CN102184238A - Vector-retrieval-supported web service management method - Google Patents

Vector-retrieval-supported web service management method Download PDF

Info

Publication number
CN102184238A
CN102184238A CN 201110125682 CN201110125682A CN102184238A CN 102184238 A CN102184238 A CN 102184238A CN 201110125682 CN201110125682 CN 201110125682 CN 201110125682 A CN201110125682 A CN 201110125682A CN 102184238 A CN102184238 A CN 102184238A
Authority
CN
China
Prior art keywords
vector
notion
dimension
web service
method vector
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
Application number
CN 201110125682
Other languages
Chinese (zh)
Inventor
彭敦陆
李云先
王晓玲
周傲英
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
University of Shanghai for Science and Technology
Original Assignee
University of Shanghai for Science and Technology
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by University of Shanghai for Science and Technology filed Critical University of Shanghai for Science and Technology
Priority to CN 201110125682 priority Critical patent/CN102184238A/en
Publication of CN102184238A publication Critical patent/CN102184238A/en
Pending legal-status Critical Current

Links

Images

Abstract

The invention discloses a vector-retrieval-supported web service management method which comprises the following steps: S1, carrying out vectorization on the corresponding description information of Web services by using a word frequency TF/reverse file frequency IDF method; S2, generating a method clustering background; S3, generating a k-dimensional method vector; and S4, generating a vector service lattice. The vector-retrieval-supported web service management method disclosed by the invention is characterized in that a vector service lattice based on a method vector concept is established through fully using the vector description and formal concept analysis of Web services, thereby realizing the method for retrieving the Web services in the vector service lattice by adopting a vector-based method. The inventor fully proves the feasibility of the method from the aspects of theory and experiment, and the method has superior performance. The advantages of the method disclosed by the invention are not possessed by other techniques and methods; and the technique disclosed by the invention fills the gaps in related research and technique fields, and has a wide practical application value.

Description

The Web service management method of support vector retrieval
Technical field
The invention belongs to service compute and Web data management technique field, relate to a kind of Web service management method, relate in particular to a kind of Web service management method of support vector retrieval.
Background technology
Web service is that those can use some standards based on XML (as SOAP, UDDI and WSDL etc.), the separate modular assembly that is described, issues, locatees and call on Internet.Web service is different from traditional dynamic Web page: the purpose of using Web service can be called assembly by other program or application for information integration, is " machine-readable "; The latter is " people is readable ", for the people directly serves.Web service depends on loose, dynamic Web assembly, service requester only need utilize interface messages such as announced service behavior, binding and I/O parameter, call by standardized XML message mechanism, and need not know the detail that service realizes.In other words, Web service be with operating system, programming language mutually independently.The ISP issues Web service standard according to the rules with the relevant interface message of service with the XML form in service register center, Web service just can serviced caller discovery and called.
Along with developing rapidly of ecommerce, more and more enterprises is transferred to its core business on the Internet.The service mode of ecommerce to B2B (Business-to-Business) pattern, also develops into data sharing and information integration in the Internet scope from simple Web information from B2C (Business-to-Customer) mode development technically.Like this, also just have higher requirement for rapidity, validity and reliability to the commercial systems for applications on the Internet.Under this background, the Web service technology is arisen at the historic moment.The increasing application shows that Web service has become the effective means that Web goes up data and information integration as a kind of novel distributed computing platform.
Service Management is as the Web service technology key link, along with the increase of Web service value volume and range of product, to the attention that has obtained industry.The Web service management can be regarded three levels as: infrastructure layer, application layer and operation layer.The infrastructure layer emphasis is to guarantee operating in the acceptable performance of carrying out from the Web service of different platform carried out, and its center of gravity is the Web service platform; Application layer refers to the Web service self-management, as tissue, data management, quality management and the access control management of information on services (during as state variation, particularly break down when service and how to handle when unavailable) etc.Operation layer refers to be organized Web service to finish the business task of expection, the combination of Web service just according to service logic.Research for infrastructure layer, develop quite rapid, existing relevant achievement forms, and relevant software systems have been developed, the Unicenter WSDM of the more famous Openview that Hewlett-Packard Corporation is arranged, CA company and OASIS thereof (Web Services Distributed Management) and WebMethods have proposed OMI (Open Management Interface), also have other major company, also all announced to support WSDM as manufacturers such as Microsoft, BEA, Sun, Collaxa, DataPower, JBoss, Systinet.
At present, in UDDI, Web service is managed by mode classification.This Classification Management is by ISP's class of service according to oneself when the Web Service Registration service to be classified.Organize Web service to have following deficiency with this traditional mode classification: (1) as a rule, a service can be finished a plurality of functions, if and these functions have very big difference on classification, which classification the service just can't accurately determined in service layer so belongs to; (2) Web service of organizing based on this mode because do not effectively utilize semantic relation between service, particularly under the situation that Web service quantity increases progressively in a large number, is difficult to realize Web service accurately and is efficiently retrieved; (3) because the Traditional Web services management method, just come Web service is organized with coarseness (service layer), and the demand that service requester will satisfy is to be realized by the method in the service, and traditional Web service Classification Management technology can not satisfy this retrieval requirement.For (1) individual weak point, a feasible solution is a Web service to be registered in a plurality of classifications according to own commercial object and the function that service provides by the ISP, but there is bulk redundancy in this Web service information that will cause service register center to be deposited.For (2), (3) individual deficiency, can't solve in present service register center.
The present invention realizes the management to Web service on the method for servicing layer.Specifically be to utilize the k-dimensional vector that service grid is expanded, set up vectorial service grid, choose strategy for the discovery algorithm and the method for servicing lexical item of method vector notion in the outgoing vector service grid.Utilize in the vectorial service grid the conceptual clustering information of method of servicing, expand, propose service search method based on vectorial service grid to serving vector.
Summary of the invention
Technical matters to be solved by this invention is: a kind of Web service management method of support vector retrieval is provided, supports the Web service retrieval based on vector effectively.
For solving the problems of the technologies described above, the present invention adopts following technical scheme:
A kind of Web service management method of support vector retrieval, described method comprises the steps:
Step S1, the reverse file frequency of employing word frequency TF/ IDF method are carried out vectorization to the corresponding descriptor of Web service;
Step S2, production method vector cluster background step; A method vector cluster background is a tlv triple
Figure BDA0000061427180000031
Figure BDA0000061427180000032
With
Figure BDA0000061427180000033
Be two finite sets, represent method of servicing collection and indexing means vector set respectively;
Figure BDA0000061427180000034
Relation between expression method of servicing collection and the indexing means vector set; For each Here
Figure BDA0000061427180000036
Figure BDA0000061427180000037
And
Figure BDA0000061427180000038
Expression
Figure BDA0000061427180000039
In each method p jCan pass through the method vector Carry out index;
Step S3, generation k-dimension method vector step; Comprise: step S31, find all 1-dimension methods vector notions; Step S32, generation k-dimension method vector notion;
Step S4, the vectorial service grid step of generation; If
Figure BDA00000614271800000311
It is method vector cluster background
Figure BDA00000614271800000312
In, by the k-dimension method vector notion that the attended operation of vectorial service grid is derived, then complete lattice Be called vectorial service grid;
If
Figure BDA00000614271800000314
Be a method vector cluster background, to each
Figure BDA00000614271800000315
Figure BDA00000614271800000316
The attended operation method representation of vector service grid is as follows:
Figure BDA00000614271800000317
Figure BDA00000614271800000318
As a preferred embodiment of the present invention, described step S4 comprises:
Step S41: scan one time
Figure BDA00000614271800000319
Produce candidate 1-dimension method vector concept set CVC 1
If
Figure BDA00000614271800000320
Be a method vector cluster background,
Figure BDA00000614271800000321
It is a method vector set;
Figure BDA00000614271800000322
Be called
Figure BDA00000614271800000323
The territory, be designated as
Figure BDA00000614271800000325
Be called
Figure BDA00000614271800000326
Closure, be designated as
Figure BDA00000614271800000327
Figure BDA00000614271800000328
In all divide vectors all to get The middle minimum value that occurs; If
Figure BDA00000614271800000330
And
Figure BDA00000614271800000331
The time, then claim
Figure BDA00000614271800000332
For about the territory
Figure BDA00000614271800000333
Method vector cluster background
Figure BDA00000614271800000334
Under method of servicing vector notion VC, abbreviate method vector notion as; If k item component is arranged in the vector, then claim
Figure BDA00000614271800000335
Be method vector cluster background
Figure BDA00000614271800000336
Under k-dimension method of servicing vector notion k-VC, abbreviate the vectorial notion of k-dimension method as;
Have only when a method vector equals group of methods vector notion closure of a set, be only a method vector; Therefore,, ask for closure by the territory then, and judge whether this closure exists in 1-dimension method vector centralized concept, if just do not add by asking for the territory of each candidate 1-dimension method vector earlier; After 1-dimension method vector notion had been arranged, algorithm was finished the generation of k+1-dimension method vector notion in two steps;
Step S42: produce candidate k+1-dimension method vector concept set CVC K+1
Judge whether a candidate k+1-method vector notion is k+1-method vector notion; Return at last
Figure BDA0000061427180000041
-Wei method vector notion; Here,
Figure BDA0000061427180000042
Expression
Figure BDA0000061427180000043
In the speech phase number that contains.
As a preferred embodiment of the present invention, described method further comprises step S5:Web service searching step; Comprise:
The similarity of step S51, calculation services method vector and query vector is determined the method greater than the threshold value of judgement Query Result given in advance, and with its reservation;
All methods vector notions that step S52, calculating remain and the similarity of query vector are if similarity is greater than threshold value, then for each method in the method vector notion territory, the similarity of computing method vector and query vector; If greater than threshold value, then this method vector is joined result set, otherwise it is joined in the non-result set; Handle all method vector notions, enter step S53;
Step S53, for the method vector of not visiting, promptly neither in result set, again not in non-result set, repeating step S52 then is up to handling all method vectors.
As a preferred embodiment of the present invention,, in realizing the Web service management,, avoid the few lexical item of those quantity of information to enter the feature lexical item collection based on method selected characteristic lexical item from the information source of Web service method of the heavy lexical item collection of k-right-safeguarding for reducing the calculating scale.
As a preferred embodiment of the present invention, for reducing the calculating scale, in realizing the Web service management, not method vector notion if there is certain subvector in certain k-dimension method vector, this method vector just is not a candidate method vector notion so; Utilize this feature, reduce the search volume that produces the alternative approach vector effectively.
A kind of Web service management method of support vector retrieval, described method comprises the steps:
Step S1, the reverse file frequency of employing word frequency TF/ IDF method are carried out vectorization to the corresponding descriptor of Web service;
Step S2, production method cluster background;
Step S3, generation k-dimension method vector; Comprise: step S31, find all 1-dimension methods vector notions; Step S32, generation k-dimension method vector notion;
Step S4, generate vectorial service grid.
Beneficial effect of the present invention is: the Web service management method of the support vector retrieval that the present invention proposes, make full use of Web service vector description and form concept analysis and set up a vectorial service grid, realized in the method for vectorial service grid employing based on the method retrieval Web service of vector based on method vector notion.The inventor has fully proved the feasibility of this technical method, and has had superior performance from theoretical and two aspects of experiment.This is that other technologies and method are not available, and this technology has been filled up the blank of correlative study and technical field, and has actual application value widely.In the medicine and hygiene fields project, obtained successful application.
Description of drawings
Fig. 1 is and the corresponding vectorial service grid synoptic diagram of table 1.
Fig. 2 is the process flow diagram of Web service management method of the present invention.
Embodiment
Describe the preferred embodiments of the present invention in detail below in conjunction with accompanying drawing.
Embodiment one
At first explanation some notion and definition relevant with the present invention:
[definition 1] (method vector) establishes p iFor being a method of Web service, in the service vector model, method of servicing vector (abbreviation method vector)
Figure BDA0000061427180000051
Represent with following formula:
Figure BDA0000061427180000052
Here, w IjExpression lexical item t iAt method of servicing p iIn significance level tolerance (or weight), w IjGet on the occasion of.K is a lexical item sum indexed in the system, the method vector
Figure BDA0000061427180000053
Also claim k-dimension method vector.Intuitively, sometimes also with the method vector
Figure BDA0000061427180000054
Be designated as
Figure BDA0000061427180000055
Method p iW IjValue is to utilize p iDescriptor (comprising the I/O message name of Service name, service description information, method name, method, the textual description of method) try to achieve by the TF/IDF method:
w ij = tf ij log N P j - - - ( 2 )
Here tf IjBe lexical item t jAt method p iWord frequency in the descriptor, N is the method sum that contains in the service library, P jRefer to comprise lexical item t in the descriptor jThe method of servicing number.With following formula to w IjCarry out standardization:
w , ij = w ij max 1 w i 1 - - - ( 3 )
Max wherein 1w I1Get
Figure BDA0000061427180000062
In former subvector maximal value.
A method of servicing vector cluster background of [definition 2] (method vector cluster background) (abbreviation method vector cluster background) is a tlv triple
Figure BDA0000061427180000063
With
Figure BDA0000061427180000065
Be two finite sets, represent method of servicing collection and indexing means vector set respectively.
Figure BDA0000061427180000066
Relation between method for expressing collection and the vector set.For each
Figure BDA0000061427180000067
Here
Figure BDA0000061427180000068
Figure BDA0000061427180000069
And
Figure BDA00000614271800000610
Expression In each method p jCan pass through the method vector
Figure BDA00000614271800000612
Carry out index.
[definition 3] (connection of vectorial service grid) establishes
Figure BDA00000614271800000613
Be a method vector cluster background, to each
Figure BDA00000614271800000614
Figure BDA00000614271800000615
The following operation of definition:
Figure BDA00000614271800000616
Figure BDA00000614271800000617
Formula (4) is called the connection of vectorial service grid.
[proposition 1] establishes
Figure BDA00000614271800000618
Following equation is set up:
Figure BDA00000614271800000619
[definition 4] (k-dimension method vector notion) established
Figure BDA00000614271800000620
Be a method vector cluster background,
Figure BDA00000614271800000621
It is a method vector set.
Figure BDA00000614271800000622
Be called
Figure BDA00000614271800000623
The territory, be designated as
Figure BDA00000614271800000624
Figure BDA00000614271800000625
Be called
Figure BDA00000614271800000626
Closure, be designated as
Figure BDA00000614271800000627
Figure BDA00000614271800000628
In all divide vectors all to get
Figure BDA00000614271800000629
The middle minimum value that occurs.If
Figure BDA00000614271800000630
And
Figure BDA00000614271800000631
The time, then claim
Figure BDA00000614271800000632
For about the territory
Figure BDA00000614271800000633
Method vector cluster background
Figure BDA00000614271800000634
Under method of servicing vector notion (or VC), abbreviate method vector notion as, if k item component is arranged in the vector, then claim
Figure BDA00000614271800000635
Be method vector cluster background
Figure BDA00000614271800000636
Under k-dimension method of servicing vector notion (or k-VC), abbreviate the vectorial notion of k-dimension method as.
[definition 5] (vectorial service grid) established
Figure BDA00000614271800000637
It is method vector cluster background
Figure BDA00000614271800000638
In, by the k-dimension method vector notion that the attended operation [definition 3] of vectorial service grid is derived, then complete lattice
Figure BDA00000614271800000639
Be called vectorial service grid, be called for short
Figure BDA00000614271800000640
Figure BDA00000614271800000641
Satisfy following character:
Figure BDA00000614271800000642
In the method vector satisfy partial ordering relation.If
Figure BDA00000614271800000643
Be vectorial service grid
Figure BDA00000614271800000644
Two methods vector notions, be designated as
Figure BDA00000614271800000645
And if only if at each t iSatisfy The time, This individual character represents that dual relationships is satisfied in any means vector territory corresponding with it in vectorial service grid:
Figure BDA0000061427180000071
Or
Figure BDA0000061427180000073
Random subset In method vector notion, have a maximum to have element, be called connection, be designated as
Figure BDA0000061427180000075
The total element of minimum is called friendship, is designated as By the fundamental theorem of Galois concept lattice, can be right with following formula With
Figure BDA0000061427180000078
Define:
Figure BDA0000061427180000079
Figure BDA00000614271800000710
Be not difficult to find out from 5 formulas,
Figure BDA00000614271800000711
All former subcomponents on all exist
Figure BDA00000614271800000712
Get minimum value, and
Figure BDA00000614271800000713
Institute importantly all exist
Figure BDA00000614271800000714
In get maximal value.
[definition 6] (minimum vectorial service grid) established
Figure BDA00000614271800000715
Be by method vector cluster background
Figure BDA00000614271800000716
A vectorial service grid, if
Figure BDA00000614271800000717
In any two methods vector notion
Figure BDA00000614271800000718
Satisfy: i) if
Figure BDA00000614271800000719
Then
Figure BDA00000614271800000720
Ii) if
Figure BDA00000614271800000721
And
Figure BDA00000614271800000722
Figure BDA00000614271800000723
Then
Figure BDA00000614271800000724
Then claim vectorial service grid
Figure BDA00000614271800000725
Be about method vector cluster background
Figure BDA00000614271800000726
Minimum vectorial service grid.The vector back of the body among the present invention all refers to the minimum cluster background about certain method vector cluster background.
In general, utilize the partial ordering relation in the above-mentioned character can give the visual expression of outgoing vector service grid: given two method vectors
Figure BDA00000614271800000727
Satisfy
Figure BDA00000614271800000728
And In do not have the 3rd method vector notion
Figure BDA00000614271800000730
Make
Figure BDA00000614271800000731
Set up, exist so
Figure BDA00000614271800000732
Between just continuous with a limit.Here, claim
Figure BDA00000614271800000733
Quilt
Figure BDA00000614271800000734
Cover, also claim
Figure BDA00000614271800000735
For
Figure BDA00000614271800000736
Parents' notion.
Based on above-mentioned key concept, give outgoing vector service grid production method, comprise that method vector generation algorithm (VC4WS) reaches the service retrieval based on vector.
At first, provide method vector production method (VC4WS), it is a kind of algorithm of Apriori style, mainly contains for two steps: the first step is to find all 1-dimension method vector notions; Second step was to produce k+1-dimension method vector notion.Arthmetic statement is as follows:
The first step: scan one time
Figure BDA00000614271800000737
Produce candidate 1-dimension method vector concept set CVC 1
By the definition of method vector notion, have only as can be known when a method vector equals group of methods vector notion closure of a set, be only a method vector notion [definition 4].Therefore, algorithm is asked for closure by the territory then by asking for the territory of each candidate 1-dimension method vector earlier, and judges whether this closure exists in 1-dimension method vector centralized concept, if just do not add.After 1-dimension method vector notion had been arranged, algorithm was finished the generation of k+1-dimension method vector notion in two steps.
Second step: produce candidate k+1-dimension method vector concept set CVC K+1
Similar with the generation of 1-dimension method vector notion, utilize the definition [definition 4] of method vector notion to judge whether a candidate k+1-method vector notion is k+1-method vector notion.Return at last
Figure BDA0000061427180000081
-Wei method vector notion.Here, Expression
Figure BDA0000061427180000083
In the speech phase number that contains.
Introduce the vector approach that in vectorial service grid, uses that proposes among the present invention below and realize the search method of Web service.
The target of Web service retrieval is to concentrate from Web service to find the Web service of satisfying query requests, and these query requests have been described demand for services information.According to the definition [definition 4] of k-dimension method vector notion, satisfy following relationship at Web service method and this method vector of occurring in the territory of Same Way vector notion: the weight of their method vector on each former subcomponent all is not less than the weight of corresponding former subcomponent in the method vector notion.In order to utilize this information when the retrieval Web service, each method vector is expanded, make it comprise two parts: one is the original method vector, another is the method vector notion that includes this method of servicing in the territory.Like this, can write out the method vector
Figure BDA0000061427180000084
New expression formula:
Figure BDA0000061427180000085
In the formula
Figure BDA0000061427180000086
It is k-weight method vector.(ranking list of VCs) is to comprise method of servicing p in all territories iK-dimension method vector notion, by they with
Figure BDA0000061427180000087
The similarity rank, with
Figure BDA0000061427180000088
The big more front more that comes of similarity.Assumption method vector notion
Figure BDA00000614271800000810
And
Figure BDA00000614271800000811
We use With Included angle cosine represent With
Figure BDA00000614271800000815
Similarity:
Figure BDA00000614271800000816
Because, at the indexing means vector set
Figure BDA00000614271800000817
In each method vector have only the k item, and method vector notion is the vector on the lexical item collection that obtains after will all k-methods vectors merging, the former subvector that it contains is not less than k.For the ease of expressing, in following formula, we suppose
Figure BDA00000614271800000818
With
Figure BDA00000614271800000819
Preceding k item identical, the lexical item that begins from the k+1 item only appears at method vector notion
Figure BDA00000614271800000820
In, and method
Figure BDA00000614271800000821
In do not have, therefore, its weight is 0.Clearly, Big more, method vector notion
Figure BDA00000614271800000823
With the method vector
Figure BDA00000614271800000824
Forward more in list.
For the lexical item weight w among the inquiry q I, q, calculate [BHM+04] with following formula:
w i , q = ( 0.5 + 0.5 tf i , q m ax 1 tf 1 , q ) × log N P j - - - ( 9 )
In the formula (9), tf I, qExpression lexical item t iThe frequency that in the text message of inquiry q, occurs.N and P jWith the same meaning in the formula (2).The vector representation of q is called query vector, is designated as
Figure BDA0000061427180000091
Equally, right
Figure BDA0000061427180000092
Also the same with the method vector, k-item before getting.
The territory that can be easy to obtain a method of servicing which method vector notion in vectorial service grid by the method vector representation after the expansion occurs, and simultaneously, also is easy to try to achieve other method in this territory.By the definition of vectorial service grid, we find, if method vector notion is similar more to the method vector, then the method for servicing that contains in the pairing territory of method vector notion is also just similar more.Therefore, the method vector sum query vector after the expansion has been arranged, had divided for three steps finished and retrieve based on the Web service of vectorial service grid:
Step 1: calculation services method vector
Figure BDA0000061427180000093
With query vector
Figure BDA0000061427180000094
Similarity
Figure BDA0000061427180000095
If
Figure BDA0000061427180000096
Wherein, s is the threshold value of judgement Query Result given in advance, then with method p iJoin in the result set (Positive Set), otherwise p iJoin in the non-result set (Negative Set), entered for second step then;
Step 2: calculate all the method vector notions in (Ranking list of VCs) With
Figure BDA0000061427180000098
Similarity
Figure BDA0000061427180000099
If Then for each method
Figure BDA00000614271800000911
Neither in result set, not in non-result set, then calculate again
Figure BDA00000614271800000912
With
Figure BDA00000614271800000913
Similarity, If
Figure BDA00000614271800000915
With p jJoin result set, otherwise with p jJoin in the non-result set.Handle all method vector notions, entered for the 3rd step.
Step 3: repeated for second step for the method vector of not visiting (neither in result set, again not in non-result set), up to handling all method vectors.
This method (certainly, needs under the situation of outermost to calculate) as a rule, need not calculate the similarity of query vector and all method vectors, and only need calculating section and inquiry similar methods, expresses obtaining other similarity method by the expansion of method vector.For the method for servicing that retrieves, can carry out rank according to the similarity of they and query vector.
See also shown in Figure 1ly, introduce the specific implementation process of the Web service management method of support vector of the present invention retrieval below in conjunction with real example in detail:
1) case introduction of method vector cluster background and vectorial service grid:
Table 1 is a method vector cluster background example.Each method is (from p 1To p 4, in the table the 1st row) the method vector by weight vectors t 1To t 4(the 2nd is listed as the 6th row) expression, for example, p 1The method vector be expressed as
Figure BDA0000061427180000101
Each w ' wherein IjMethod for expressing p iMiddle lexical item t jWeight, can try to achieve by through type (3).Last row are Web services of supporting corresponding method in first row in the table.
Method t 1 t 2 t 3 t 4 t 5 Web service
p 1 0.5 0.7 1 0.8 0.6 ws 1
p 2 1 0.8 0.6 0.7 1 ws 2
p 3 0.6 1 0.6 0.1 0.8 ws 3
p 4 0.7 0.3 1 0.6 0.7 ws 4
Table 1: the example of a vectorial cluster background of service
Figure BDA0000061427180000102
Table 2: the k-dimension method vector list of notion 2 corresponding with table 1 is the 5-dimension direction vector notions that obtain according to [definition 4].By table as seen, with territory { p 2, p 3, p 4Corresponding method vector set is
Figure BDA0000061427180000111
Figure BDA0000061427180000112
It is a 5-dimension method vector.Notice that the value of each component must be the value that occurs in the method vector set in the closure in a territory, vector
Figure BDA0000061427180000113
The territory be { p 2, p 3, p 4And But not about territory { p 2, p 3, p 45-dimension method vector because
Figure BDA0000061427180000115
Not at the method vector set
Figure BDA0000061427180000116
The middle appearance.(note,
Figure BDA0000061427180000117
Expression α 〉=β.Any two method vectors
Figure BDA0000061427180000118
If at each former subcomponent t i,
Figure BDA0000061427180000119
All set up, we just say
Figure BDA00000614271800001110
At the further part of this chapter, t i＞t jRepresent in the ti lexicographic order greater than t j).
Fig. 1 is the visual expression of method vector notion among the table 5-2.If From Fig. 1, we can be clear that,
Figure BDA00000614271800001112
Each former subcomponent is got
Figure BDA00000614271800001113
Minimum value;
Figure BDA00000614271800001114
Because what we discussed is minimum vectorial service grid, owing to there is another method vector notion
Figure BDA00000614271800001115
Satisfy
Figure BDA00000614271800001116
Therefore, get
Figure BDA00000614271800001117
As VC 5And VC 9Friendship.
2) be to adopt the vector index method to search the Web service example at vectorial service grid
For example, according to formula (7), contain method p in the territory in the table (2) 1Method vector notion have
Figure BDA00000614271800001119
Figure BDA00000614271800001120
Figure BDA00000614271800001122
Utilize formula (8) try to achieve they with Similarity be respectively 0.88,0.96,0.89,0.97,0.98 and 1.0, therefore, the order in list is followed successively by (VC 5, VC 5, VC 4, VC 2, VC 3, VC 1).Like this, by formula (7),
Figure BDA00000614271800001125
Again be expressed as:
Figure BDA00000614271800001126
Will Output just obtains at the vector of the method shown in the table 2 notion, p 1Approaching with Search Requirement.
Embodiment two
See also Fig. 2, the present invention has disclosed a kind of Web service management method of support vector retrieval, and described method comprises the steps:
Step S1, the reverse file frequency of employing word frequency TF/ IDF method are carried out vectorization to the corresponding descriptor of Web service;
Step S2, production method vector cluster background step; A method vector cluster background is a tlv triple
Figure BDA0000061427180000122
With
Figure BDA0000061427180000123
Be two finite sets, represent method of servicing collection and indexing means vector set respectively;
Figure BDA0000061427180000124
Relation between expression method of servicing collection and the indexing means vector set; For each Here
Figure BDA0000061427180000126
And
Figure BDA0000061427180000128
Expression
Figure BDA0000061427180000129
In each method p jCan pass through the method vector
Figure BDA00000614271800001210
Carry out index;
Step S3, generation k-dimension method vector step; Comprise: step S31, find all 1-dimension methods vector notions; Step S32, generation k-dimension method vector notion;
Step S4, the vectorial service grid step of generation; If
Figure BDA00000614271800001211
It is method vector cluster background In, by the k-dimension method vector notion that the attended operation of vectorial service grid is derived, then complete lattice Be called vectorial service grid;
If
Figure BDA00000614271800001214
Be a method vector cluster background, to each
Figure BDA00000614271800001215
Figure BDA00000614271800001216
The attended operation method representation of vector service grid is as follows:
Figure BDA00000614271800001217
Figure BDA00000614271800001218
In the present embodiment, described step S4 comprises:
Step S41: scan one time
Figure BDA00000614271800001219
Produce candidate 1-dimension method vector concept set CVC 1
If
Figure BDA00000614271800001220
Be a method vector cluster background,
Figure BDA00000614271800001221
It is a method vector set; Be called
Figure BDA00000614271800001223
The territory, be designated as
Figure BDA00000614271800001224
Figure BDA00000614271800001225
Be called
Figure BDA00000614271800001226
Closure, be designated as
Figure BDA00000614271800001228
In all divide vectors all to get
Figure BDA00000614271800001229
The middle minimum value that occurs; If
Figure BDA00000614271800001230
And
Figure BDA00000614271800001231
The time, then claim
Figure BDA00000614271800001232
For about the territory
Figure BDA00000614271800001233
Method vector cluster background
Figure BDA00000614271800001234
Under method of servicing vector notion VC, abbreviate method vector notion as; If k item component is arranged in the vector, then claim
Figure BDA00000614271800001235
Be method vector cluster background
Figure BDA00000614271800001236
Under k-dimension method of servicing vector notion k-VC, abbreviate the vectorial notion of k-dimension method as;
Have only when a method vector equals group of methods vector notion closure of a set, be only a method vector; Therefore,, ask for closure by the territory then, and judge whether this closure exists in 1-dimension method vector centralized concept, if just do not add by asking for the territory of each candidate 1-dimension method vector earlier; After 1-dimension method vector notion had been arranged, algorithm was finished the generation of k+1-dimension method vector notion in two steps;
Step S42: produce candidate k+1-dimension method vector concept set CVC K+1
Judge whether a candidate k+1-method vector notion is k+1-method vector notion; Return at last
Figure BDA0000061427180000131
-Wei method vector notion; Here,
Figure BDA0000061427180000132
Expression
Figure BDA0000061427180000133
In the speech phase number that contains.
In addition, described method further comprises step S5:Web service searching step; Comprise:
The similarity of step S51, calculation services method vector and query vector is determined the method greater than the threshold value of judgement Query Result given in advance, and with its reservation;
All methods vector notions that step S52, calculating remain and the similarity of query vector are if similarity is greater than threshold value, then for each method in the method vector notion territory, the similarity of computing method vector and query vector; If greater than threshold value, then this method vector is joined result set, otherwise it is joined in the non-result set; Handle all method vector notions, enter step S53;
Step S53, for the method vector of not visiting, promptly neither in result set, again not in non-result set, repeating step S52 then is up to handling all method vectors.
For reducing the calculating scale, in realizing the Web service management, can use following strategy:
Strategy 1:, avoid the few lexical item of those quantity of information to enter the feature lexical item collection based on method selected characteristic lexical item from the information source of Web service method of the heavy lexical item collection of k-right-safeguarding.
Strategy 2: if there is certain subvector in certain k-dimension method vector is not method vector notion, and this method vector just is not a candidate method vector notion so.Utilize this feature, can reduce the search volume that produces the alternative approach vector effectively.
In sum, the Web service management method of the support vector retrieval that the present invention proposes, make full use of Web service vector description and form concept analysis and set up a vectorial service grid, realized in the method for vectorial service grid employing based on the method retrieval Web service of vector based on method vector notion.The inventor has fully proved the feasibility of this technical method, and has had superior performance from theoretical and two aspects of experiment.This is that other technologies and method are not available, and this technology has been filled up the blank of correlative study and technical field, and has actual application value widely.In the medicine and hygiene fields project, obtained successful application.
Here description of the invention and application is illustrative, is not to want with scope restriction of the present invention in the above-described embodiments.Here the distortion of disclosed embodiment and change are possible, and the various parts of the replacement of embodiment and equivalence are known for those those of ordinary skill in the art.Those skilled in the art are noted that under the situation that does not break away from spirit of the present invention or essential characteristic, and the present invention can be with other form, structure, layout, ratio, and realize with other assembly, material and parts.Under the situation that does not break away from the scope of the invention and spirit, can carry out other distortion and change here to disclosed embodiment.

Claims (6)

1. the Web service management method of a support vector retrieval is characterized in that described method comprises the steps:
Step S1, the reverse file frequency of employing word frequency TF/ IDF method are carried out vectorization to the corresponding descriptor of Web service;
Step S2, production method vector cluster background step; A method vector cluster background is a tlv triple With Be two finite sets, represent method of servicing collection and indexing means vector set respectively;
Figure FDA0000061427170000014
Relation between expression method of servicing collection and the indexing means vector set; For each
Figure FDA0000061427170000015
Here
Figure FDA0000061427170000016
Figure FDA0000061427170000017
And
Figure FDA0000061427170000018
Expression
Figure FDA0000061427170000019
In each method p jCan pass through the method vector Carry out index;
Step S3, generation k-dimension method vector step; Comprise: step S31, find all 1-dimension methods vector notions; Step S32, generation k-dimension method vector notion;
Step S4, the vectorial service grid step of generation; If
Figure FDA00000614271700000111
It is method vector cluster background
Figure FDA00000614271700000112
In, by the k-dimension method vector notion that the attended operation of vectorial service grid is derived, then complete lattice
Figure FDA00000614271700000113
Be called vectorial service grid;
If
Figure FDA00000614271700000114
Be a method vector cluster background, to each
Figure FDA00000614271700000115
Figure FDA00000614271700000116
The attended operation method representation of vector service grid is as follows:
Figure FDA00000614271700000117
Figure FDA00000614271700000118
2. the Web service management method of support vector retrieval according to claim 1 is characterized in that:
Described step S4 comprises:
Step S41: scan one time Produce candidate 1-dimension method vector concept set CVC 1
If
Figure FDA00000614271700000120
Be a method vector cluster background,
Figure FDA00000614271700000121
It is a method vector set;
Figure FDA00000614271700000122
Be called
Figure FDA00000614271700000123
The territory, be designated as
Figure FDA00000614271700000124
Figure FDA00000614271700000125
Be called
Figure FDA00000614271700000126
Closure, be designated as
Figure FDA00000614271700000127
Figure FDA00000614271700000128
In all divide vectors all to get
Figure FDA00000614271700000129
The middle minimum value that occurs; If
Figure FDA00000614271700000130
And
Figure FDA00000614271700000131
The time, then claim
Figure FDA00000614271700000132
For about the territory Method vector cluster background
Figure FDA00000614271700000134
Under method of servicing vector notion VC, abbreviate method vector notion as; If k item component is arranged in the vector, then claim
Figure FDA00000614271700000135
Be method vector cluster background
Figure FDA00000614271700000136
Under k-dimension method of servicing vector notion k-VC, abbreviate the vectorial notion of k-dimension method as;
Have only when a method vector equals group of methods vector notion closure of a set, be only a method vector; Therefore,, ask for closure by the territory then, and judge whether this closure exists in 1-dimension method vector centralized concept, if just do not add by asking for the territory of each candidate 1-dimension method vector earlier; After 1-dimension method vector notion had been arranged, algorithm was finished the generation of k+1-dimension method vector notion in two steps;
Step S42: produce candidate k+1-dimension method vector concept set CVV K+1
Judge whether a candidate k+1-method vector notion is k+1-method vector notion; Return at last
Figure FDA0000061427170000021
-Wei method vector notion; Here,
Figure FDA0000061427170000022
Expression In the speech phase number that contains.
3. the Web service management method of support vector retrieval according to claim 1 is characterized in that:
Described method further comprises step S5:Web service searching step; Comprise:
The similarity of step S51, calculation services method vector and query vector is determined the method greater than the threshold value of judgement Query Result given in advance, and with its reservation;
All methods vector notions that step S52, calculating remain and the similarity of query vector are if similarity is greater than threshold value, then for each method in the method vector notion territory, the similarity of computing method vector and query vector; If greater than threshold value, then this method vector is joined result set, otherwise it is joined in the non-result set; Handle all method vector notions, enter step S53;
Step S53, for the method vector of not visiting, promptly neither in result set, again not in non-result set, repeating step S52 then is up to handling all method vectors.
4. the Web service management method of support vector retrieval according to claim 1 is characterized in that:
For reducing the calculating scale, in realizing the Web service management,, avoid the few lexical item of those quantity of information to enter the feature lexical item collection based on method selected characteristic lexical item from the information source of Web service method of the heavy lexical item collection of k-right-safeguarding.
5. the Web service management method of support vector retrieval according to claim 1 is characterized in that:
For reducing the calculating scale, in realizing the Web service management, not method vector notion if there is certain subvector in certain k-dimension method vector, this method vector just is not a candidate method vector notion so; Utilize this feature, reduce the search volume that produces the alternative approach vector effectively.
6. the Web service management method of a support vector retrieval is characterized in that described method comprises the steps:
Step S1, the reverse file frequency of employing word frequency TF/ IDF method are carried out vectorization to the corresponding descriptor of Web service;
Step S2, production method cluster background;
Step S3, generation k-dimension method vector; Comprise: step S31, find all 1-dimension methods vector notions; Step S32, generation k-dimension method vector notion;
Step S4, generate vectorial service grid.
CN 201110125682 2011-05-16 2011-05-16 Vector-retrieval-supported web service management method Pending CN102184238A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN 201110125682 CN102184238A (en) 2011-05-16 2011-05-16 Vector-retrieval-supported web service management method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN 201110125682 CN102184238A (en) 2011-05-16 2011-05-16 Vector-retrieval-supported web service management method

Publications (1)

Publication Number Publication Date
CN102184238A true CN102184238A (en) 2011-09-14

Family

ID=44570415

Family Applications (1)

Application Number Title Priority Date Filing Date
CN 201110125682 Pending CN102184238A (en) 2011-05-16 2011-05-16 Vector-retrieval-supported web service management method

Country Status (1)

Country Link
CN (1) CN102184238A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103440308A (en) * 2013-08-27 2013-12-11 北京理工大学 Digital thesis retrieval method based on formal concept analyses

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
《上海理工大学学报》 20071231 彭敦陆 向量服务格支持下的Web服务检索 476-480 1-6 第29卷, 第5期 *
《中国博士学位论文全文数据库（电子期刊）信息科技辑》 20070815 彭敦陆 基于形式概念分析的Web服务管理及检索技术 1-93 , 第2期 *
《计算机应用》 20071031 彭敦陆等 基于方法聚类的Web服务检索技术 2365-2368 1-6 第27卷, 第10期 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103440308A (en) * 2013-08-27 2013-12-11 北京理工大学 Digital thesis retrieval method based on formal concept analyses
CN103440308B (en) * 2013-08-27 2016-08-10 北京理工大学 A kind of digital thesis search method based on form concept analysis

Similar Documents

Publication Publication Date Title
Gibson Cultural economy: Achievements, divergences, future prospects
CN103186612B (en) A kind of method of classified vocabulary, system and implementation method
CN103268348B (en) A kind of user's query intention recognition methods
Luo Which industries to bail out first in economic recession? Ranking US industrial sectors by the power-of-pull
Gardashova Z-number based TOPSIS method in multi-criteria decision making
Champin et al. SIOC in action representing the dynamics of online communities
Shambour et al. A hybrid collaborative filtering recommendation algorithm for requirements elicitation
Jun et al. A novel method of IP R&D using patent analysis and expert survey
Dharavath et al. An apriori-based vertical fragmentation technique for heterogeneous distributed database transactions
CN102184238A (en) Vector-retrieval-supported web service management method
Luo et al. Discovery of textual knowledge flow based on the management of knowledge maps
CN105787072A (en) Field knowledge extracting and pushing method oriented to progress
CN101340458B (en) Grid data copy generation method based on time and space limitation
CN114186689B (en) Methods, systems, apparatus, and media for path discovery in a knowledge graph
Cao et al. Control of 2D minimally persistent formations with three co-leaders in a cycle
US8019814B2 (en) Service for standardization of resource metadata models via social networking—arriving at an agreed upon (standard) resource meta-model via social consensus
CN104598613A (en) Conceptual relation construction method and device for vertical field
Kumar et al. Function-technology-based product platform formation
Ma et al. Personalized geo-social group queries in location-based social networks
Gu et al. Risk assessment method for mass unexpected incident in city with hesitant fuzzy linguistic information
Vats et al. A Collaborative Filtering Recommender System using Apache Mahout, Ontology and Dimensionality Reduction Technique
Dang et al. A semantic-aware collaborative filtering recommendation method for emergency plans in response to meteorological hazards
Liu et al. Bootstrapping operation-level web service ontology: A bottom-up approach
Giachanou et al. USI Participation at SMERP 2017 Text Retrieval Task.
Chen et al. Service discovery based on tree structure

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C02 Deemed withdrawal of patent application after publication (patent law 2001)
WD01 Invention patent application deemed withdrawn after publication

Application publication date: 20110914