CN102508828A - Method for finding path relationship of graph based on multiple agent routes - Google Patents
Method for finding path relationship of graph based on multiple agent routes Download PDFInfo
- Publication number
- CN102508828A CN102508828A CN2011102909458A CN201110290945A CN102508828A CN 102508828 A CN102508828 A CN 102508828A CN 2011102909458 A CN2011102909458 A CN 2011102909458A CN 201110290945 A CN201110290945 A CN 201110290945A CN 102508828 A CN102508828 A CN 102508828A
- Authority
- CN
- China
- Prior art keywords
- semantic
- relation
- subagent
- path
- route
- 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
Abstract
The invention discloses a method for finding a path relationship of a graph based on multiple agent routes. The method comprises the following steps of: firstly, respectively acquiring a semantic relationship graph from each knowledge field by using resource description language and a semantic extracting and editing tool based on W3C standard and taking the semantic relationship graph as a bottom-layer knowledge base for finding the correlation relationship; secondly, constructing a multi-agent distributed route framework through a main agent and a plurality of subagents; thirdly, providing a user interaction interface; fourthly, receiving a hypothesis command of a user and then distributing the hypothesis command to subagent groups for executing a query task by the main agent; and fifthly, receiving the query task and then carrying out route query on a correlation relationship path of the bottom layer knowledge base by the subagent groups and feeding back answers to the user by the main agent after the route query is ended. The method has the advantages that a similar vertex incidence matrix with a weight is established between knowledge from every two different fields; and a distributed route support for querying and demonstrating the interdisciplinary knowledge and the like are provided by adopting a multi-agent technology.
Description
Technical field
The present invention relates to semantic network technology, data mining technology, graph theory are used and intelligent agent technology, relate in particular to the related information of multi-agent technology on extensive grapheme and excavate.
Background technology
One, semantic network technology
The core of semantic net is: through adding the semanteme that can be understood by computing machine to the document on the WWW, thereby make whole internet become a general message exchange media.Semantic web is expanded the ability of WWW through use standard, markup language and relevant handling implement.So-called " semanteme " is exactly the implication of text, and semantic net is exactly the network that can judge according to semanteme, just a kind ofly can understand human language, can make people and interchange between the computer interpersonal interchange of the picture intelligent network easily that becomes.
Except providing a kind of mode to allow the text representation semanteme, another purpose of semantic network technology makes data be convenient to computer more to carry out Intelligent treatment and search exactly.Through tlv triple relationship expression and semantic tagger technology, computing machine can utilize intelligence software, from large-scale semantic relation figure or semantic network, filters out relevant useful information through intelligent agent.
Resource description framework (Resource Description Framework; Be called for short RDF) be a language that is used for expression about the information of the resource on WWW (World Wide Web). it is specifically designed to the metadata of expression about the Web resource; Title, author and modification time such as the Web page; The copyright of Web document and License Info, certain is by available planning chart of shared resource etc.RDF is based on such thought: identify things with Web identifier (being called the unified resource identifier, Uniform Resource Identifier s or URIs), describe resource with simple attribute (property) and property value.This makes RDF can one or more plain statements about resource be expressed as a figure who is made up of node and arc, and node wherein and arc are represented resource, attribute or property value.
Two, many agency's (agent) technology
Along with the foundation of information infrastructure and perfect, people are increasingly high to requiring of using.High intelligent, networked, high reliability and fast adaptability become the target that application system is pursued, and this is the characteristics of many agent system just.So-called many agent system is meant a collective that is made up of the agent of a plurality of autonomous operations, and in the Open Distributed network environment, agent is an abstract entity, and it is autonomous, can take action to self environment, operating environment and environmental change.
Three, based on the clustering method of semantic similarity
Semantic similarity refers to the similarity degree between the semanteme that text comprises, just in the human language of actual environment, and the similarity of the meaning of text indication.Cluster is the method for often using in the data mining, and purpose is to make the set of physics or abstract object be divided into a plurality of types the process of being made up of similar object.Clustering algorithm normally carries out based on similarity, finally can make things convenient for calculation task to carry out, and reduces computation complexity.
Four, graph theory is used
Graph theory is a research object to scheme, and the figure in the graph theory is by some given points and connects 2 the figure that line constituted.This figure is commonly used to describe certain particular kind of relationship between some things, representes things with node, representes to have this relation between corresponding two things with the line that connects at 2.This is the speciality that possessed of the semantic relation figure that finally changes into of semantic network exactly.Therefore, graph-theoretical algorithm for example path search algorithm can be applied to usually in the semantic relation search of semantic net and go.
Summary of the invention
The present invention is directed to the computational problem that needs extensive relation data, overcome the shortcoming that unit is difficult to obtain fast the A to Z of, propose a kind of more based on the figure path relation discover method of acting on behalf of route.
In order to solve the problems of the technologies described above, technical scheme of the present invention is following:
A kind of based on the figure path relation discover method of acting on behalf of route more, comprise the steps:
1) utilize resource description language to obtain semantic relation figure respectively from each ken with semantic extraction and edit tool based on the W3C standard, and with its bottom knowledge base of finding as incidence relation;
2) build the distributed route framework of many agencies through a master agent and a plurality of subagent;
3) the user interactions interface is provided;
4) after master agent reception user's the hypothesis order it is distributed to subagent's group and carry out query task;
5) behind the said query task of subagent's group of received its bottom knowledge base carried out the routing inquiry in incidence relation path, after routing inquiry finishes through master agent to the user feedback answer.
As possibility, said semantic relation figure adopt semantic triple (s, p o) express the semantic knowledge in field, and said s representes main body, said p representation attribute relation, said o representes by the object of main body through the relation on attributes constraint; Semantic similarity between said each ken resource can extract and become similarity matrix; Similarity and its number field scope that the ranks of said similarity matrix are respectively any among the said semantic relation figure at 2 are [0; 1], said semantic similarity is the close degree of said each ken resource indication implication.
As possibility; In said subagent's group; When the routing inquiry that the subagent carries out the incidence relation path to its bottom knowledge base is not enough to accomplish query task; Then feed back local answer and add new hypothesis order to said master agent, said master agent receives said new hypothesis order and it is redistributed the routing inquiry that carries out the incidence relation path to other subagents; But the said process iteration supposes that until the user order is verified correctly or mistake, and said master agent is integrated the local answer of each subagent's feedback.
As possibility, the new hypothesis that said master agent reception user's hypothesis order or subagent add orders the back according to the semantic similarity between each semantic relation figure, and the subagent accepted query task like selection and hypothesis order were nearest.
Beneficial effect of the present invention is:
The first, between different field knowledge, set up the similar incidence matrix that has weights, for cross discipline provides the knowledge foundation.
The second, adopt multi-agent technology,, make different interdisciplinary knowledge be able to that positions different on network is edited and visited for the inquiry and the demonstration of cross-cutting knowledge provides distributed route support.
Three, the user capture interface can adopt the Flex technology for the network user web access interface to be provided, and accepts user's input, and the route results of showing the grapheme incidence relation.
Description of drawings
Fig. 1 is based on the figure path relation discover method implementation framework figure that acts on behalf of route more;
Fig. 2 is a multi-proxy collaboration formula hypothesis verification process flow diagram;
Fig. 3 multi-proxy collaboration formula pathfinding scene.
Embodiment
To combine accompanying drawing and specific embodiment that the present invention is done further explanation below.
1) based on the figure path relation discover method implementation framework of acting on behalf of route more
(Resource Description Framework RDF) obtains semantic relation figure with semantic extraction and edit tool Prot é g é respectively from each ken, promptly through semantic triple (s at first to use the resource description language; P o) expresses the semantic knowledge in field, and is not only the impenetrable text knowledge of machine; Wherein s representes main body; P representation attribute relation, o representes by the object of main body through the relation on attributes constraint, with its bottom knowledge base of finding as incidence relation.As shown in Figure 1, general frame of the present invention is formed by a master agent with by the subagent of master agent control.Master agent through the user capture interface (the user capture interface be system for ease the user propose proposition hypothesis and check the system queries result;, it is used for the output interface of user input and system) accept the proposition hypothesis of user's input; Similarity between this proposition hypothesis and each subagent is mated; Carry out cluster earlier, press packet scheduling again and give each subagent.Semantic similarity is mainly confirmed based on the common frequency that occurs of phrase in the document; In the figure path relation was found, semantic similarity was equivalent to the edge lengths between the node among the figure, and the semantic similarity between each ken resource can extract and become similarity matrix; The ranks of said similarity matrix are respectively any similarity among the said semantic relation figure at 2; Its number field scope is [0,1], and for example " father " and " father " is although there is difference on text; But implication is approximate, and semantic similarity may be defined as 1.Knowledge similarity between each subagent is calculated and is divided into groups to submit to input to carry out prior to the user based on the cluster of edge lengths.Each subagent respectively manages after a part of knowledge; The proposition hypothesis is sent to by master agent and the highest subagent of hypothesis semantic similarity that assigns a topic carries out route; The subagent accepts the laggard walking along the street of proposition hypothesis by inquiry; Semantic relation figure in inside seeks associated pathway, and constantly local result is put into relationship chain.If proposition hypothesis can be acted on behalf of back warp at one and cross route and obtain path validation and promptly can obtain the assign a topic correctness of hypothesis of user, then directly return final associated path to master agent by this subagent; If this proposition hypothesis can't obtain the correctness that final associated path can not verify that promptly the user assigns a topic and supposes an agency inside; Then produce new proposition hypothesis; New proposition hypothesis is submitted in the proposition pond of master agent, to carry out the iteration route, till obtaining final associated path.The user is integrated and returned to hypothesis associated path through a plurality of proxy collaboration routes produce by master agent.
Submit as follows the false code of means of proof to for the false code of method for routing and subagent:
A) route Router (s, c, o ∈ resources, p ∈ graph), Q:task queue
B) subagent submits evidence Replier (h to
c∈ closed hypotheses)
2) multi-proxy collaboration formula hypothesis verification process flow diagram is as shown in Figure 2; Based on the concrete route implementation framework of acting on behalf of more; Multi-proxy collaboration formula hypothesis verification flow process is following: the user propose proposition hypothesis
when master agent will assign a topic hypothesis send to the subagent in organizing act on behalf of A the time; Act on behalf of the routing program of A and find that there is not complete associated path in this agency inside; Submit new proposition hypothesis
master agent to send to act on behalf of B in subagent group to new proposition hypothesis so act on behalf of A to the proposition pond of master agent but there is possible associated path
; Act on behalf of the routing program of B according to the semantic relation figure
of inside produced again the new hypothesis of new proposition hypothesis
then act on behalf of found in the C can be used as evidence local association path
so far; The incidence relation of in the evidence chain, storing is able to form a complete evidence path
thus, and proposition hypothesis
is able to obtain checking.
3) multi-proxy collaboration formula pathfinding application scenarios is as shown in Figure 3, and supposing the system has the domain knowledge of three aspects, Chinese medicine preparation, the liver five-element, disease treatment.The domain knowledge of this three aspect is respectively by different subagent management and safeguard, but for " Wish i knew certain Chinese medicine and the five-element relation " that the user proposes, then needs the cooperation of a plurality of domain knowledges to contribute.Domain knowledge is used RDF tlv triple (s, p, o) definition, for example (irascibility is flourishing, and prescription is Longdan Xiegan Tang), (irascibility is flourishing, and main body is liver), (liver, belong to, wood), (Radix Angelicae Sinensis is returned through, liver) usually.Suppose that the user asks a question " what relation this flavor herbal medicine of Radix Angelicae Sinensis and five-element's element exist? " Such problem can be supposed to carry out at semantic relation figure internal verification via a plurality of agencies; Reach a conclusion at last, Radix Angelicae Sinensis can be treated liver disease, and has certain anonymous relation with element wood.
The above only is a preferred implementation of the present invention; Should be pointed out that for those skilled in the art, under the prerequisite that does not break away from the present invention's design; Can also make some improvement and retouching, these improvement and retouching also should be regarded as in the protection domain of the present invention.
Claims (4)
1. one kind based on the figure path relation discover method of acting on behalf of route more, it is characterized in that, comprises the steps:
1) utilize resource description language to obtain semantic relation figure respectively from each ken with semantic extraction and edit tool based on the W3C standard, and with its bottom knowledge base of finding as incidence relation;
2) build the distributed route framework of many agencies through a master agent and a plurality of subagent;
3) the user interactions interface is provided;
4) after master agent reception user's the hypothesis order it is distributed to subagent's group and carry out query task;
5) behind the said query task of subagent's group of received its bottom knowledge base carried out the routing inquiry in incidence relation path, after routing inquiry finishes through master agent to the user feedback answer.
2. according to claim 1 a kind of based on the figure path relation discover method of acting on behalf of route more; It is characterized in that said semantic relation figure adopts semantic triple (s, p; O) express the semantic knowledge in field; Said s representes main body, said p representation attribute relation, and said o representes by the object of main body through the relation on attributes constraint; Semantic similarity between said each ken resource can extract and become similarity matrix; Similarity and its number field scope that the ranks of said similarity matrix are respectively any among the said semantic relation figure at 2 are [0; 1], said semantic similarity is the close degree of said each ken resource indication implication.
3. according to claim 1 and 2 a kind of based on the figure path relation discover method of acting on behalf of route more; It is characterized in that; In said subagent's group; When the routing inquiry that the subagent carries out the incidence relation path to its bottom knowledge base is not enough to accomplish query task, then feeds back local answer and add new hypothesis order to said master agent, said master agent receives said new hypothesis order and it is redistributed the routing inquiry that carries out the incidence relation path to other subagents; But the said process iteration supposes that until the user order is verified correctly or mistake, and said master agent is integrated the local answer of each subagent's feedback.
4. according to claim 3 a kind of based on the figure path relation discover method of acting on behalf of route more; It is characterized in that; The new hypothesis that said master agent reception user's hypothesis order or subagent add orders the back according to the semantic similarity between each semantic relation figure, and the subagent accepted query task like selection and hypothesis order were nearest.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2011102909458A CN102508828A (en) | 2011-09-16 | 2011-09-16 | Method for finding path relationship of graph based on multiple agent routes |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2011102909458A CN102508828A (en) | 2011-09-16 | 2011-09-16 | Method for finding path relationship of graph based on multiple agent routes |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN102508828A true CN102508828A (en) | 2012-06-20 |
Family
ID=46220919
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2011102909458A Pending CN102508828A (en) | 2011-09-16 | 2011-09-16 | Method for finding path relationship of graph based on multiple agent routes |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102508828A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106354728A (en) * | 2015-07-16 | 2017-01-25 | 富士通株式会社 | Method and device for generating association intensity between objects with semantic graph |
| CN107862037A (en) * | 2017-11-03 | 2018-03-30 | 哈尔滨工业大学 | A kind of event masterplate building method based on entity connected graph |
| CN110413989A (en) * | 2019-06-19 | 2019-11-05 | 北京邮电大学 | A kind of text field based on domain semantics relational graph determines method and system |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20040167862A1 (en) * | 2003-02-05 | 2004-08-26 | Leonid Yabloko | Method and apparatus for mediated cooperation |
| CN101079026A (en) * | 2007-07-02 | 2007-11-28 | 北京百问百答网络技术有限公司 | Text similarity, acceptation similarity calculating method and system and application system |
| CN101140588A (en) * | 2007-10-10 | 2008-03-12 | 华为技术有限公司 | Method and apparatus for ordering incidence relation search result |
| CN102122306A (en) * | 2011-03-28 | 2011-07-13 | 中国人民解放军国防科学技术大学 | Data processing method and distributed file system applying same |
-
2011
- 2011-09-16 CN CN2011102909458A patent/CN102508828A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20040167862A1 (en) * | 2003-02-05 | 2004-08-26 | Leonid Yabloko | Method and apparatus for mediated cooperation |
| CN101079026A (en) * | 2007-07-02 | 2007-11-28 | 北京百问百答网络技术有限公司 | Text similarity, acceptation similarity calculating method and system and application system |
| CN101140588A (en) * | 2007-10-10 | 2008-03-12 | 华为技术有限公司 | Method and apparatus for ordering incidence relation search result |
| CN102122306A (en) * | 2011-03-28 | 2011-07-13 | 中国人民解放军国防科学技术大学 | Data processing method and distributed file system applying same |
Non-Patent Citations (2)
| Title |
|---|
| QINGZHAO ZHENG等: "Collaborative Semantic Association Discovery from Linked Data", 《INFORMATION REUSE & INTEGRATION, 2009. IRI 09. IEEE INTERNATIONAL CONFERENCE ON》, 12 August 2009 (2009-08-12), pages 394 - 399, XP031517249 * |
| 毛郁欣等: "《基于语义的多Agent网构软件模型》", 《情报学报》, vol. 28, no. 5, 31 October 2009 (2009-10-31) * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106354728A (en) * | 2015-07-16 | 2017-01-25 | 富士通株式会社 | Method and device for generating association intensity between objects with semantic graph |
| CN107862037A (en) * | 2017-11-03 | 2018-03-30 | 哈尔滨工业大学 | A kind of event masterplate building method based on entity connected graph |
| CN107862037B (en) * | 2017-11-03 | 2021-07-16 | 哈尔滨工业大学 | Event template construction method based on entity connected graph |
| CN110413989A (en) * | 2019-06-19 | 2019-11-05 | 北京邮电大学 | A kind of text field based on domain semantics relational graph determines method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Das et al. | Social BIMCloud: a distributed cloud-based BIM platform for object-based lifecycle information exchange | |
| US20170139991A1 (en) | Dynamic query plan based on skew | |
| CN103473224B (en) | Problem semantization method based on problem solving process | |
| CN101655862A (en) | Method and device for searching information object | |
| CN111026874A (en) | Data processing method and server of knowledge graph | |
| CN103023714A (en) | Activeness and cluster structure analyzing system and method based on network topics | |
| CN104199893B (en) | A kind of system and method for quickly issuing full media content | |
| CN102508828A (en) | Method for finding path relationship of graph based on multiple agent routes | |
| CN104765763B (en) | A kind of semantic matching method of the Heterogeneous Spatial Information classification of service based on concept lattice | |
| CN106685707A (en) | Asset information control method in distributed infrastructure system | |
| CN101916260A (en) | Method for establishing semantic mapping between disaster body and relational database | |
| KR100912190B1 (en) | System and method for converting rdql query to sql query using optimization transformation rule | |
| McCusker et al. | Functional requirements for information resource provenance on the web | |
| 신지유 et al. | A study on the design of a Topic Map-based retrieval system for the academic administration records of universities | |
| TW578067B (en) | Knowledge graphic system and method based on ontology | |
| Yan et al. | CADRE: A Cloud-Based Data Service for Big Bibliographic Data | |
| CN108763583A (en) | A kind of microblog hot topic extracting method and system based on keyword search | |
| Lili et al. | Semantic ontology method of learning resource based on the approximate subgraph isomorphism | |
| Zhang et al. | CoStorm: a term map system to aid in a collaborative ideation process | |
| El-Gamil et al. | Deep web integrated systems: current achievements and open issues | |
| Lu et al. | Research on the Design of Education Resource Service Cloud Platform of Wide-range District and Metadata Standars | |
| Bandyopadhyay et al. | Automatic Table completion using Knowledge Base | |
| Nie et al. | Design of big data integration platform based on hybrid hierarchy architecture | |
| Tiwari et al. | Semantically ontology merging with intelligent information retrieval using SPARQL on education domain | |
| Henley | Monastic Manuscript Networks of the Anglo-Welsh March: A Study in Literary Transmission |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C12 | Rejection of a patent application after its publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20120620 |