the modularity of each cluster reaches the preset threshold of the modularity, step S40 is ended.
Steps S41-S45 are the process of clustering nodes, and when the modularity converges to the threshold of the modularity, the clustering is deemed to be completed. The threshold of the modularity may be set according to the actual application scenario.
S50 determines a population strongly related to the target based on the similarity.
Specifically, the final similarity of any two adjacent node vectors is compared with a preset similarity threshold, when the similarity of the node vectors is lower than the preset similarity threshold, it is judged that the node does not have a relationship with the target, and the node and the connected edge thereof are deleted; otherwise, judging that the node has no relation with the target, and reserving the node and the edge connected with the node, wherein the main body represented by the reserved node is a group strongly related to the target. The similarity threshold can be set according to the actual application scene.
According to an embodiment of the present invention, as shown in fig. 2, the present invention further provides a system for implementing the above-mentioned target relationship detection method based on spatiotemporal association, including:
a construction module for constructing the knowledge-graph based on the information acquired by the acquisition device;
an acquisition module for acquiring information related to the target based on the known characteristics of the target;
the screening module is used for screening out a candidate node set related to the target based on the information related to the target;
a calculation module, configured to calculate a similarity between each node in the candidate node set and the node representing the target;
and the determining module is used for determining the group which is strongly related to the target based on the similarity calculated by the calculating module.
There is also provided, in accordance with an embodiment of the present invention, a storage medium having stored thereon a plurality of instructions adapted to be loaded by a processor to perform a method according to any one of the preceding claims.
The above description is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiments, and all technical solutions belonging to the idea of the present invention belong to the protection scope of the present invention. It should be noted that modifications and embellishments within the scope of the present invention may be made by those skilled in the art without departing from the principle of the present invention, and such modifications and embellishments should also be considered as within the scope of the present invention.
Claims (9)
1. The method for detecting the target relationship based on the space-time association is characterized by comprising the following steps of:
constructing a knowledge graph based on the information acquired in the acquisition region;
inputting known characteristics of the target based on the knowledge graph, and determining a node representing the target;
screening out a candidate node set related to the target based on the nodes representing the target;
calculating the similarity between each node in the candidate node set and the node representing the target;
based on the similarity, a population strongly correlated with the target is determined.
2. The method for detecting the relation of the targets based on the space-time association as claimed in claim 1, wherein the knowledge-graph is constructed based on the information collected in the collection area, further comprising the following steps:
calling information acquired by acquisition equipment preset in the acquisition area;
generating subgraphs of corresponding layers based on the information acquired by the acquisition equipment;
and associating the information of the subgraph to obtain a corresponding knowledge graph.
3. The method for detecting the relation of the targets based on the spatiotemporal correlation as claimed in claim 2, wherein the information related to the targets at least comprises position information and time information of the occurrence of the targets.
4. The method for detecting the object relationship based on the spatiotemporal association as claimed in claim 3, wherein the method for screening out the candidate node set related to the object based on the information related to the object further comprises the following steps:
screening out nodes which meet the condition that the nodes appear in a preset range of the place where the target appears in a preset time range based on the place information and the time information of the target;
and classifying the screened nodes into a candidate node set, calling other information of the screened nodes and performing associated storage.
5. The method for detecting a target relationship based on spatio-temporal correlation according to claim 4, wherein the similarity between each node in the candidate node set and the node representing the target is calculated, further comprising the steps of:
vectorizing each Node in the candidate Node set through a Node2vec algorithm to obtain a corresponding Node vector;
and calculating the similarity of the node vectors by using a cosine similarity algorithm based on the information of the node vectors.
6. The method for detecting the target relationship based on the spatio-temporal correlation according to claim 5, wherein after calculating the similarity of the node vectors by using a cosine similarity algorithm based on the information of the node vectors, the method further comprises the following steps:
dividing node vectors meeting the preset same similarity range into the same cluster;
calculating the modularity of each cluster;
and comparing the modularity of each cluster with a preset modularity threshold, when the modularity of each cluster does not reach the preset modularity threshold, adjusting the node vector and repeating the steps, otherwise, ending the current step.
7. The method for detecting the relation of the targets based on the spatio-temporal correlation as claimed in claim 6, wherein the group strongly related to the targets is determined based on the similarity, further comprising the following steps:
comparing the final similarity of any two adjacent node vectors with a preset similarity threshold, judging that the node is not related to the target when the similarity of the node vectors is lower than the preset similarity threshold, and deleting the node and the edge connected with the node; otherwise, judging that the node has no relation with the target, and reserving the node and the edge connected with the node, wherein the reserved node is a group strongly related to the target.
8. A system for implementing a spatiotemporal correlation-based target relationship detection method according to any one of claims 1 to 7, comprising:
a construction module for constructing the knowledge-graph based on the information acquired by the acquisition device;
an acquisition module for acquiring information related to the target based on the known characteristics of the target;
the screening module is used for screening out a candidate node set related to the target based on the information related to the target;
a calculation module, configured to calculate a similarity between each node in the candidate node set and the node representing the target;
and the determining module is used for determining the group which is strongly related to the target based on the similarity calculated by the calculating module.
9. A storage medium storing a plurality of instructions adapted to be loaded by a processor to perform the method of any one of claims 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010720486.1A CN112052337A (en) | 2020-07-24 | 2020-07-24 | Target relation detection method, system and storage medium based on time-space correlation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010720486.1A CN112052337A (en) | 2020-07-24 | 2020-07-24 | Target relation detection method, system and storage medium based on time-space correlation |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112052337A true CN112052337A (en) | 2020-12-08 |
Family
ID=73601451
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010720486.1A Pending CN112052337A (en) | 2020-07-24 | 2020-07-24 | Target relation detection method, system and storage medium based on time-space correlation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112052337A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113239203A (en) * | 2021-06-02 | 2021-08-10 | 北京金山数字娱乐科技有限公司 | Knowledge graph-based screening method and device |
| CN113256804A (en) * | 2021-06-28 | 2021-08-13 | 湖北亿咖通科技有限公司 | Three-dimensional reconstruction scale recovery method and device, electronic equipment and storage medium |
-
2020
- 2020-07-24 CN CN202010720486.1A patent/CN112052337A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113239203A (en) * | 2021-06-02 | 2021-08-10 | 北京金山数字娱乐科技有限公司 | Knowledge graph-based screening method and device |
| CN113256804A (en) * | 2021-06-28 | 2021-08-13 | 湖北亿咖通科技有限公司 | Three-dimensional reconstruction scale recovery method and device, electronic equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110047095B (en) | Tracking method and device based on target detection and terminal equipment | |
| CN111355697B (en) | Detection method, device, equipment and storage medium for botnet domain name family | |
| CN110807385A (en) | Target detection method and device, electronic equipment and storage medium | |
| CN107665289B (en) | Operator data processing method and system | |
| KR20170100535A (en) | Transaction risk detection method and apparatus | |
| CN109190588A (en) | A kind of method and device of population classification | |
| CN112052337A (en) | Target relation detection method, system and storage medium based on time-space correlation | |
| CN109996245B (en) | Communication resource delivery evaluation method and device, electronic equipment and storage medium | |
| CN111753642B (en) | Method and device for determining key frame | |
| CN110633279A (en) | Multi-source space-time trajectory data processing method, device, equipment and storage medium | |
| CN106530331A (en) | Video monitoring system and method | |
| KR20200123537A (en) | Method for detecting sinkhole using deep learning and data association and sinkhole detecting system using it | |
| CN109947817B (en) | Six-dimensional spatial relationship analysis method and system based on associated graph | |
| CN110891071A (en) | Network traffic information acquisition method, device and related equipment | |
| CN114492590A (en) | Boundary channel generation method and device based on track clustering | |
| CN112070035A (en) | Target tracking method and device based on video stream and storage medium | |
| CN109871456B (en) | Method and device for analyzing relationship between watchmen and electronic equipment | |
| CN111897993A (en) | Efficient target person track generation method based on pedestrian re-recognition | |
| CN111460246A (en) | Real-time activity abnormal person discovery method based on data mining and density detection | |
| CN111325185B (en) | Face fraud prevention method and system | |
| CN114299231A (en) | Construction method of 3D model for river water pollution | |
| CN115187884A (en) | High-altitude parabolic identification method and device, electronic equipment and storage medium | |
| CN112052338A (en) | Method, system and storage medium for constructing knowledge graph | |
| CN111382628B (en) | Method and device for judging peer | |
| CN112182658A (en) | Track data acquisition system and method based on privacy perception |
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 |