CN103810266A - Semantic network object identification and judgment method - Google Patents
Semantic network object identification and judgment method Download PDFInfo
- Publication number
- CN103810266A CN103810266A CN201410040106.4A CN201410040106A CN103810266A CN 103810266 A CN103810266 A CN 103810266A CN 201410040106 A CN201410040106 A CN 201410040106A CN 103810266 A CN103810266 A CN 103810266A
- Authority
- CN
- China
- Prior art keywords
- msub
- attribute
- evidence
- mrow
- semantic
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 38
- 238000004364 calculation method Methods 0.000 claims abstract description 9
- 230000015572 biosynthetic process Effects 0.000 claims description 26
- 238000003786 synthesis reaction Methods 0.000 claims description 26
- 239000011159 matrix material Substances 0.000 claims description 17
- 238000000605 extraction Methods 0.000 claims description 7
- 239000000284 extract Substances 0.000 claims description 5
- 238000012546 transfer Methods 0.000 claims description 5
- 238000012545 processing Methods 0.000 claims description 3
- 101100272279 Beauveria bassiana Beas gene Proteins 0.000 claims description 2
- 238000010606 normalization Methods 0.000 claims description 2
- 238000013329 compounding Methods 0.000 abstract 2
- 150000001875 compounds Chemical class 0.000 abstract 1
- 230000003993 interaction Effects 0.000 abstract 1
- 230000004927 fusion Effects 0.000 description 10
- 238000011161 development Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 239000008186 active pharmaceutical agent Substances 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000001133 acceleration Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000001364 causal effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000012854 evaluation process Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
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
- G06F16/35—Clustering; Classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/28—Databases characterised by their database models, e.g. relational or object models
- G06F16/284—Relational databases
- G06F16/285—Clustering or classification
Landscapes
- Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Machine Translation (AREA)
Abstract
The invention provides a semantic network object identification and judgment method. The semantic network object identification and judgment method aims to provide a comprehensive judgment method capable of lowering calculated amount of evidence compound and update, lowering confliction among evidences with multiple attributes, and improving consistency of identification results. The method is achieved through the following technical scheme that a target semantic database, an evidence receiving module, an evidence semantic knowledge extracting module and an evidence semantic knowledge extension module are established between evidence receiving and evidence output, wherein data interaction is carried out between the target semantic database and the evidence receiving module, the evidence semantic knowledge extracting module and the evidence semantic knowledge extension module, the evidence receiving module receives object identification evidences from sensor identification sources of different types in real time, an evidence semantic knowledge cluster module carries out clustering on elements in multiple attribute collection related to an expanded attribute constraint relation to obtain a plurality of attributive classifications, an evidence compounding and updating module carries out orthogonal calculation and orthogonal compounding and updating on identification evidences and all attributive classification evidences of different levels of attributes from restricted relationship among multiple attributes to obtain judgment results of updated identification evidences.
Description
Technical Field
The invention relates to a target identification method based on multi-sensor data fusion in the field of target identification and tracking pattern identification, in particular to a comprehensive evidence judgment method of multiple identification sources and multiple attributes under a multi-identification framework of a target comprehensive identification system.
Background
With the development of target identification technology, it is a development trend to comprehensively identify targets by using various types of sensors and various identification technical means. At present, because different types of identification sources independently identify targets, given identification results are often different in attribute level and identification results of the same identification level are often inconsistent in a target identification and tracking system based on multi-sensor (radar and infrared) signal fusion. In order to make the comprehensive identification system have uniform output, the target identification result needs to be provided by various identification means for evidence judgment and decision making. The judgment decision processing comprises single attribute, namely single identification frame judgment and multiple attribute, namely multiple identification frame judgment. However, at present, comprehensive judgment is carried out on multiple attributes of multiple identification sources, so that the conflict among the multiple attribute evidences is difficult to reduce, and the consistency of identification results is improved.
In the prior art, multi-sensor data fusion has gained wide attention and rapid development in recent years as an emerging interdiscipline. The multi-sensor data fusion technology can synthesize the information of each side surface provided by a plurality of sensors, and can obtain more comprehensive and accurate information of an observed object, thereby obtaining accurate and rapid decision and judgment. Object recognition is an important component of data fusion techniques. The definition of the target identification is to make some kind of discrimination on the type or attribute of the target, and the target identification is also called attribute classification or identity estimation. The output data from the multiple target sources can be either dynamic information, i.e., dynamic parameters of the target's motion, typically including position, velocity, and acceleration, or identity information. Because the single sensor system usually only provides partial information for identifying the tracked object, the multi-sensor system utilizes a data fusion technology to synthesize information from different information sources to overcome the defect of the single sensor, and the multi-sensor system utilizes the data complementation and redundancy of different sensors to acquire information from respective independent measuring spaces. Because the target is in continuous motion and the posture is in continuous change, the images of the posture are different, and the difficulty of target identification in a three-dimensional space is greatly increased. Traditional multi-sensor data fusion is performed at the data, feature, and decision levels. In multi-sensor target recognition, the traditional method is to directly send a plurality of local decisions to a fusion center to carry out final overall decision. The prior art methods for judging multiple attributes of multiple identification sources can be roughly divided into three categories: (1) bayesian network inference; (2) reasoning on evidence theory; (3) and (6) judging in a heuristic manner. A Bayesian network (Bayesian network) is a probabilistic network based on probabilistic reasoning mathematical model, which is a graphical network based on probabilistic reasoning, and a Bayesian formula is the basis of this probabilistic network. Probabilistic reasoning is the process of obtaining other probability information through some variable information. The Bayesian network based on the probabilistic reasoning is proposed for solving the problems of uncertainty and incompleteness, has good advantages for solving the faults caused by the uncertainty and relevance of complex equipment, and is widely applied to multiple fields. The bayesian network, also known as belief network, is an uncertainty causal association model. The Bayesian network is different from other decision models, is a probabilistic knowledge expression and inference model for visualizing a multivariate knowledge graph, and is one of the most effective theoretical models in the field of uncertain knowledge expression and inference at present. It is a Directed Acyclic Graph (DAG) consisting of representative variable nodes and Directed edges connecting these nodes. The nodes represent random variables, the directed edges among the nodes represent the mutual correlation system (from father nodes to descendant nodes thereof) among the nodes, the relation strength is expressed by using conditional probability, and the prior probability is used for expressing information without father nodes. Node variables may be abstractions of any problem, such as: test values, observations, opinion polls, etc. The decision making method is applicable to expressing and analyzing uncertain and probabilistic events and is applied to the decision making which is conditionally dependent on various control factors, and can be incomplete. Make inferences about inaccurate or uncertain knowledge or information. The construction of the bayesian network is a complex task, needs the participation of knowledge engineers and domain experts, needs to establish a conditional probability table among various attributes, and represents the constraint relation among different attributes in a quantitative form, and in reality, the quantitative relation is not easy to obtain; meanwhile, bayesian network reasoning cannot solve the "unknown" situation. The evidence theory reasoning is simple for a single identification frame, and has no clear implementation mode for a multi-identification frame of a multi-information fusion identification frame. Although the evidence theory reasoning DS method is widely applied to various data fusion systems, due to the computational complexity of a core Dempster synthesis rule of the DS method, the realization of the algorithm is a difficult problem. Heuristic judgment is mainly based on constraint knowledge among attributes, and according to a certain rule, such as scoring, reliability of the attributes is updated, so that judgment decision is made. In addition, the existing evidence judging method does not well utilize semantic knowledge among target attributes, does not have clear mathematical realization of multiple attribute evidence judgment, has poor reliability and can not ensure the reasonability of results.
Disclosure of Invention
The invention aims at the defects of the comprehensive information of different information sources and the comprehensive identification system of multiple attributes of a multi-target identification source, and provides a comprehensive evidence judgment method which is more reasonable in evidence judgment, can reduce the calculation amount of evidence synthesis and update, and can be used for judging multiple attributes of target attributes, types, models and the like from multiple identification sources containing 'unknown' statements so as to reduce the conflict among the multiple attribute evidences and improve the consistency of identification results.
The scheme adopted by the invention for solving the problems in the prior art is as follows: a semantic network target identification evidence judgment method has the following technical characteristics: the evidence collection and evidence output method comprises the steps of establishing a target semantic library for storing various target entities, semantic membership knowledge among multiple attributes and attribute constraint relations of different levels of targets between evidence receiving and evidence output, carrying out classification processing on the attributes related to the received evidence and interacting data with the target semantic library, and further comprising an evidence receiving module, an evidence semantic knowledge extraction module, an evidence semantic knowledge expansion module and an evidence output module for outputting evidence judgment data results sequentially through an evidence semantic knowledge clustering module and an evidence synthesis and update module; the evidence receiving module receives target identification evidence from different types of sensor identification sources in real time, wherein the target identification evidence comprises target attributes, target types, target models, identification statements of one or more attributes in a target attribute hierarchy and/or included 'unknown' identification evidence statements; the evidence semantic knowledge extraction module extracts semantic knowledge related to the identification evidence statement received in the evidence receiving module from the membership relation among various target attributes stored in the target semantic library; the evidence semantic knowledge expansion module expands the extracted semantic knowledge according to the transfer rule of the membership between the target attributes to obtain an expanded attribute support constraint relation and an attribute conflict constraint relation; the evidence semantic knowledge clustering module clusters elements in a multiple attribute set S related to the expanded attribute constraint relationship to obtain a plurality of attribute classifications, and attributes in each classification are minimum attribute sets with mutual constraint relationship; and the evidence output module outputs the updated identification evidence judgment result to other modules for calling the method.
Compared with the prior art, the invention has the following beneficial effects.
The invention introduces the following steps between evidence receiving and evidence outputting: the evidence evaluation system comprises a target semantic library, an evidence receiving module, an evidence semantic knowledge extracting module, an evidence semantic knowledge expanding module, an evidence semantic knowledge clustering module, an evidence synthesis updating module and an evidence output module, wherein the constraint relation between different levels of attributes of a target described by the target semantic library is used as the basis of the evidence evaluation process, so that the evidence evaluation is more reasonable. The target semantic library is independently stored in the evidence judging process, so that the maintenance is facilitated; the evidence semantic knowledge extraction module, the evidence semantic knowledge expansion module and the evidence semantic knowledge clustering module classify the attributes related to the received evidence, so that the calculation amount of evidence synthesis updating is reduced; the evidence synthesis updating module starts from the constraint relation among the multiple attributes globally and obtains a globally superior evidence judgment result by performing strict orthogonal calculation, orthogonal synthesis and other processes on the identification evidences of the attributes at different levels; the participation of the unknown attributes in the multiple types of attributes in judgment makes the description of the judgment result more reasonable.
The invention adopts a target semantic library, an evidence receiving module for interacting data with the target semantic library, an evidence semantic knowledge extraction module and an evidence semantic knowledge expansion module to classify the attributes related to the received evidence and comprehensively judge multiple identification sources and multiple attribute evidences of target attributes, types, models and the like which are derived from multiple identification sources and contain unknown statements, thereby further reducing the conflict among the multiple attribute evidences and improving the consistency of the identification results.
Drawings
In order that the invention may be more clearly understood, it will now be described by way of example with reference to the accompanying drawings, in which:
FIG. 1 is a schematic diagram of semantic network object recognition evidence judgment principle of the present invention.
FIG. 2 is a flow diagram of an evidence semantic knowledge clustering module to which the present invention relates.
FIG. 3 is a flow diagram of an evidence composition update module in accordance with the present invention.
Detailed Description
See fig. 1. In the semantic Web object recognition evidence embodiment described below, there is included between evidence reception and evidence output: the system comprises a target semantic library, an evidence receiving module, an evidence semantic knowledge extracting module, an evidence semantic knowledge expanding module, an evidence semantic knowledge clustering module, an evidence synthesis updating module and an evidence output module. Semantic membership knowledge about multiple attributes of various target entities is stored in a target semantic library; the evidence receiving module receives target identification evidence which is sourced from different types of sensor identification sources and can contain multiple attributes of 'unknown' declarations in real time; the evidence semantic knowledge extraction module extracts semantic knowledge related to the received identification evidence statement from the membership between various target attributes stored in the target semantic library; the evidence semantic knowledge expansion module expands the extracted semantic knowledge according to the transfer rule of the membership between the target attributes to obtain an expanded attribute support constraint relation and an attribute conflict constraint relation; clustering elements in an attribute type set related to the expanded attribute constraint relationship by an evidence semantic knowledge clustering module to obtain a plurality of attribute classifications, wherein the attributes in each classification are minimum attribute sets with mutual constraint relationship; the evidence synthesis updating module performs orthogonal calculation and orthogonal synthesis updating on the identification evidence and each attribute classification evidence of different levels of attributes from the constraint relation among the multiple attributes to obtain an updated identification evidence judgment result; and the evidence output module outputs the updated identification evidence judgment result. The multiple attribute set is S = { model, size type, platform type, environment type, attribute, nationality, … }; the specific implementation steps are as follows: in step S0, the target semantic library stores semantic descriptions of different targets about each object included in each level of target attributes, such as model, size type, platform type, and attribute. Taking an object "F16" contained in the target model as an example, the semantic membership knowledge is described as: the small-sized aerial target takes an object contained in a platform type as an example, and the semantic membership knowledge of the small-sized aerial target is described as follows: fixed wing aircraft, small aerial targets. For descriptive convenience, the constraint relationship between different target attribute objects is abbreviated as R [ (T)1,t1),r,(T2,t2)]. Wherein R represents a piece of constraint knowledge; (T)1,t1)、(T2,t2) Respectively represent attributes T1Object t of1And an attribute T2Object t of2R represents (T)1，t1) And (T)2,t2) The relationship between them has three relationships of "support", "conflict" and "do not declare", which are respectively marked as rs、rc、ru. Wherein, T1、T2Respectively as target attributes; t is t1、t2Respectively target attribute T1The corresponding attribute object and target attribute T2The corresponding attribute object.
In step S1, the evidence receiving module receives in real time the multiple-attribute target identification evidence which may include the "unknown" statement and is derived from different types of sensor identification sources such as radar, photoelectric, electronic reconnaissance, recognizer, etc., and stores the multiple-attribute target identification evidence in the cache.
In step S2, the evidence semantic knowledge extracting module extracts the multiple attribute recognition evidence currently captured and received in the cache, and then extracts semantic knowledge related to the received recognition evidence in the membership between various target attributes stored in the target semantic library.
In step S3, the evidence semantic knowledge expansion module expands the extracted semantic knowledge according to the transfer rule of the membership between the target attributes to obtain an expanded attribute support constraint relationship and an attribute conflict constraint relationship; the transfer rule of the membership between the involved target attributes is as follows:
[ support for extended rules ]: if R [ (T)1，t1)，rs，(T2,t2)]And R [ (T)2,t2),rs,(T3,t3)]Then R [ (T)1，t1),rs，(T3,t3)]；
[ conflict extension rules ]: if R [ (T)1，t1)，rs，(T2,t2)]And R [ (T)2，t2),rc，(T3,t3)]Then R [ (T)1，t1)，rc,(T3,t3)]。
In step S4, the evidence semantic knowledge clustering module clusters the elements in the attribute type set related to the expanded attribute constraint relationship.
See fig. 2. In step S41, the clustering module initializes the clustering result, and takes each target attribute as a cluster; the clustering module traverses all current clustering results in the step S42, selects one clustering pair, uses one clustering pair as a reference class and uses the other clustering pair as a class to be compared, and the clustering module judges whether a constraint relation exists between the object related to the reference class and the object related to the class to be compared in the clustering pair selected in the step S43. If the constraint relation exists, the step S44 is entered, and the clustering module merges the attributes of the classes to be compared into the reference class; otherwise, the clustering module proceeds to step S45 to determine whether all the clustering pairs formed in the existing clusters have been traversed. If the traversal is finished, the step S46 is entered, and the clustering module stores the obtained clustering result in a cache; otherwise, go to step S42;
see fig. 3. In step S5, the evidence synthesis and update module performs synthesis and update on the confidence of each attribute classification from each cluster in the cached clustering results by performing processes such as orthogonal operation and orthogonal synthesis on the input identification evidence in combination with the evidence related to the attribute set corresponding to the cluster, so as to obtain an updated identification evidence. In step S51, the evidence synthesis update module selects an unprocessed cluster from the attribute clusters in the cache, and performs the following steps for the attributes involved in the cluster: performing orthogonal calculation of the attributes in the cluster selected in step S51 by step S52, the calculation method including: selected cluster CqContains m attributes, the ith attribute contains niA number of definite variables, in total, with 1 "unknown" variable (n)i+1) attribute objects. J-th attribute assigned by input evidenceConfidence of each object is recorded as
Computing elements in an orthogonal product matrix
In step S53, the evidence synthesis update module normalizes each element in the orthogonal matrix to obtain a normalized matrix, and calculates a normalized value in the normalized matrix
In the formula,
are the orthogonal elements in the orthogonal matrix obtained by step S52.
In step S54, the evidence synthesis updating module performs marginalization on the orthogonal elements in the normalized orthogonal matrix to obtain confidence values of the updated attribute objects
In step S55, the evidence synthesis updating module estimates, according to the updated evidence of each attribute, the object reliability values included in each attribute by using the following formula; jth variable object for ith attribute
Representing an "unknown" variable.Is estimated as
In step S56, for each class of attributes in the cluster, determining the maximum confidence value of all objects thereof; the maximum value of the credibility of all the objects of the ith attribute is
And determine
The storage sequence number j of the corresponding object in all the objects of the attribute*(ii) a Given a threshold value gammathIf, if
Then will be
As a final decision of the ith attribute; otherwise will t0I.e. "unknown" as the final decision for the ith attribute. Wherein,
for the normalized values of the elements in the normalization matrix obtained in step S53,
the updated confidence level of the jth object corresponding to the ith attribute,
the jth object representing the kth attribute,
j object, n, representing the ith attributeiThe number of objects other than "unknown" for the ith attribute,
the trustworthiness of the jth object showing the ith attribute,
representing the maximum value of confidence, gamma, for all objects in the ith attributethIndicating a decision threshold.
In step S57, it is determined whether the evidence synthesis update module has processed all the attributes in a cluster, and if not, the process returns to step S51; otherwise, step S58 is executed to store the confidence updating of all the clustered attribute objects at each level and the attribute judgment results at each level into the designated cache area.
In step S6, the evidence output module outputs the attribute judgment result saved in the buffer in step S58 to the network for reception by other devices.
What has been described above is merely a preferred embodiment of the invention. It should be noted that variations and modifications can be made by those skilled in the art without departing from the principle of the present invention, and these variations and modifications should be construed as falling within the scope of the present invention.
Claims (10)
1. A semantic network target identification evidence judgment method has the following technical characteristics: the evidence collection and evidence output method comprises the steps of establishing a target semantic library for storing various target entities, semantic membership knowledge among multiple attributes and attribute constraint relations of different levels of targets between evidence receiving and evidence output, carrying out classification processing on the attributes related to the received evidence and interacting data with the target semantic library, and further comprising an evidence receiving module, an evidence semantic knowledge extraction module, an evidence semantic knowledge expansion module and an evidence output module for outputting evidence judgment data results sequentially through an evidence semantic knowledge clustering module and an evidence synthesis and update module; the evidence receiving module receives target identification evidence from different types of sensor identification sources in real time, wherein the target identification evidence comprises target attributes, target types, target models, identification statements of one or more attributes in a target attribute hierarchy and/or included 'unknown' identification evidence statements; the evidence semantic knowledge extraction module extracts semantic knowledge related to the identification evidence statement received in the evidence receiving module from the membership relation among various target attributes stored in the target semantic library; the evidence semantic knowledge expansion module expands the extracted semantic knowledge according to the transfer rule of the membership between the target attributes to obtain an expanded attribute support constraint relation and an attribute conflict constraint relation; the evidence semantic knowledge clustering module clusters elements in a multiple attribute set S related to the expanded attribute constraint relationship to obtain a plurality of attribute classifications, and attributes in each classification are minimum attribute sets with mutual constraint relationship; and the evidence output module outputs the updated identification evidence judgment result.
2. The semantic network object recognition forensic method according to claim 1, wherein: the multiple attribute set is S = { model, size type, platform type, environment type, attribute, nationality, … }.
3. The semantic web object recognition forensic method according to claim 1, wherein: the target semantic library stores semantic descriptions of different targets about model, size type, platform type, attribute and each object contained in each level of target attribute.
4. The semantic network object recognition forensic method according to claim 1, wherein: the constraint relationship between different target attribute objects is R [ (T)1，t1)，r，（T2，t2)]Wherein R represents a piece of constraint knowledge; (T)1,t1)、(T2,t2) Respectively represent attributes T1Object t of1And an attribute T2Object t of2(ii) a r represents (T)1,t1) And (T)2,t2) The relation between the two has three relations of 'support', 'conflict' and 'no declaration', wherein T1、T2Respectively as target attributes; t is t1、t2Respectively target attribute T1The corresponding attribute object and target attribute T2The corresponding attribute object.
5. The semantic network object recognition forensic method according to claim 1, wherein: the evidence semantic knowledge clustering module clusters elements in an attribute type set related to the expanded attribute constraint relationship, initializes clustering results, respectively uses each target attribute as a cluster, traverses all current clustering results, selects a cluster pair, uses one of the cluster pairs as a reference class and uses the other cluster pair as a class to be compared, judges whether a constraint relationship exists between an object related to the reference class and an object related to the class to be compared in the selected cluster pair, and if the constraint relationship exists, merges the attributes in the class to be compared into the reference class; otherwise, judging whether all the cluster pairs formed in the existing clusters are traversed; and if the traversal is finished, storing the obtained clustering result in a cache.
6. The semantic network object recognition forensic method according to claim 1, wherein: and the evidence synthesis updating module performs orthogonal operation and orthogonal synthesis on the input identification evidence by combining each cluster in the cluster results in the cache and the evidence related to the attribute set corresponding to the cluster, performs synthesis updating on the confidence coefficient of each attribute classification, and obtains the updated identification evidence.
7. The language of claim 6The network object identification and evidence judgment method is characterized by comprising the following steps: the evidence synthesis updating module selects an unprocessed cluster from the attribute clusters in the cache, and carries out orthogonal calculation on the attributes in the selected cluster according to the attributes involved in the cluster, wherein the calculation method comprises the following steps: selected cluster CqContains m attributes, the ith attribute contains niA number of definite variables, in total, with 1 "unknown" variable (n)i+1) attribute objects, and the confidence degree assigned by the input evidence about the jth object corresponding to the ith attribute is recorded as
Computing elements in an orthogonal product matrix
8. The semantic network object recognition forensic method according to claim 7 wherein: elements in an orthogonal product matrix
Wherein m is the number of attributes; j is a function of1、j2、jmRespectively represent the j-th corresponding to the 1 st attribute1The j (j) th object corresponding to the 2 nd attribute2J-th object corresponding to m-th attributemAn object;
the confidence of the jth object corresponding to the ith attribute;
is the orthogonal element in the orthogonal matrix obtained by step S52;
taking a value of 0 or 1 for the orthogonal product validity flag, wherein the determination method comprises the following steps: if p e [1, m ] is present]And q ∈ [1, m ]]Having the relationship R [ ()
]Then, then Otherwise
Wherein, TpFor the p-th attribute, the attribute,
is the j-th attribute in the p-th attributepAn attribute object; r iscIndicating a "conflict" relationship.
9. The semantic network object recognition forensic method according to claim 8 wherein: the evidence synthesis updating module normalizes each element in the orthogonal matrix to obtain a normalized matrix, and calculates a normalized value in the normalized matrix
In the formula,
is an orthogonal element, j, in the orthogonal matrix obtained by step S521、j2、jmRespectively represent the j-th corresponding to the 1 st attribute1The j (j) th object corresponding to the 2 nd attribute2J-th object corresponding to m-th attributemAn object.
10. The semantic network object recognition forensic method according to claim 9 wherein: the evidence synthesis updating module marginalizes orthogonal elements in the normalized orthogonal matrix to obtain confidence values of the updated attribute objects
The evidence synthesis updating module estimates the credibility values of all objects contained in all attributes by using the following formula according to the updated evidences of all attributes; jth variable object for ith attribute
Represents an "unknown" variable;
For each type of attribute in the cluster, determining the maximum value of the credibility of all objects of the attribute; the maximum value of the credibility of all the objects in the ith attribute is
And determine
The storage sequence number j of the corresponding object in all the objects of the attribute*Given a threshold value gammath) If it is
Then will be
As a final decision of the ith attribute; otherwise will t0Namely 'unknown' as the final judgment of the ith attribute; wherein,
for the normalized values of the elements in the normalization matrix obtained in step S53,
the updated confidence level of the jth object corresponding to the ith attribute,the jth object representing the kth attribute,j object, n, representing the ith attributeiThe number of objects other than "unknown" for the ith attribute,
the trustworthiness of the jth object showing the ith attribute,
representing the maximum value of confidence, gamma, for all objects in the ith attributethIndicating a decision threshold.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410040106.4A CN103810266B (en) | 2014-01-27 | 2014-01-27 | Semantic network target recognition sentences card method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410040106.4A CN103810266B (en) | 2014-01-27 | 2014-01-27 | Semantic network target recognition sentences card method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103810266A true CN103810266A (en) | 2014-05-21 |
| CN103810266B CN103810266B (en) | 2017-04-05 |
Family
ID=50707036
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410040106.4A Active CN103810266B (en) | 2014-01-27 | 2014-01-27 | Semantic network target recognition sentences card method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103810266B (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106326928A (en) * | 2016-08-24 | 2017-01-11 | 四川九洲电器集团有限责任公司 | Target identification method and equipment |
| CN106650785A (en) * | 2016-11-09 | 2017-05-10 | 河南大学 | Weighted evidence fusion method based on evidence classification and conflict measurement |
| CN106970957A (en) * | 2017-03-17 | 2017-07-21 | 福州大学 | A kind of digital evidence chain overall analysis system and method |
| CN108256401A (en) * | 2016-12-29 | 2018-07-06 | 杭州海康威视数字技术股份有限公司 | A kind of method and device for obtaining objective attribute target attribute Feature Semantics |
| CN108351971A (en) * | 2015-10-12 | 2018-07-31 | 北京市商汤科技开发有限公司 | The method and system that the object that label has is clustered |
| CN109189848A (en) * | 2018-09-19 | 2019-01-11 | 平安科技（深圳）有限公司 | Abstracting method, system, computer equipment and the storage medium of knowledge data |
| CN110110089A (en) * | 2018-01-09 | 2019-08-09 | 网智天元科技集团股份有限公司 | Cultural relations drawing generating method and system |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1137320A (en) * | 1993-10-29 | 1996-12-04 | 瓦尔数据公司 | Semantic object modeling system for creating relational database schemas |
| US20040068697A1 (en) * | 2002-10-03 | 2004-04-08 | Georges Harik | Method and apparatus for characterizing documents based on clusters of related words |
| CN101986296A (en) * | 2010-10-28 | 2011-03-16 | 浙江大学 | Noise data cleaning method based on semantic ontology |
| CN102884779A (en) * | 2010-02-24 | 2013-01-16 | 数字标记公司 | Intuitive computing methods and systems |
| CN103500208A (en) * | 2013-09-30 | 2014-01-08 | 中国科学院自动化研究所 | Deep layer data processing method and system combined with knowledge base |
-
2014
- 2014-01-27 CN CN201410040106.4A patent/CN103810266B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1137320A (en) * | 1993-10-29 | 1996-12-04 | 瓦尔数据公司 | Semantic object modeling system for creating relational database schemas |
| US20040068697A1 (en) * | 2002-10-03 | 2004-04-08 | Georges Harik | Method and apparatus for characterizing documents based on clusters of related words |
| CN102884779A (en) * | 2010-02-24 | 2013-01-16 | 数字标记公司 | Intuitive computing methods and systems |
| CN101986296A (en) * | 2010-10-28 | 2011-03-16 | 浙江大学 | Noise data cleaning method based on semantic ontology |
| CN103500208A (en) * | 2013-09-30 | 2014-01-08 | 中国科学院自动化研究所 | Deep layer data processing method and system combined with knowledge base |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108351971B (en) * | 2015-10-12 | 2022-04-22 | 北京市商汤科技开发有限公司 | Method and system for clustering objects marked with attributes |
| CN108351971A (en) * | 2015-10-12 | 2018-07-31 | 北京市商汤科技开发有限公司 | The method and system that the object that label has is clustered |
| CN106326928A (en) * | 2016-08-24 | 2017-01-11 | 四川九洲电器集团有限责任公司 | Target identification method and equipment |
| CN106326928B (en) * | 2016-08-24 | 2020-01-07 | 四川九洲电器集团有限责任公司 | Target identification method and device |
| CN106650785B (en) * | 2016-11-09 | 2019-05-03 | 河南大学 | A Weighted Evidence Fusion Method Based on Evidence Classification and Conflict Measurement |
| CN106650785A (en) * | 2016-11-09 | 2017-05-10 | 河南大学 | Weighted evidence fusion method based on evidence classification and conflict measurement |
| CN108256401A (en) * | 2016-12-29 | 2018-07-06 | 杭州海康威视数字技术股份有限公司 | A kind of method and device for obtaining objective attribute target attribute Feature Semantics |
| CN108256401B (en) * | 2016-12-29 | 2021-03-26 | 杭州海康威视数字技术股份有限公司 | Method and device for obtaining target attribute feature semantics |
| CN106970957B (en) * | 2017-03-17 | 2020-01-14 | 福州大学 | Digital evidence chain comprehensive analysis system and method |
| CN106970957A (en) * | 2017-03-17 | 2017-07-21 | 福州大学 | A kind of digital evidence chain overall analysis system and method |
| CN110110089A (en) * | 2018-01-09 | 2019-08-09 | 网智天元科技集团股份有限公司 | Cultural relations drawing generating method and system |
| CN110110089B (en) * | 2018-01-09 | 2021-03-30 | 网智天元科技集团股份有限公司 | Cultural relation graph generation method and system |
| CN109189848A (en) * | 2018-09-19 | 2019-01-11 | 平安科技（深圳）有限公司 | Abstracting method, system, computer equipment and the storage medium of knowledge data |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103810266B (en) | 2017-04-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Feng et al. | A review and comparative study on probabilistic object detection in autonomous driving | |
| CN103810266B (en) | Semantic network target recognition sentences card method | |
| US11341424B2 (en) | Method, apparatus and system for estimating causality among observed variables | |
| US10073887B2 (en) | System and method for performing k-nearest neighbor search based on minimax distance measure and efficient outlier detection | |
| Shao et al. | A novel outlier detection method for monitoring data in dam engineering | |
| CN110348437B (en) | A Target Detection Method Based on Weakly Supervised Learning and Occlusion Awareness | |
| JP2012194967A (en) | Finding of abnormal pattern | |
| CN106682502A (en) | Intrusion intension recognition system and method based on hidden markov and probability inference | |
| Zhang et al. | Energy theft detection in an edge data center using threshold-based abnormality detector | |
| CN112884742A (en) | Multi-algorithm fusion-based multi-target real-time detection, identification and tracking method | |
| CN104717468B (en) | Cluster scene intelligent monitoring method and system based on the classification of cluster track | |
| CN112116019B (en) | Multi-sensor Vine Copula heterogeneous information decision fusion method | |
| Pang et al. | Anomaly detection based on data stream monitoring and prediction with improved Gaussian process regression algorithm | |
| CN108447057A (en) | SAR image change detection based on conspicuousness and depth convolutional network | |
| CN106203521A (en) | Based on disparity map from the SAR image change detection of step study | |
| Ye et al. | A deep learning-based method for automatic abnormal data detection: Case study for bridge structural health monitoring | |
| KR20220047228A (en) | Method and apparatus for generating image classification model, electronic device, storage medium, computer program, roadside device and cloud control platform | |
| Gligorijevic et al. | Uncertainty propagation in long-term structured regression on evolving networks | |
| Hong et al. | Wildfire detection via transfer learning: a survey | |
| CN114064928A (en) | Knowledge inference method, knowledge inference device, knowledge inference equipment and storage medium | |
| US8301584B2 (en) | System and method for adaptive pruning | |
| Li et al. | Structural health monitoring data anomaly detection by transformer enhanced densely connected neural networks | |
| CN113408663B (en) | Fusion model construction method, fusion model using device and electronic equipment | |
| CN117612277A (en) | Unsupervised Internet of Vehicles anomaly detection method and system based on physically invariant subspace | |
| CN117195969A (en) | Ship track prediction method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |