CN112085458A - Cloud manufacturing platform resource service matching method - Google Patents
Cloud manufacturing platform resource service matching method Download PDFInfo
- Publication number
- CN112085458A CN112085458A CN202010699532.4A CN202010699532A CN112085458A CN 112085458 A CN112085458 A CN 112085458A CN 202010699532 A CN202010699532 A CN 202010699532A CN 112085458 A CN112085458 A CN 112085458A
- Authority
- CN
- China
- Prior art keywords
- service
- resource
- information
- similarity
- matching
- 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
- 238000004519 manufacturing process Methods 0.000 title claims abstract description 72
- 238000000034 method Methods 0.000 title claims abstract description 43
- 238000013507 mapping Methods 0.000 claims abstract description 30
- 238000007781 pre-processing Methods 0.000 claims abstract description 21
- 238000004364 calculation method Methods 0.000 claims description 81
- 239000000126 substance Substances 0.000 claims description 20
- 230000008569 process Effects 0.000 claims description 12
- 238000012545 processing Methods 0.000 claims description 12
- 238000010606 normalization Methods 0.000 claims description 11
- 239000011159 matrix material Substances 0.000 claims description 9
- 230000000694 effects Effects 0.000 claims description 8
- 238000011156 evaluation Methods 0.000 claims description 8
- 238000005457 optimization Methods 0.000 claims description 8
- 238000007514 turning Methods 0.000 claims description 6
- 239000002131 composite material Substances 0.000 claims description 5
- 238000009826 distribution Methods 0.000 claims description 5
- 230000004931 aggregating effect Effects 0.000 claims description 3
- 230000001419 dependent effect Effects 0.000 claims description 3
- 238000012163 sequencing technique Methods 0.000 claims description 3
- 241000287196 Asthenes Species 0.000 claims description 2
- 230000009286 beneficial effect Effects 0.000 abstract description 2
- 238000010586 diagram Methods 0.000 description 8
- 238000005516 engineering process Methods 0.000 description 8
- 238000003754 machining Methods 0.000 description 8
- 238000011161 development Methods 0.000 description 5
- 230000000877 morphologic effect Effects 0.000 description 4
- 238000011160 research Methods 0.000 description 4
- 239000013598 vector Substances 0.000 description 4
- 241000764238 Isis Species 0.000 description 3
- 239000008186 active pharmaceutical agent Substances 0.000 description 3
- 101100518501 Mus musculus Spp1 gene Proteins 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 238000005553 drilling Methods 0.000 description 2
- 238000010438 heat treatment Methods 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 238000005192 partition Methods 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 206010063385 Intellectualisation Diseases 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000000052 comparative effect Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000003064 k means clustering Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/10—Office automation; Time management
- G06Q10/103—Workflow collaboration or project management
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/04—Manufacturing
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Entrepreneurship & Innovation (AREA)
- General Business, Economics & Management (AREA)
- Economics (AREA)
- Tourism & Hospitality (AREA)
- Marketing (AREA)
- Manufacturing & Machinery (AREA)
- Primary Health Care (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Probability & Statistics with Applications (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
A cloud manufacturing platform resource service matching method comprises a model establishing step, a preprocessing step, a mapping matching step and an optimizing step, and is characterized in that the model establishing step, the preprocessing step, the mapping matching step and the optimizing step are sequentially carried out. The beneficial effects are as follows: the matching method has high matching efficiency while keeping high accuracy, can greatly reduce the matching range of resource services through resource service preprocessing, reduces the operation scale and has high user satisfaction.
Description
Technical Field
The invention relates to the technical field of cloud manufacturing, in particular to a cloud manufacturing platform resource service matching method.
Background
The manufacturing industry plays a crucial role in the economic development of various countries, and directly determines the economic level and comprehensive national strength of one country. With the rapid development of science and technology and the global integration of economy, the international manufacturing industry is developing towards the direction of greening, intellectualization, servitization and informatization, and in order to adapt to the international development trend, different manufacturing industry transformation planning targets are provided for each country. As a large country of manufacturing industry, the manufacturing industry in China also faces the problems of backward production mode, low utilization rate of manufacturing resources, insufficient innovation capacity and the like, and in order to solve the problems, the traditional manufacturing industry needs to be continuously integrated with emerging technologies such as cloud computing, Internet of things, intelligent science and the like to form a new intelligent manufacturing mode, and the manufacturing industry is continuously promoted to be developed to high-end. Cloud manufacturing is firstly proposed in 2010, as an advanced manufacturing concept and an advanced manufacturing mode, advanced technologies such as cloud computing, the Internet of things and big data are fused, and the problems of manufacturing resource waste and scheduling imbalance in China can be effectively solved. The development and popularization of the cloud manufacturing technology have profound significance for transformation and upgrade of the manufacturing industry in China, and the efficient matching technology of resource service and demand in the cloud platform is one of the key technologies for cloud manufacturing and popularization.
At present, research on a manufacturing resource virtualization technology and a resource service matching method in a cloud environment has achieved certain achievements, but some problems still exist:
the manufacturing resources are various in types and complicated in content, various resource attributes are different and are connected, most of the existing research focuses on description of a single resource, and the existing research has no universality and lacks a uniform and effective resource description model;
in the aspect of resource service matching, the research mostly focuses on the recall ratio and precision ratio of matching, but neglects the matching efficiency of resource requirements, and with the development of the cloud manufacturing technology, the number of resource services in the cloud manufacturing platform will be more and more, and the matching between the resource services and the requirements needs to be realized quickly and accurately.
Disclosure of Invention
The invention aims to solve the problems and designs a cloud manufacturing platform resource service matching method. The specific design scheme is as follows:
a cloud manufacturing platform resource service matching method comprises a model establishing step, a preprocessing step, a mapping matching step and an optimizing step, and is characterized in that the model establishing step, the preprocessing step, the mapping matching step and the optimizing step are sequentially carried out,
in the step of establishing the model, the model is established,
establishing a formalized description model of a cloud manufacturing platform resource provider, namely a provider model:
RS={ID,BI,ST,FI,SI},
wherein the content of the first and second substances,
the ID is an identification code of the cloud manufacturing resource, cannot be changed and does not participate in matching, and is mainly used for positioning and searching the cloud manufacturing resource;
the BI indicates Basic Information (Basic Information) of a resource provider, the Basic Information mainly comprises a brief introduction of a resource service, a name of the resource service, a contact way of the resource provider, an address of the resource provider and the like, wherein the brief introduction of the resource service is main Information which is described by adopting a simple text and participates in a matching process of the resource service, so that the filling of the brief introduction of the resource service emphasizes the characteristics of the resource;
ST represents the State information (State information) of the resource provider, and the State information of the resource service is set by the resource provider, and comprises an idle State, a non-full-load State, a full-load State and an invalid State;
FI denotes function Information (Functional Information) of the resource provider, which is Information that the resource demander pays attention to when searching for the resource service;
SI represents Service Information (Service Information) of a resource provider, and the Service Information refers to various performance indexes required by the resource provider when providing a Service;
establishing a formalized description model of a resource demander of a cloud manufacturing platform, namely a demander model:
RD={ID,DBI,DST,DFI,DSI},
wherein the content of the first and second substances,
DBI represents Basic Information (Demand Basic Information) of a resource demander;
the DST represents State information (Demand State information) of the resource Demand side, including an unexecuted State, an executing State, a completed State and a failure Demand State;
the DFI represents function Information (Demand function Information) of the resource demander;
the DSI represents Service attribute Information (Demand Service Information) of the resource demander,
in the preprocessing step, resource services similar to the text description of the basic information BI are aggregated into service clusters, and the service cluster most relevant to the basic information of the demand side is selected as a candidate resource service set so as to narrow the matching range, wherein the similarity of the text description comprises word form similarity, word order similarity, sentence length similarity and comprehensive similarity,
in the mapping and matching step, the state information ST, the function information FI and the service information SI of the alternative resource service selected by the pretreatment are respectively matched with the state information DST, the function information DFI and the service information DSI of the resource demand party, corresponding matching threshold values are set, the resource service meeting the matching threshold values is reserved, the resource service not meeting the matching threshold values is removed,
in the optimization step, comprehensive similarity matching is carried out on the resource service sets which are matched with the service information, sequencing is carried out according to the comprehensive similarity, and the resource with the maximum similarity is the optimal matching resource recommended by the system.
In the preprocessing step, the comprehensive similarity sim of the basic information text is calculated according to the basic information of the service in the cloud service pool and the number of resource service clustersT(T1,T2) The method comprises the following specific steps:
the morphological similarity reflects the morphological similarity of the two texts, which is determined by the number of the same words in the two texts and the total length of the two texts, and the calculation formula is as follows:
in the formula:
simF(T1,T2) As a text T1And text T2Similarity of parts of speech;
SameWord(T1,T2) As a text T1And text T2The number of the same word;
Len(T1) As a text T1Length of (d);
Len(T2) As a text T2Length of (d);
simF(T1,T2)∈[0,1]the more words two texts are identical, the more similar the two texts are in terms of morphology, simF(T1,T2) The closer to 1.
The word order similarity refers to the similarity of the same words in two texts on the position, and is measured by the number of the same words in the reverse direction of the adjacent sequence, and the calculation formula is as follows:
in the formula, the first step is that,
simO(T1,T2) As a text T1And text T2Word order similarity of parts of speech;
OnceWord(T1,T2) As a text T1And text T2A set of identical words;
|OnceWord(T1,T2) I is text T1And text T2The number of elements in the set of identical words;
if Pfirst(T1,T2) Is an OnecWord (T)1,T2) In the text T1Position number in (1), Psecond(T1,T2) Is Pfirst(T1,T2) Component(s) in text T by corresponding word2In (3) vector(s) formed by sequentially arranging the generated vectors, Revord (T)1,T2) Then is Psecond(T1,T2) The inverse of each adjacent component in the sequence, max (Len (T)1),Len(T2) Is text T)1And text T2Maximum value of sentence length;
the sentence length similarity also reflects the similarity of the two descriptive texts on the form, and the text T1Text T2The formula for calculating the sentence length similarity is as follows:
wherein the content of the first and second substances,
simL(T1,T2) As a text T1And text T2Sentence similarity of parts of speech;
Len(T1) As a text T1Length of (d);
Len(T2) As a text T2Length of (d);
simL(T1,T2)∈[0,1]the closer the lengths of the two texts are, simL(T1,T2) The closer to 1.
The comprehensive similarity of the texts reflects the overall similarity of two description texts, namely the text T1And T2The comprehensive similarity calculation formula is as follows:
wherein, simT(T1,T2) Is the integrated similarity of the two texts,
α1、α2、α3respectively representing influence weights representing the similarity of the word shapes, the similarity of the word orders and the similarity of the sentence lengths,
simT(T1,T2)∈[0,1]the closer the two texts are, simT(T1,T2) The closer to 1.
In the step of aggregating the resource services into service clusters, the base between the resource services
simT(T1,T2)=α1simF(T1,T2)+α2simO(T1,T2)+α3simL(T1,T2)
The information text similarity matrix may be expressed as:
wherein m is the number of all resource services in the cloud service pool;
Tiserve the ith riThe basic information text of (1);
Tjserve the jth rjThe basic information text of (1);
each row of the matrix represents the similarity value of the basic information text of the resource service and other services, each column of the matrix represents the similarity value of the basic information text of other services and the service, and sim (T)i,Tj) And sim (T)j,Ti) Equal;
setting ith service r in cloud service pooliThe average similarity of the basic information text of other resource services is aiAnd then:
wherein, simT(Ti,Tj) For the ith service andsimilarity of jth service basic information text;
the intra-class similarity of the resource service cluster represents the tightness degree of the services in the resource service cluster, the larger the intra-class similarity of the resource service cluster is, the tighter the intra-cluster services are, all the services in the cloud service pool are clustered into q service clusters, and if the theta-th service cluster C isθHas a cluster center ofAnd the service cluster contains t services, the similarity E in the theta resource service clusterθComprises the following steps:
wherein the content of the first and second substances,serving a resource with a cluster CθMiddle ith service riAnd cluster centerSimilarity of basic information text, t is resource service cluster CθThe number of services contained in;
the clustering effect of the resource service can be measured by an evaluation function, and the evaluation function of the resource service clustering is as follows:
wherein the content of the first and second substances,the similarity of the basic information text of the theta-th resource service cluster center and the lambda-th resource service cluster center is expressed and called as the similarity between clusters and is used for measuring the similarity of the basic information between resource service clusters,
the numerator represents the average intra-class similarity of all resource service clusters, the denominator represents the average inter-class similarity between resource service clusters,
since the greater the similarity in the average class of the resource service cluster, the smaller the similarity between the average classes, and the better the clustering effect of the resource service cluster, when F isqWhen the value is maximum, the best clustering effect is achieved.
The algorithm flow of the source clustering is as follows:
And 2, calculating the comprehensive similarity of any two service basic information texts in the cloud service pool.
And 4, the value range of the number q of the resource service clusters is more than or equal to 2 and less than or equal to m 1/2. At the beginning of the clustering, q is 2, so thatAndfor the cluster center, calculate EθAnd FqThe resource service cluster with the minimum similarity in the cluster is selected, and the service with the minimum similarity to the cluster center is found in the resource service cluster and is used as the next cluster center.
And 5, q is q +1, if q is less than or equal to m1/2, the step 4 is carried out, and if not, the step 6 is carried out.
Determination of alternative resource service cluster:
when the resource demander submits own requirements on the platform after all the services in the cloud platform service pool are clustered, the platform matches the basic information of the resource demander with the basic information of all the resource service cluster clustering centers so as to screen out the resource service clusters meeting the requirements,
setting theta cluster center of resource service clusterThe similarity with the basic information of the demand side isSelecting the resource service cluster with the maximum similarity as an alternative resource service cluster, wherein the alternative resource service set D is as follows:
wherein the content of the first and second substances,to show the maximum value of the similarity between each resource service cluster center and the basic information of the demand, cluster () is the corresponding resource service cluster,
by calculating the similarity of the basic information of each resource service cluster center, the matching of the basic information and all manufacturing services in a cloud service pool is avoided, and finally, an alternative service set related to the requirement is obtained.
The method is characterized in that in the mapping and matching step, the object elements are quoted into a provider model and a demander model, and specifically:
the object element M includes a name N of the object, an attribute feature g, and a magnitude v corresponding to the attribute feature, i.e. M ═ N (g, v),
for the provider model, it is disclosed after introducing the object as:
wherein v isBI、vST、vFI、vSIQuantities, N, corresponding to basic information, status information, function information and service information of the provider, respectivelysAs provider service name, GsFor provider service attribute features, VsFor the magnitude corresponding to the provider attribute feature,
the function information and the service information of the resource service comprise a plurality of sub-attributes, the sub-attributes are independent from each other, and each sub-attribute module corresponds to an attribute value, so the disclosure further comprises:
wherein gamma is the number of the function information sub-attribute modules, and each sub-function attribute value is { v }FI1,vFI2,...vFIγE is the number of the service information sub-attribute modules, the service information comprises e sub-attributes, and the value of each sub-attribute is { v }SI1,vSI2,...vSIe},
Similarly, for the demand side model, the model is disclosed after introducing the object elements as follows:
wherein v isDBI、vDST、vDFI、vDSIThe quantity values, N, corresponding to the basic information, the state information, the function information and the service information of the demand side respectivelysServing name for demander, GsFor demander service attribute features, VdsThe magnitude corresponding to the attribute feature of the demander,
the mapping matching is the matching of the Chinese character parameters and the numerical parameters of the Vs and the Vds, namely the calculation process of the similarity values of the Vs and the Vds.
In the mapping matching step, the state information of the provider and the demander is matched, and the specific steps are as follows:
Step 5.kD=kD+1, if kD≤nDTurning to the step 4, otherwise, turning to the step 6;
In the mapping matching step, the functional information is matched based on the calculation of the similarity of the concept and the similarity of the numerical parameters in the functional information of the provider and the demander, and the specific steps are as follows:
in the calculation of the concept similarity, the semantic distance and the semantic overlap ratio are respectively calculated, then the concept similarity is finally calculated according to the calculation result,
in the semantic distance calculation, an ontology tree of the concept is established, and the weight W expression of an edge between a node of the nth layer of concept of the ontology tree and a node of the (n-1) th layer of concept is as follows:
the formula for calculating the semantic distance is as follows:
wherein the content of the first and second substances,
dist(C1,C2) Is a concept C1And concept C2Semantic distance between, lens is concept C1And concept C2The shortest path length between, ∑ W (n) is concept C1And concept C2Path weights between;
the semantic overlap ratio is measured according to the number of the same upper concept nodes contained in the two concept nodes in the resource ontology tree, the more the two concept nodes contain the same upper concept nodes, the higher the semantic overlap ratio of the two concepts is, the greater the similarity degree between the concepts is, and the calculation formula of the semantic overlap ratio is as follows:
wherein, OL (C)i,Cj) Is the degree of semantic overlap between two concepts, Path (C)i) As concept node CiTo the set of upper concepts experienced by Root of ontology tree, | Path (Ci) Path (cj) | CiAnd CjTwo concept nodes contain the same number of upper concept nodes, | Path (Ci) Path (cj) | CiAnd CjThe number of nodes of all upper concepts of the two concepts,
according to the calculation structure of the semantic distance and the semantic overlap ratio and the weight of the semantic distance and the semantic overlap ratio, the concept similarity simCThe formula for the calculation of (Ci, Cj) is:
wherein eta 1 and eta 2 are the influence weights of the semantic distance and the semantic overlap ratio respectively,
when the algorithm is applied to the matching process of the function information, the w-th concept parameter in the function information of the resource provider is set as FIwThe w-th concept parameter in the functional information of the resource demander is DFIwThen, the concept semantic similarity formula is:
the numerical parameter similarity calculation formula is as follows:
wherein A ishIs the h-th numerical parameter information in the function information of the resource provider, AhIs the h-th numerical parameter information, | A, in the functional information of the resource demand sideh| is a numerical range AhLength of (d).
The calculation formula for matching the functional information similarity of the supply and demand parties is as follows:
wherein, γ1As a quantity of a conceptual parameter, γ2Is the number of the numerical parameters,
based on the calculation formula of the functional information similarity matching, the matching steps are as follows:
And step 2, outputting: resource service set D for completing function information matchingF;
In the mapping matching step, normalization processing is carried out on a plurality of service information to obtain attribute values of the information service, and matching of the information service of the provider and the information service of the demander is realized according to a fuzzy number calculation formula, which comprises the following specific steps:
service information to providersService information with a requesting partyMatching is carried out, and the fuzzy number formula is as follows:
wherein the content of the first and second substances,and isIs composed ofSmall element of (a), x+Is composed ofIs a large element of (a), x isThe bit of (which represents the value with the highest probability of taking a value in this interval, usually called the information preference value),
in the same way, the method for preparing the composite material,and 0 < y-≤y≤y+,y-Is composed ofSmall element of (a), y+Is composed ofIs a large element of (a), y isThe characteristic of (a) is that,
if a decision scheme includes multiple service information, i.e. ThenThe fuzzy number calculation formula is as follows:
setting the resource service set matched with the function information as DF={df1,df2...dfu...dfnF},DFThe attribute value of the service information of the u-th resource is dfu={SIu1,SIu2...SIueIn which n isFIs DFThe number of medium resource services (resource providers), e, indicates the dimension of the service information attribute, and since the physical meanings of the service information attributes are different and the measurement units are different, the dimensions and the magnitude of the attribute values are different. For the convenience of comparison and calculation, the attribute values of the service information need to be normalized,
in the normalization processing step, the normalization processing formula is as follows:
wherein, SIuvServing information attribute's of the u-th candidate resourceIndividual attribute value, minSIvServing information attributes for all candidate resourcesMinimum value of individual attribute, maxSIvServing information attributes for all candidate resourcesMaximum value of attribute, SIuvFor the u-th candidate resource service information attribute after normalization processingThe value of each of the attributes is,
let SIuvIs expressed as a fuzzy number ofIn the same way, SIuvIs expressed as a fuzzy number ofThe service information attribute value of the resource demander may be expressed as DSI ═ DSI1,DSI2...DSIeThe first in the normalized requirementsDimension attribute value DSIvThe fuzzy number of (a) means that the result isThe calculation formula for obtaining the service information matching between the provider and the demander according to the fuzzy number formula is as follows:
based on the calculation formula for matching the service information of the demand party, the following steps are adopted to complete the matching of the service information:
And step 2, outputting: completing a resource service set S matched with the service information;
And 6.u is equal to u +1, and if u is less than or equal to nFGo to step 5, otherwise go to step 7;
and 7, finishing the algorithm, and outputting the resource service set S matched with the service information.
In the optimization step, objective weight and subjective weight of the comprehensive similarity are calculated respectively, and then comprehensive similarity matching is performed, and the specific steps are as follows:
in the calculation of the objective weight,
let the tetrad I ═ (B, a ═ C ═ D, V, F) be the cloud manufacturing platform resource service information system,
wherein B is a non-empty subset of the finite universe of resources served, and B ═ x1,x2,x3...xnIs history data, a is a set of attributes, C ═ a1,a2,a3...aoD ═ a, is a set of respective sub-attributes of the function information and the service informationo+1And F is an information function of mapping each sub-attribute to the value range.
Let X and Y be non-empty subsets of the finite discourse domain, haveIn a partial order relationship, then
Wherein | X | is the cardinal number of the set X and is the number of the objects in the equivalence class; c (X, Y) is the misclassification rate of set X relative to set Y;
beta is more than or equal to 0 and less than 0.5, existIf c (X, Y) ═ 0, then X is included by the Y criteria,
let X be B/C be X1,X2...X|B/C|B is an equivalence class derived from each sub-attribute C, Y ═ B/D ═ Y1,Y2...Y|B/D|B is an equivalence class obtained by dividing through a decision attribute set D,
in the case of beta ∈ [0,0.5), YjThe following approximate distribution of β for C is:
wherein the content of the first and second substances,is a positive region of beta, which may also be denoted as Cβ(Yj) Is the whole decision class YjC of lower approximation { Cβ(Y1),Cβ(Y2)...Cβ(Y|B/D|) Represents the probability case that its decision class is based on B/D,
the information quantity of each sub-attribute reflects the classification capability of the attribute on the data object, the size of the information quantity reflects the strength of the classification capability, and for any sub-attribute apInformation amount γ (a) thereofp) The calculation method of (2) is as follows:
wherein, XiFor set B by apEach equivalence class of the partition, | Xi| is the number of resource services corresponding to each equivalence class,
the dependency degree of each sub-attribute is reflected by the dependency degree of the decision attribute classification on the index attribute, the magnitude of the dependency degree indicates the importance degree of the index, and the dependency degree is reflected by the arbitrary sub-attribute apWhich is dependent on a degree lambda (a)p) The calculation method of (2) is as follows:
gamma (a) of each sub-attributep) And λ (a)p) The importance degree of the sub-attribute is described from different angles, and the calculation formula of the objective weight of each sub-attribute can be obtained by comprehensively considering the importance degree of the sub-attribute:
wherein, gamma (a)p) Is the amount of information of the sub-attribute, λ (a)p) Is the degree of dependence of a sub-attribute, and ω1+ω2+...+ωo=1,
Because different users have different preferences for each sub-attribute of the resource service, if the objective weight obtained by the algorithm is directly adopted, the preferences of the users can be ignored; if only subjective weight is adopted, the weight setting is unreasonable when the experience of the user is insufficient, therefore, the importance degree of each sub-attribute is measured together through the subjective and objective weight,
in the step of calculating the subjective weight, the subjective weight is calculated,
setting the subjective weight of a resource demander asThen each sub-attribute ω'pThe comprehensive weight calculation method comprises the following steps:
wherein, the importance degree of the subjective weight is reflected, the more important the user's preference for each sub-attribute is, the larger is, and the is [0,1 ].
The comprehensive similarity matching formula is as follows:
simZ(RS,RD)=ω′1sim1+ω′2sim2+...+ω′osimo
wherein, sim1,sim2...simoRepresenting the magnitude of similarity, ω 'of each sub-attribute'1,ω′2...ω′oRepresents the corresponding comprehensive weight, omega'1+ω′2+...+ω′o=1,
The matching steps are as follows:
And step 2, outputting: resource service set S finally matched with requirementE,
Step 5.kZ=kZ+1, if kZ≤nZGo to step 4, otherwise go to step 6,
The cloud manufacturing platform resource service matching method obtained by the technical scheme of the invention has the beneficial effects that:
the matching of the cloud manufacturing platform resource service is divided into three stages of resource service preprocessing, resource service mapping matching and resource service optimal configuration.
The optimized K-means clustering algorithm is introduced into the resource service preprocessing, the matching range of the resource service can be greatly reduced through the resource service preprocessing, and the operation scale is reduced.
In the optimization configuration stage of the resource service, the objective weights of all the sub-attributes are firstly calculated from the transaction history by adopting a variable-precision rough set theory, and then the comprehensive matching weights are calculated by combining the personalized preference of the resource demand party, so that the user satisfaction is high.
Drawings
FIG. 1 is a flow chart of a resource service clustering algorithm in a cloud service pool according to the present invention;
FIG. 2 is cloud manufacturing resource demander information in accordance with the present invention
FIG. 3 is a cloud manufacturing resource provider information diagram in accordance with the present invention
FIG. 4 is a graph of the average similarity result of the basic information texts according to the present invention;
FIG. 5 is a diagram illustrating the similarity between the basic information text of each resource service and r21 according to the present invention;
FIG. 6 is a graph showing the relationship between the number of clusters and the evaluation function according to the present invention
FIG. 7 is a resource service clustering result diagram according to the present invention
FIG. 8 is a tree diagram of partially fabricated resource entities according to the present invention
FIG. 9 is a diagram of the matching result of the functional information according to the present invention
FIG. 10 is a diagram of historical transaction records of a resource requiring party in accordance with the present invention
FIG. 11 is a diagram of the result of the historical data clustering according to the present invention
FIG. 12 is a graph of the result of the integrated similarity matching according to the present invention
FIG. 13 is a comparison graph of the matching algorithm of the present invention
FIG. 14 is a diagram of a matching model of cloud manufacturing platform resources and requirements according to the present invention.
Detailed Description
The present invention will be described in detail below with reference to the accompanying drawings.
A cloud manufacturing platform resource service matching method comprises a model establishing step, a preprocessing step, a mapping matching step and an optimizing step, and is characterized in that the model establishing step, the preprocessing step, the mapping matching step and the optimizing step are sequentially carried out,
in the step of establishing the model, the model is established,
establishing a formalized description model of a cloud manufacturing platform resource provider, namely a provider model:
RS={ID,BI,ST,FI,SI},
wherein the content of the first and second substances,
the ID is an identification code of the cloud manufacturing resource, cannot be changed and does not participate in matching, and is mainly used for positioning and searching the cloud manufacturing resource;
the BI indicates Basic Information (Basic Information) of a resource provider, the Basic Information mainly comprises a brief introduction of a resource service, a name of the resource service, a contact way of the resource provider, an address of the resource provider and the like, wherein the brief introduction of the resource service is main Information which is described by adopting a simple text and participates in a matching process of the resource service, so that the filling of the brief introduction of the resource service emphasizes the characteristics of the resource;
ST represents the State information (State information) of the resource provider, and the State information of the resource service is set by the resource provider, and comprises an idle State, a non-full-load State, a full-load State and an invalid State;
FI denotes function Information (Functional Information) of the resource provider, which is Information that the resource demander pays attention to when searching for the resource service;
SI represents Service Information (Service Information) of a resource provider, and the Service Information refers to various performance indexes required by the resource provider when providing a Service;
establishing a formalized description model of a resource demander of a cloud manufacturing platform, namely a demander model:
RD={ID,DBI,DST,DFI,DSI},
wherein the content of the first and second substances,
DBI represents Basic Information (Demand Basic Information) of a resource demander;
the DST represents State information (Demand State information) of the resource Demand side, including an unexecuted State, an executing State, a completed State and a failure Demand State;
the DFI represents function Information (Demand function Information) of the resource demander;
the DSI represents Service attribute Information (Demand Service Information) of the resource demander,
in the preprocessing step, resource services similar to the text description of the basic information BI are aggregated into service clusters, and the service cluster most relevant to the basic information of the demand side is selected as a candidate resource service set so as to narrow the matching range, wherein the similarity of the text description comprises word form similarity, word order similarity, sentence length similarity and comprehensive similarity,
in the mapping and matching step, the state information ST, the function information FI and the service information SI of the alternative resource service selected by the pretreatment are respectively matched with the state information DST, the function information DFI and the service information DSI of the resource demand party, corresponding matching threshold values are set, the resource service meeting the matching threshold values is reserved, the resource service not meeting the matching threshold values is removed,
in the optimization step, comprehensive similarity matching is carried out on the resource service sets which are matched with the service information, sequencing is carried out according to the comprehensive similarity, and the resource with the maximum similarity is the optimal matching resource recommended by the system.
In the preprocessing step, the comprehensive similarity sim of the basic information text is calculated according to the basic information of the service in the cloud service pool and the number of resource service clustersT(T1,T2) The method comprises the following specific steps:
the morphological similarity reflects the morphological similarity of the two texts, which is determined by the number of the same words in the two texts and the total length of the two texts, and the calculation formula is as follows:
in the formula:
simF(T1,T2) As a text T1And text T2Similarity of parts of speech;
SameWord(T1,T2) As a text T1And text T2The number of the same word;
Len(T1) As a text T1Length of (d);
Len(T2) As a text T2Length of (d);
simF(T1,T2)∈[0,1]the more words two texts are identical, the more similar the two texts are in terms of morphology, simF(T1,T2) The closer to 1.
The word order similarity refers to the similarity of the same words in two texts on the position, and is measured by the number of the same words in the reverse direction of the adjacent sequence, and the calculation formula is as follows:
in the formula, the first step is that,
simO(T1,T2) As a text T1And text T2Word order similarity of parts of speech;
OnceWord(T1,T2) As a text T1And text T2A set of identical words;
|OnceWord(T1,T2) I is text T1And text T2The number of elements in the set of identical words;
if Pfirst(T1,T2) Is an OnecWord (T)1,T2) In the text T1Position number in (1), Psecond(T1,T2) Is Pfirst(T1,T2) Component(s) in text T by corresponding word2In (3) vector(s) formed by sequentially arranging the generated vectors, Revord (T)1,T2) Then is Psecond(T1,T2) All phases ofInverse order of adjacent components, max (Len (T)1),Len(T2) Is text T)1And text T2The maximum value of sentence length.
The sentence length similarity also reflects the similarity of the two descriptive texts on the form, and the text T1Text T2The formula for calculating the sentence length similarity is as follows:
wherein the content of the first and second substances,
simL(T1,T2) As a text T1And text T2Sentence similarity of parts of speech;
Len(T1) As a text T1Length of (d);
Len(T2) As a text T2Length of (d);
simL(T1,T2)∈[0,1]the closer the lengths of the two texts are, simL(T1,T2) The closer to 1.
The comprehensive similarity of the texts reflects the overall similarity of two description texts, namely the text T1And T2The comprehensive similarity calculation formula is as follows:
wherein, simT(T1,T2) Is the integrated similarity of the two texts,
α1、α2、α3respectively representing influence weights representing the similarity of the word shapes, the similarity of the word orders and the similarity of the sentence lengths,
simT(T1,T2)∈[0,1]the closer the two texts are, simT(T1,T2) The closer to 1.
simT(T1,T2)=α1simF(T1,T2)+α2simO(T1,T2)+α3simL(T1,T2)
Applying WOL-S-TC test setRespectively calculating the similarity of word shapes Sim by 20 semantic Web services provided in the systemwordAfter the sentence length similarity Simlen, calculating the lambda through a linear optimization method1=0.45、λ1=0.45、λ1=0.10。
In the step of aggregating the resource services into a service cluster, the basic information text similarity matrix between the resource services can be expressed as:
wherein m is the number of all resource services in the cloud service pool;
Tiserve the ith riThe basic information text of (1);
Tjserve the jth rjThe basic information text of (1);
each row of the matrix represents the similarity value of the basic information text of the resource service and other services, each column of the matrix represents the similarity value of the basic information text of other services and the service, and sim (T)i,Tj) And sim (T)j,Ti) Equal;
setting ith service r in cloud service pooliThe average similarity of the basic information text of other resource services is aiAnd then:
wherein, simT(Ti,Tj) Similarity of basic information texts of the ith service and the jth service;
the intra-class similarity of the resource service cluster represents the tightness degree of the services in the resource service cluster, the larger the intra-class similarity of the resource service cluster is, the tighter the intra-cluster services are, all the services in the cloud service pool are clustered into q service clusters, and if the theta-th service cluster C isθHas a cluster center ofAnd the service cluster contains t services, the similarity E in the theta resource service clusterθComprises the following steps:
wherein the content of the first and second substances,serving a resource with a cluster CθMiddle ith service riAnd cluster centerSimilarity of basic information text, t is resource service cluster CθThe number of services contained in;
the clustering effect of the resource service can be measured by an evaluation function, and the evaluation function of the resource service clustering is as follows:
wherein the content of the first and second substances,the similarity of the basic information text of the theta-th resource service cluster center and the lambda-th resource service cluster center is expressed and called as the similarity between clusters and is used for measuring the similarity of the basic information between resource service clusters,
the numerator represents the average intra-class similarity of all resource service clusters, the denominator represents the average inter-class similarity between resource service clusters,
since the greater the similarity in the average class of the resource service cluster, the smaller the similarity between the average classes, and the better the clustering effect of the resource service cluster, when F isqWhen the value is maximum, the best clustering effect is achieved.
The algorithm flow of the source clustering is as follows:
And 2, calculating the comprehensive similarity of any two service basic information texts in the cloud service pool.
And 4, the value range of the number q of the resource service clusters is more than or equal to 2 and less than or equal to m 1/2. At the beginning of the clustering, q is 2, so thatAndfor the cluster center, calculate EθAnd FqThe resource service cluster with the minimum similarity in the cluster is selected, and the service with the minimum similarity to the cluster center is found in the resource service cluster and is used as the next cluster center.
And 5, q is q +1, if q is less than or equal to m1/2, the step 4 is carried out, and if not, the step 6 is carried out.
Determination of alternative resource service cluster:
when the resource demander submits own requirements on the platform after all the services in the cloud platform service pool are clustered, the platform matches the basic information of the resource demander with the basic information of all the resource service cluster clustering centers so as to screen out the resource service clusters meeting the requirements,
setting theta cluster center of resource service clusterThe similarity with the basic information of the demand side isSelecting the resource service cluster with the maximum similarity as an alternative resource service cluster, wherein the alternative resource service set D is as follows:
wherein the content of the first and second substances,to show the maximum value of the similarity between each resource service cluster center and the basic information of the demand, cluster () is the corresponding resource service cluster,
by calculating the similarity of the basic information of each resource service cluster center, the matching of the basic information and all manufacturing services in a cloud service pool is avoided, and finally, an alternative service set related to the requirement is obtained.
The method is characterized in that in the mapping and matching step, the object elements are quoted into a provider model and a demander model, and specifically:
the object element M includes a name N of the object, an attribute feature g, and a magnitude v corresponding to the attribute feature, i.e. M ═ N (g, v),
for the provider model, it is disclosed after introducing the object as:
wherein v isBI、vST、vFI、vSIQuantities, N, corresponding to basic information, status information, function information and service information of the provider, respectivelysAs provider service name, GsTo be liftedSupplier service attribute characteristics, VsFor the magnitude corresponding to the provider attribute feature,
the function information and the service information of the resource service comprise a plurality of sub-attributes, the sub-attributes are independent from each other, and each sub-attribute module corresponds to an attribute value, so the disclosure further comprises:
wherein gamma is the number of the function information sub-attribute modules, and each sub-function attribute value is { v }FI1,vFI2,...vFIγE is the number of the service information sub-attribute modules, the service information comprises e sub-attributes, and the value of each sub-attribute is { v }SI1,vSI2,...vSIe},
Similarly, for the demand side model, the model is disclosed after introducing the object elements as follows:
wherein v isDBI、vDST、vDFI、vDSIThe quantity values, N, corresponding to the basic information, the state information, the function information and the service information of the demand side respectivelysServing name for demander, GsFor demander service attribute features, VdsThe magnitude corresponding to the attribute feature of the demander,
the mapping matching is the matching of the Chinese character parameters and the numerical parameters of the Vs and the Vds, namely the calculation process of the similarity values of the Vs and the Vds.
In the mapping matching step, the state information of the provider and the demander is matched, and the specific steps are as follows:
Step 5.kD=kD+1, if kD≤nDTurning to the step 4, otherwise, turning to the step 6;
In the mapping matching step, the functional information is matched based on the calculation of the similarity of the concept and the similarity of the numerical parameters in the functional information of the provider and the demander, and the specific steps are as follows:
in the calculation of the concept similarity, the semantic distance and the semantic overlap ratio are respectively calculated, then the concept similarity is finally calculated according to the calculation result,
in the semantic distance calculation, an ontology tree of the concept is established, and the weight W expression of an edge between a node of the nth layer of concept of the ontology tree and a node of the (n-1) th layer of concept is as follows:
the formula for calculating the semantic distance is as follows:
wherein the content of the first and second substances,
dist(C1,C2) Is a concept C1And concept C2Semantic distance between, lens is concept C1And concept C2The shortest path length between, ∑ W (n) is concept C1And concept C2Path weights between;
the semantic overlap ratio is measured according to the number of the same upper concept nodes contained in the two concept nodes in the resource ontology tree, the more the two concept nodes contain the same upper concept nodes, the higher the semantic overlap ratio of the two concepts is, the greater the similarity degree between the concepts is, and the calculation formula of the semantic overlap ratio is as follows:
wherein, OL (C)i,Cj) Is the degree of semantic overlap between two concepts, Path (C)i) As concept node CiTo the set of upper concepts experienced by Root of ontology tree, | Path (Ci) Path (cj) | CiAnd CjTwo concept nodes contain the same number of upper concept nodes, | Path (Ci) Path (cj) | CiAnd CjThe number of nodes of all upper concepts of the two concepts,
according to the calculation structure of the semantic distance and the semantic overlap ratio and the weight of the semantic distance and the semantic overlap ratio, the concept similarity simCThe formula for the calculation of (Ci, Cj) is:
wherein eta 1 and eta 2 are the influence weights of the semantic distance and the semantic overlap ratio respectively,
when the algorithm is applied to the matching process of the function information, the w-th concept parameter in the function information of the resource provider is set as FIwThe w-th concept parameter in the functional information of the resource demander is DFIwThen, the concept semantic similarity formula is:
the numerical parameter similarity calculation formula is as follows:
wherein A ishIs the h-th numerical parameter information in the function information of the resource provider, AhIs the h-th numerical parameter information, | A, in the functional information of the resource demand sideh| is a numerical range AhLength of (d).
The calculation formula for matching the functional information similarity of the supply and demand parties is as follows:
wherein, γ1As a quantity of a conceptual parameter, γ2Is the number of the numerical parameters,
based on the calculation formula of the functional information similarity matching, the matching steps are as follows:
And step 2, outputting: resource service set D for completing function information matchingF;
In the mapping matching step, normalization processing is carried out on a plurality of service information to obtain attribute values of the information service, and matching of the information service of the provider and the information service of the demander is realized according to a fuzzy number calculation formula, which comprises the following specific steps:
service information to providersService information with a requesting partyMatching is carried out, and the fuzzy number formula is as follows:
wherein the content of the first and second substances,and isIs composed ofSmall element of (a), x+Is composed ofIs a large element of (a), x isThe bit of (which represents the value with the highest probability of taking a value in this interval, usually called the information preference value),
in the same way, the method for preparing the composite material,and isIs composed ofSmall element of (a), y+Is composed ofIs a large element of (a), y isThe characteristic of (a) is that,
if a decision scheme includes multiple service information, i.e. ThenThe fuzzy number calculation formula is as follows:
setting the resource service set matched with the function information as DF={df1,df2...dfu...dfnF},DFThe attribute value of the service information of the u-th resource is dfu={SIu1,SIu2...SIueIn which n isFIs DFThe number of medium resource services (resource providers), e, indicates the dimension of the service information attribute, and since the physical meanings of the service information attributes are different and the measurement units are different, the dimensions and the magnitude of the attribute values are different. For the convenience of comparison and calculation, the attribute values of the service information need to be normalized,
in the normalization processing step, the normalization processing formula is as follows:
wherein, SIuvFor the v attribute value of the u candidate resource service information attribute, minSIvIs the minimum value of the v attribute of all candidate resource service information attributes, maxSIvIs the maximum value of the v-th attribute, SI' of all candidate resource service information attributesuvFor the v attribute value of the u candidate resource service information attribute after the normalization process,
let SIuvIs expressed as a fuzzy number ofIn the same way, SIuvIs expressed as a fuzzy number ofThe service information attribute value of the resource demander may be expressed as DSI ═ DSI1,DSI2...DSIeAnd v-dimension attribute value DSI in the normalized requirementvThe fuzzy number of (a) means that the result isThe calculation formula for obtaining the service information matching between the provider and the demander according to the fuzzy number formula is as follows:
based on the calculation formula for matching the service information of the demand party, the following steps are adopted to complete the matching of the service information:
And step 2, outputting: completing a resource service set S matched with the service information;
And 6.u is equal to u +1, and if u is less than or equal to nFGo to step 5, otherwise go to step 7;
and 7, finishing the algorithm, and outputting the resource service set S matched with the service information.
In the optimization step, objective weight and subjective weight of the comprehensive similarity are calculated respectively, and then comprehensive similarity matching is performed, and the specific steps are as follows:
in the calculation of the objective weight,
let the tetrad I ═ (B, a ═ C ═ D, V, F) be the cloud manufacturing platform resource service information system,
wherein B is a non-empty subset of the finite universe of resources served, and B ═ x1,x2,x3...xnIs history data, a is a set of attributes, C ═ a1,a2,a3...aoD ═ a, is a set of respective sub-attributes of the function information and the service informationo+1And F is an information function of mapping each sub-attribute to the value range.
Let X and Y be non-empty subsets of the finite discourse domain, haveIn a partial order relationship, then
Wherein | X | is the cardinal number of the set X and is the number of the objects in the equivalence class; c (X, Y) is the misclassification rate of set X relative to set Y;
beta is more than or equal to 0 and less than 0.5, existIf c (X, Y) ═ 0, then X is included by the Y criteria,
let X be B/C be X1,X2...X|B/C|B is an equivalence class derived from each sub-attribute C, Y ═ B/D ═ Y1,Y2...Y|B/D|B is an equivalence class obtained by dividing through a decision attribute set D,
in the case of beta ∈ [0,0.5), YjThe following approximate distribution of β for C is:
wherein the content of the first and second substances,is a positive region of beta, which may also be denoted as Cβ(Yj) Is the whole decision class YjC of lower approximation { Cβ(Y1),Cβ(Y2)...Cβ(Y|B/D|) Represents the probability case that its decision class is based on B/D,
the information quantity of each sub-attribute reflects the classification capability of the attribute on the data object, and the size of the information quantity is oppositeReflecting the strength of the classification capability of the attribute, for any sub-attribute apInformation amount γ (a) thereofp) The calculation method of (2) is as follows:
wherein, XiFor set B by apEach equivalence class of the partition, | Xi| is the number of resource services corresponding to each equivalence class,
the dependency degree of each sub-attribute is reflected by the dependency degree of the decision attribute classification on the index attribute, the magnitude of the dependency degree indicates the importance degree of the index, and the dependency degree is reflected by the arbitrary sub-attribute apWhich is dependent on a degree lambda (a)p) The calculation method of (2) is as follows:
gamma (a) of each sub-attributep) And λ (a)p) The importance degree of the sub-attribute is described from different angles, and the calculation formula of the objective weight of each sub-attribute can be obtained by comprehensively considering the importance degree of the sub-attribute:
wherein, gamma (a)p) Is the amount of information of the sub-attribute, λ (a)p) Is the degree of dependence of a sub-attribute, and ω1+ω2+...+ωo=1,
Because different users have different preferences for each sub-attribute of the resource service, if the objective weight obtained by the algorithm is directly adopted, the preferences of the users can be ignored; if only subjective weight is adopted, the weight setting is unreasonable when the experience of the user is insufficient, therefore, the importance degree of each sub-attribute is measured together through the subjective and objective weight,
in the step of calculating the subjective weight, the subjective weight is calculated,
setting the subjective weight of a resource demander asThe respective sub-attribute ωpThe comprehensive weight calculation method comprises the following steps:
wherein, the importance degree of the subjective weight is reflected, the more important the user's preference for each sub-attribute is, the larger is, and the is [0,1 ].
The comprehensive similarity matching formula is as follows:
simZ(RS,RD)=ω′1sim1+ω′2sim2+...+ω′osimo
wherein, sim1,sim2...simoRepresenting the magnitude of similarity, ω 'of each sub-attribute'1,ω′2...ω′oRepresents the corresponding comprehensive weight, omega'1+ω′2+...+ω′o=1,
The matching steps are as follows:
And step 2, outputting: resource service set S finally matched with requirementE,
Step 5.kZ=kZ+1, if kZ≤nZGo to step 4, otherwise go to step 6,
Example 1
Measured in terms of the number of words in the text, Revord (T) is explained in detail in two contexts as follows1,T2) The method of calculation is carried out by taking the measured values,
T1this is a gear manufacturing service
This is 1, 2, 3, 4, 5, 6, 7
T2Manufacturing and processing shaft parts
Manufacture is 1, process is 2, shaft is 3, part is 4
Wherein, Revord (T)1,T2) Machining and manufacturing, Pfirst(T1,T2)={4,5};Psecond(T1,T2) 1, {5, 4}, since 5>4, so Revord (T)1,T2)=1。
Example 2
As shown in fig. 1, the resource service clustering specifically includes the following steps:
(1) initializing basic information of services in the cloud service pool, the number q of resource service clusters and the like
(2) And calculating the comprehensive similarity of any two service basic information texts in the cloud service pool, and solving a similarity matrix.
(3) Computing the ith service riAverage similarity a of basic information textsiSelecting the service with the maximum average similarity of the basic information texts as a first clustering centerSelecting andthe service with the least similarity serves as the second cluster center rc2。
(4) The value range of the number q of the resource service clusters isWhen the resource service starts clustering, q is 2, so as toAndfor the clustering centers to start clustering, calculate EθAnd FqThe resource service cluster with the minimum similarity in the cluster is selected, the service with the minimum similarity to the basic information text of the cluster center is found in the resource service cluster, and the service is used as the next cluster center.
(6) Calculate and compare FqTo find FqThe size of q at this time is recorded, and the result of clustering is output.
Example 3
Taking an example that an enterprise requests machining and manufacturing of a gear part, in order to improve production efficiency of a product, the enterprise completes machining of part of the part by relying on a cloud manufacturing platform, and requirement information of the enterprise is shown in fig. 2: assuming that 30 resource services are provided in the cloud manufacturing service platform at this time, r is used respectively1-r30Indicating, resource information r1-r10Basic information BIfTo "this is a machining and manufacturing service of one gear part", resource information r11-r20Basic information BIsFor manufacturing and processing shaft parts "Resource information r21-r30Basic information BItTo "provide bolt manufacturing service", all resource providers' information is as shown in figure 3,
firstly, calculating the comprehensive similarity of basic information texts of each resource service, and taking a1=0.45、a2=0.45、a3Calculating the text comprehensive similarity of any two resource basic information in the cloud resource pool, wherein the calculation result of the similarity matrix R is as follows:
the average similarity results of the basic information texts are shown in figures 4 and 5,
as shown in FIG. 4, the average similarity of the basic information text, a, is known21-a23Equal and all maximum value selecting resource service r21-r30Any one of them is used as the first cluster center (r is selected)21As cluster centers), compute each resource service and r21Similarity of basic information;
as shown in FIG. 5, resource service r11-r20And resource service r21Has the minimum similarity to the basic information text, so the resource service r is selected11-r20Any resource service as the second cluster center (r-select)11) Then, for 30 resource service clusters, q is more than or equal to 2 and less than or equal to 5, and the relation between the cluster number and the evaluation function is shown in FIG. 6,
as shown in fig. 6, when q is 3, the evaluation function value reaches the maximum value, and the clustering effect is the best, and then the clustering result is as shown in fig. 7,
the similarity between the user requirement and the basic information text of each clustering center is as follows:
simT(T21,TDBI)=0.55;
simT(T11,TDBI)=0.55;
simT(T1,TDBI)=1;
it can be seen that resourcesService cluster C3Most similar to the user needs, so the resource service r1-r10As an alternative resource service.
Through the calculation and analysis, 30 resource service ranges in the cloud platform can be narrowed to 10 through the preprocessing of the resource services, the clustering stage of the resource services in the preprocessing does not need to participate in the matching process and can be completed before matching, the matching range can be quickly narrowed only by calculating the similarity between the basic information required by the user and each resource service clustering center during service matching, the matching with all resources in the cloud platform is avoided, and the matching time is saved.
Example 4
As shown in FIG. 8, if the depth of the concept in the ontology tree and the width of the concept are not considered, let SimC(C1、C2) 1/(1+ p) (p ≧ 0), where p is concept C1And C2The semantic distance of the concept and the path length between the equipment resource and the manufacturing equipment, the coordinate boring machine and the boring machine are all 1, so the similarity Sim of the two pairs of conceptsCAre the same; but in practical situations, the concept similarity of the coordinate boring machine and the boring machine is considered to be higher;
based on the depth at which the concept is located and the width of the concept in the ontology tree, the weight of the edge between the device resource and the computing device in fig. 8 is 0.25, the weight of the edge between the machining device and the boring machine is 0.0625, the weight of the edge between the boring machine and the coordinate boring machine is 0.03125,
the semantic distance dist (boring machine ) between the boring machine and the boring machine is 0; the semantic distance between the boring machine and the horizontal boring machine is dist (boring machine, horizontal boring machine) ═ 0.03125, the semantic distance between the manufacturing equipment and the drilling machine is dist (manufacturing equipment, drilling machine): 0.375, and the semantic distance between the equipment resource and the horizontal boring machine is dist (equipment resource, horizontal boring machine): 1.875; since there is no inheritance relationship between the two concepts of the machining equipment and the heat treatment equipment, the semantic distance between the two concepts is infinity (machining equipment and heat treatment equipment)
Example 5
Precision machining of certain part by resource demand sideValue range B of degree1=[0.09,0.1]The length is 0.01,
if the resource provider can reach the precision of A1=[0.09,0.1]If the numerical parameter similarity is simN(A1,B1)=1;
The resource provider can achieve the precision of a1 ═ 0.07,0.095]If the numerical parameter similarity is simN(A1,B1)=0.5;
If the resource provider can achieve the precision of A1 ═ 0.07,0.085]If the numerical parameter similarity is simN(A1,B1)=0。
Example 6
Based on embodiment 3, it can be known that the alternative resource service set is the resource service r1-r10,
(1) State information matching
Since the resource serves r2、r9For a full load state, the resource service r3、r10If the state information is invalid, the state information is rejected, and the matching result of the state information is DS={r1,r4,r5,r6,r7,r8};
(2) Functional information matching
Get eta1=0.9、η10.1, 0.8, DSThe similarity of the function information of the service and the requirement of each resource is shown in FIG. 9, and the service r4、r7The function matching similarity is smaller than the threshold value and is removed, and the matching result of the function information is DS={r1,r5,r6,r8}。
(3) Service information matching
Setting service information matching threshold 0.9, DFThe result of the matching degree of the service information of each resource service and the requirement is as follows, and it can be seen that the resource service r5Is less than the threshold value and is eliminated, and the result of the service information matching is S ═ r1,r6,r8}。
simS(r1.SI,DSI)=1
simS(r5·SI,DSI)=0.89
simS(r6.SI,DSI)=0.91
simS(r8,SI,DSI)=0.96
The result of the resource service mapping matching is S ═ { r ═ r1,r6,r8And comparing the information with the information of the demand side to find that the three resource services can basically meet the demand of the demand side.
Example 7
On the basis of embodiment 6, 8 groups of data are randomly generated to represent 8 groups of historical transaction records in the cloud platform, as shown in fig. 10, for the function sub-attribute and the service information sub-attribute in the historical records, since the function sub-attribute and the service information sub-attribute of different task demands are different, and there is no comparability between the historical data, the similarity values are used to represent the function sub-attribute, the technical level and the reliability, and the ratio of the provider to the demander is used to represent the time and the price sub-attribute.
Because the rough set can only be used for clustering data, the data in FIG. 10 is clustered by k-medoids, the number of clusters is 5, the clustering result is shown in FIG. 11, and the selected resource service history record Y1=B/d1={H1,H3,H4,H8}, resource service history record Y not selected2=B/d2={H2,H5,H6,H7According to the sub-attribute a7The equivalence class of the division is B/a7={{H1},{H2,H3,H4,H8}},{H5,H6,H7Get the sub-attribute a according to the information quantity calculation formula7Comprises the following steps:
similarly, calculate the sub-attribute a1-a8The information amounts of (a) are 32/64, 32/64, 18/64, 18/64, 26/64, 22/64, 26/64 and 16/64 respectively,
let beta equal to 0.4 to obtain Y1About sub-attribute a7The lower distribution of β of (a) is:
Y2about sub-attribute a7The lower distribution of β of (a) is:
according to the dependency calculation formula, a is obtained7The dependence of (A) is:
respectively calculating according to an objective weight calculation formula:
ω1=0.18,ω2=0.18,ω3=0.08,ω4=0.11,ω5=0.11,ω6=0.09,ω7=0.17,ω8=0.08,
the subjective weight of each sub-attribute given by the user is set as:
ω*={0.125,0.125,0.125,0.125,0.16,0.1,0.14,0.1},=0.6,
and (3) calculating according to a comprehensive weight matching formula to obtain:
ω·={0.147,0.147,0.107,0.119,0.14,0.096,0.152,0.092},
let the overall similarity match threshold beE0.9, the overall matching degree is obtained according to the overall similarity matching formula as shown in fig. 12,
as shown in FIG. 12, all three resource services satisfy the integrated similarity match threshold, and resource service r1Has the maximum integrated similarity value, serves the most satisfied resource of the user, and finallyIs r1>r8>r6。
Comparative example 1
In order to verify the high efficiency of the matching method, the advantages and disadvantages of the matching method are evaluated from three dimensions of recall ratio, precision ratio and operation times, and compared with the existing intelligent searching and matching method for manufacturing cloud services on the basis of the embodiment 7,
wherein
tru is a resource service set meeting a resource demand party in the cloud manufacturing platform, Ans is a result of algorithm matching, and Tru n Ans is a resource service set meeting the resource demand party in the matching result.
The comparison result is shown in fig. 13, compared with the existing method, the precision ratio of the matching strategy is obviously improved, and the operation times are greatly reduced. Analyzing the matching process of the two algorithms can know that a preprocessing stage is added, clustering is carried out on the resource services before matching, the services with similar basic information are gathered into resource service clusters, when the services are matched, only the similarity between the clustering center of each resource service cluster and the basic information of a demand side is needed to be calculated to select alternative resource services, and all the resource services do not need to be matched with the demands one by one, so that the operation times are greatly reduced, and certain improvement is realized in different matching stages, so that the accuracy is higher. In the comprehensive matching stage, the weights of the sub-attributes are determined by comprehensively considering two aspects of historical records and user preferences, and then comprehensive matching sorting is performed, so that the satisfaction degree of the user on the matching result is improved.
The technical solutions described above only represent the preferred technical solutions of the present invention, and some possible modifications to some parts of the technical solutions by those skilled in the art all represent the principles of the present invention, and fall within the protection scope of the present invention.
Claims (8)
1. A cloud manufacturing platform resource service matching method comprises a model establishing step, a preprocessing step, a mapping matching step and an optimizing step, and is characterized in that the model establishing step, the preprocessing step, the mapping matching step and the optimizing step are sequentially carried out,
in the step of establishing the model, the model is established,
establishing a provider model:
RS={ID,BI,ST,FI,SI},
wherein the content of the first and second substances,
ID is an identification code; BI is basic information;
ST is state information, including idle state, not full load state, invalid state; FI is function information; the SI is service information;
establishing a demand side model:
RD={ID,DBI,DST,DFI,DSI},
wherein the content of the first and second substances,
DBI is basic information; DST is state information including an unexecuted state, an executing state, a completed state and a failure demand state; DFI is functional information;
the DSI is service attribute information that is,
in the preprocessing step, resource services similar to the text description of the basic information BI are aggregated into service clusters, and the service cluster most relevant to the basic information of the demand side is selected as a candidate resource service set, wherein the similarity of the text description comprises word form similarity, word order similarity, sentence length similarity and comprehensive similarity,
in the mapping and matching step, the state information ST, the function information FI and the service information SI of the alternative resource service selected by the pretreatment are respectively matched with the state information DST, the function information DFI and the service information DSI of the resource demand party, corresponding matching threshold values are set, the resource service meeting the matching threshold values is reserved, the resource service not meeting the matching threshold values is removed,
in the optimization step, comprehensive similarity matching is carried out on the resource service sets which are matched with the service information, sequencing is carried out according to the comprehensive similarity, and the resource with the maximum similarity is the optimal matching resource recommended by the system.
2. The cloud manufacturing platform resource service matching method according to claim 1, wherein in the preprocessing step, the comprehensive similarity sim of the basic information text is calculated according to the basic information and the number of resource service clusters of the service in the cloud service poolT(T1,T2) The method comprises the following specific steps:
the calculation formula of the word shape similarity is as follows:
simF(T1,T2)∈[0,1]the more words two texts are identical, the more similar the two texts are in terms of morphology, simF(T1,T2) The closer to 1.
The word order similarity calculation formula is as follows:
the sentence length similarity calculation formula is as follows:
simL(T1,T2)∈[0,1]the closer the lengths of the two texts are, simL(T1,T2) The closer to 1 the more the number of pixels is,
the comprehensive similarity calculation formula of the text is as follows:
simT(T1,T2)=α1sinF(T1,T2)+α2simO(T1,T2)+α3minL(T1,T2)
simT(T1,T2)∈[0,1]the closer the two texts are, simT(T1,T2) The closer to 1.
3. The cloud manufacturing platform resource service matching method according to claim 1, wherein in the step of aggregating the resource services into a service cluster, the basic information text similarity matrix between resource services can be expressed as:
wherein m is the number of all resource services in the cloud service pool;
Tiserve the ith riThe basic information text of (1);
Tjserve the jth rjThe basic information text of (1);
sim(Ti,Tj) And sim (T)j,Ti) Equal;
setting ith service r in cloud service pooliThe average similarity of the basic information text of other resource services is aiAnd then:
if all the services in the cloud service pool are clustered into q service clusters, if the theta service cluster CθHas a cluster center ofAnd the service cluster contains t services, the similarity E in the theta resource service clusterθComprises the following steps:
the evaluation function of the resource service cluster is:
the numerator represents the average intra-class similarity of all resource service clusters, the denominator represents the average inter-class similarity between resource service clusters,
when F is presentqWhen the value is maximum, the best clustering effect is achieved,
the algorithm flow of the source clustering is as follows:
step 1, initializing basic information of services in a cloud service pool, the number q of resource service clusters and the like.
And 2, calculating the comprehensive similarity of any two service basic information texts in the cloud service pool.
Step 3 calculating the ith service riAverage similarity a of basic information textsiSelecting the service with the maximum average similarity of the basic information texts as a first clustering center rC1Is selected fromC1The service with the least similarity serves as the second cluster center rC2。
And 4, the value range of the number q of the resource service clusters is more than or equal to 2 and less than or equal to m 1/2. When starting clustering, q is 2 and r isC1And rC2For the cluster center, calculate EθAnd FqThe resource service cluster with the minimum similarity in the cluster is selected, and the service with the minimum similarity to the cluster center is found in the resource service cluster and is used as the next cluster center.
And 5, q is q +1, if q is less than or equal to m1/2, the step 4 is carried out, and if not, the step 6 is carried out.
Step 6, calculating and comparing FqTo find maxFqAnd recording the size of q at the moment, and outputting a clustering result.
Determination of alternative resource service cluster:
setting theta cluster center of resource service clusterThe similarity with the basic information of the demand side isSelecting the resource service cluster with the maximum similarity as an alternative resource service cluster, wherein the alternative resource service set D is as follows:
by calculating the similarity of the basic information of each resource service cluster center, the matching of the basic information and all manufacturing services in a cloud service pool is avoided, and finally, an alternative service set related to the requirement is obtained.
4. The cloud manufacturing platform resource service matching method according to claim 1, wherein in the mapping matching step, an object is referred to a provider model and a demander model, specifically:
the function information and the service information of the resource service comprise a plurality of sub-attributes, the sub-attributes are independent from each other, and each sub-attribute module corresponds to an attribute value, so the disclosure further comprises:
each sub-function attribute value is { v }FI1,vFI2,...vFIγThe sub-attribute values are { v }SI1,vSI2,...vSIe},
For the demand side model, the introduced object elements are disclosed as follows:
the mapping matching is the matching of the Chinese character parameters and the numerical parameters of the Vs and the Vds, namely the calculation process of the similarity values of the Vs and the Vds.
5. The cloud manufacturing platform resource service matching method according to claim 4, wherein in the mapping matching step, the status information of the provider and the demander is matched, and the specific steps are as follows:
Step 2. resource set D for completing state information matchings;
Step 3, initial state orderkD1, wherein kDIs the sequence number of the resource in D, nDThe total number of the alternative resources;
Step 5.kD=kD+1, if kD≤nDTurning to the step 4, otherwise, turning to the step 6;
step 6, finishing the algorithm, and outputting the resource service set D matched with the completion state informations。
6. The cloud manufacturing platform resource service matching method according to claim 4, wherein in the mapping matching step, the function information is matched based on calculation of the similarity between the concept similarity and the similarity between the numerical parameters in the function information of the provider and the demander, and the specific steps are as follows:
in the calculation of the concept similarity, the semantic distance and the semantic overlap ratio are respectively calculated, then the concept similarity is finally calculated according to the calculation result,
in the semantic distance calculation, an ontology tree of the concept is established, and the weight W expression of an edge between a node of the nth layer of concept of the ontology tree and a node of the (n-1) th layer of concept is as follows:
the formula for calculating the semantic distance is as follows:
the calculation formula of the semantic overlap ratio is as follows:
according to the calculation structure of the semantic distance and the semantic overlap ratio and the weight of the semantic distance and the semantic overlap ratio, the concept similarity simCThe formula for the calculation of (Ci, Cj) is:
when the algorithm is applied to the functional information matching process, the concept semantic similarity formula is as follows:
the numerical parameter similarity calculation formula is as follows:
the calculation formula for matching the functional information similarity of the supply and demand parties is as follows:
based on the calculation formula of the functional information similarity matching, the matching steps are as follows:
step 1, inputting: resource service set D matched by state informations={ds1,ds2...dsk...dsnThe matching threshold of the function information isF;
And step 2, outputting: resource service set D for completing function information matchingF;
Step 3, initial state orderk is 1, wherein k is DsThe serial number of the medium resource, n is DsThe total number of medium resource services;
step 4, calculating simG (d) of similarity between the function information of the kth resource provider and the function information of the resource demander by adopting a calculation formula matched with the similarity of the function informationskFI, DFI), if simG (d)sk.FI,DFI)≥FThen D isF=DF∪dsk,
Step 5, k is equal to k +1, if k is equal to or less than n, the step 4 is carried out, and if not, the step 6 is carried out;
step 6, finishing the algorithm, and outputting the resource service set D matched with the completion state informationF。
7. The cloud manufacturing platform resource service matching method according to claim 4, wherein in the mapping matching step, normalization processing is performed on a plurality of service information to obtain an attribute value of the information service, and then matching between the provider and the demander information services is achieved according to a fuzzy number calculation formula, and the specific steps are as follows:
service information to providersService information with a requesting partyMatching is carried out, and the fuzzy number formula is as follows:
if a decision scheme includes multiple service information, i.e. ThenThe fuzzy number calculation formula is as follows:
in the normalization processing step, the normalization processing formula is as follows:
let SIuvIs expressed as a fuzzy number ofIn the same way, SIuvIs expressed as a fuzzy number ofThe service information attribute value of the resource demander may be expressed as DSI ═ DSI1,DSI2...DSIeAnd v-dimension attribute value DSI in the normalized requirementvThe fuzzy number of (a) means that the result isThe calculation formula for obtaining the service information matching between the provider and the demander according to the fuzzy number formula is as follows:
based on the calculation formula for matching the service information of the demand party, the following steps are adopted to complete the matching of the service information:
step 1, inputting: the resource service set matched by the function information is DF={df1,df2...dfu...dfnFService information matching thresholds;
And step 2, outputting: completing a resource service set S matched with the service information;
step 3, initial state:u is 1, wherein u is DFNumber of medium resource, nFIs DFThe total number of the medium alternative resources;
step 4, for DFNormalizing the service information attribute value of the medium resource;
step 5. calculate the u-th dataSimilarity sim between source service and demander service informations(dfuSI, DSI), if sims(dfu.SI,DSI)≥sThen S ═ S ≡ U ∞ dfu;
And 6.u is equal to u +1, and if u is less than or equal to nFGo to step 5, otherwise go to step 7;
and 7, finishing the algorithm, and outputting the resource service set S matched with the service information.
8. The cloud manufacturing platform resource service matching method according to claim 1, wherein in the optimization step, objective weights and subjective weights of the comprehensive similarity are calculated respectively, and then the comprehensive similarity matching is performed, and the specific steps are as follows:
in the calculation of the objective weight,
let the tetrad I ═ (B, a ═ C ═ D, V, F) be the cloud manufacturing platform resource service information system.
Let X and Y be non-empty subsets of the finite discourse domain, have In a partial order relationship, then
Beta is more than or equal to 0 and less than 0.5, existIf c (X, Y) ═ 0, then X is included by the Y criteria,
let X be B/C be X1,X2...X|B/C|B is an equivalence class derived from each sub-attribute C, Y ═ B/D ═ Y1,Y2...Y|B/D|B is an equivalence class obtained by dividing through a decision attribute set D,
in the case of β ∈ [0, 0.5)),Yjthe following approximate distribution of β for C is:
for arbitrary sub-attributes apInformation amount γ (a) thereofp) The calculation method of (2) is as follows:
for arbitrary sub-attributes apWhich is dependent on a degree lambda (a)p) The calculation method of (2) is as follows:
gamma (a) of each sub-attributep) And λ (a)p) The importance degree of the sub-attribute is described from different angles, and the calculation formula of the objective weight of each sub-attribute can be obtained by comprehensively considering the importance degree of the sub-attribute:
in the step of calculating the subjective weight, the subjective weight is calculated,
setting the subjective weight of a resource demander asThen each sub-attribute ω'pThe comprehensive weight calculation method comprises the following steps:
and ∈ [0,1 ].
The comprehensive similarity matching formula is as follows:
simZ(RS,RD)=ω′1sim1+ω′2sim2+...+ω′osimo
ω′1+ω′2+...+ω′o=1,
the matching steps are as follows:
And step 2, outputting: resource service set S finally matched with requirementE,
Step 3, initial state:kZ=1,kZsequence number, n, for resource service in SZThe number of resource services in S is counted,
step 4, calculating the k-th time in SZComprehensive similarity value of individual resource provider and demanderIf it isThen
Step 5.kZ=kZ+1, if kZ≤nZGo to step 4, otherwise go to step 6,
step 6, finishing the algorithm, and finally obtaining the resource service set S matched with the requirementE。
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010699532.4A CN112085458A (en) | 2020-07-20 | 2020-07-20 | Cloud manufacturing platform resource service matching method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010699532.4A CN112085458A (en) | 2020-07-20 | 2020-07-20 | Cloud manufacturing platform resource service matching method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN112085458A true CN112085458A (en) | 2020-12-15 |
Family
ID=73735471
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010699532.4A Pending CN112085458A (en) | 2020-07-20 | 2020-07-20 | Cloud manufacturing platform resource service matching method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112085458A (en) |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101820444A (en) * | 2010-03-24 | 2010-09-01 | 北京航空航天大学 | Descriptive information matching similarity-based resource service matching and searching method |
CN103984714A (en) * | 2014-05-07 | 2014-08-13 | 湖北工业大学 | Ontology semantics-based supply and demand matching method for cloud manufacturing service |
CN106570080A (en) * | 2016-10-18 | 2017-04-19 | 河海大学常州校区 | Multilevel semantic matching method for cloud manufacturing resource services |
CN109063171A (en) * | 2018-08-20 | 2018-12-21 | 同济大学 | Semantic-based reso urce matching method |
CN109636168A (en) * | 2018-12-05 | 2019-04-16 | 北京工业大学 | A kind of cloud platform scheduling resource intelligent matching process based on Ontology |
-
2020
- 2020-07-20 CN CN202010699532.4A patent/CN112085458A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101820444A (en) * | 2010-03-24 | 2010-09-01 | 北京航空航天大学 | Descriptive information matching similarity-based resource service matching and searching method |
CN103984714A (en) * | 2014-05-07 | 2014-08-13 | 湖北工业大学 | Ontology semantics-based supply and demand matching method for cloud manufacturing service |
CN106570080A (en) * | 2016-10-18 | 2017-04-19 | 河海大学常州校区 | Multilevel semantic matching method for cloud manufacturing resource services |
CN109063171A (en) * | 2018-08-20 | 2018-12-21 | 同济大学 | Semantic-based reso urce matching method |
CN109636168A (en) * | 2018-12-05 | 2019-04-16 | 北京工业大学 | A kind of cloud platform scheduling resource intelligent matching process based on Ontology |
Non-Patent Citations (2)
Title |
---|
张健: "云制造平台资源服务匹配方法研究", 中国优秀硕士学位论文全文数据库 信息科技辑, no. 7, pages 16 - 27 * |
贾海利;张健;祁宇明;邓三鹏;: "云制造平台资源需求的高效匹配策略研究", 现代制造工程, no. 04, pages 36 - 43 * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107515938B (en) | Intelligent supply and demand matching method in cloud manufacturing environment | |
CN111324642A (en) | Model algorithm type selection and evaluation method for power grid big data analysis | |
CN108921604B (en) | Advertisement click rate prediction method based on cost-sensitive classifier integration | |
CN109784722A (en) | Web service selection method and system based on user preference | |
CN110147808A (en) | A kind of novel battery screening technique in groups | |
CN109686402A (en) | Based on key protein matter recognition methods in dynamic weighting interactive network | |
CN112148986A (en) | Crowdsourcing-based top-N service re-recommendation method and system | |
CN112070136A (en) | Method for classifying unbalanced data based on boost decision tree and improved SMOTE | |
CN114781717A (en) | Network point equipment recommendation method, device, equipment and storage medium | |
Chen et al. | Efficient online ml api selection for multi-label classification tasks | |
CN114580847A (en) | Energy efficiency assessment service resource supply and demand matching system and method considering similarity and credibility | |
CN107066328A (en) | The construction method of large-scale data processing platform | |
CN116663964B (en) | Engineering unit price rapid calculation method and system based on characteristic values of list items | |
Zhong et al. | Parallel machine scheduling with restricted job rejection | |
CN112085458A (en) | Cloud manufacturing platform resource service matching method | |
CN115619236A (en) | Decision method based on hesitation fuzzy diffusion decision model and related device | |
CN113095604B (en) | Fusion method, device and equipment of product data and storage medium | |
CN111127184B (en) | Distributed combined credit evaluation method | |
CN108256083A (en) | Content recommendation method based on deep learning | |
CN111125541B (en) | Method for acquiring sustainable multi-cloud service combination for multiple users | |
CN112464101A (en) | Electronic book sorting recommendation method, electronic device and storage medium | |
CN110110130A (en) | Personalized labels recommended method and system based on convolution feature Yu Weighted random migration | |
CN117638950B (en) | Electricity utilization strategy recommendation method and device, electronic equipment and storage medium | |
CN111709532B (en) | Online shopping representative sample selection system based on model-independent local interpretation | |
CN112988564B (en) | SRGM decision model considering cost-reliability and construction method thereof |
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 |