40 11 2015 11 GeomaticsandInformationScienceofWuhanUniversity Vol.40No.11 Nov.2015 DOI:10.13203/j.whugis20130662 :1671-8860(2015)11-1557-06 1 2 2 1 1 210007 2 100141 : 与已有的侧重于服务内容的方法不同 本文根据服务组合的历史记录 提出了一种依据服务间依赖关 系发现重要服务 以及其他可与之进行组合的服务的方法在对 ProgrammableWeb 真实数据的实验结果表 明 发现的重要 WebAPI 在 Mashup 中出现率达 95% 且在分析给定 WebAPI 集的可组合性方面 成功率高 达 98% : 软件网络 ; 服务发现 ; 角色 ; 可组合性 :P208 :A 21 ( (smalworld) (scalefree) )Web Web [1] Web Web (service-oriented architecturesoa) SOAP Web (simpleobjectaccessprotocol) Web REST(representationalstatetransfer) [2] WebAPI(applicationprogramminginter- WebAPI face) Web [3-11] WebAPI [6-7] [8] WebAPI Web QoS(qualityofservice) ( ) [11-20] ( ) Web [1220] [13] [814] WebAPI [4] Oh 1 [9-10] Web WSDL(Web servicesdescriptionlanguage) WebAPI :2013-11-08 : (61075089) : E-mail:xufeng 61s@163.com : E-mail:yjsnudt@163.com
1558 2015 11 (Mashup) IT Google API Foursquare API IT API 2 Mashup Fig.2 CaseofComplexNetworkandNodeDegree USPSTracking 100 MostPowerfulCeleb- USPS Tracking USPostalService Google Maps API ;100 MostPowerful Celebrities GoogleMaps YahooGeocoding YouTube3 API ;alsongsby YouTube itunesanditunes Connect [614] (close- API 1(a) 3 Mashup ; ( 1~3) WebAPI( PageR- 1~5) WebAPI ank [16] 1 Mashup 1 ( ) Mashup-API [8] 1(b) WebAPI 2 rities alsongsby programma- bleweb.com 3 Mashup ness) (betweenness) ( ) ( 3) Fig.3 3 MainFlowofthePresented Method 1 WebAPI Fig.1 ExampleofWebAPINetworks 2.1 (community) [6] ( 654 8494 ) ( 573 7838 ) 2(a) (tag) ( ) Cytoscape [15] 2.385 WebAPI 0.679 ; -0.847 2(b) Web API API (modularity) Mashup [17] Newman
40 11 : 1559 k i i ;k mj i O(n lg 2 n) m j I 0 [18] 2 4 8 5 179 :1 ( R 5 ) 3 4 Fig.4 ResultofCommunityPartition 2.2 (structural equivalence) (block model) ( ) 4 5 ( ) ( 6 ) :R 1 ;R 2 : ; (centrality) (interconnectivity) R 3 ;R 5 ; ;R 7 WebAPI m i i ( ) : c i = ki - 珔 k mi (1) 2.3 σk mi k i ; 珔 k mi k ; 2.385 σk mi k c Web API [19] m i i : :1 N M I i =1-2 k ( mj j=1 k ) i (2) [19] 4 5 R 1 R 2 ;2 ( R 2 R 3 R 7 ) ;3 (nearestneighbors) - ; 4 R 7 ( Google Maps Flickr Twiter) ;5 R 4 5 Fig.5 DistributionofNode srole ( ) (R 5~R 7 ) ;2 ( )
1560 2015 11 6 Fig.6 CommunityNetworkwithRoleTag R 5 ProgrammableWeb k%(k ) ; [8] 3 Mashup WebAPI Mashup R 7 100 Mashup k% ;4 [8] API API Mashup API : Mashup n(n 2) WebAPI API 10 50 <n s n e> 100 8 n s n e ( n s n e ) 98% API Mashup n Dijkstra O(( V 95% Web API 7 + E )lg V )(n << V ) API Mashup WebAPI API 4 5 Kendal stau [20] R 5 R 7 ProgrammableWeb ( ) 1 (Backbone) Top-k k 1 7 Google Maps Flickr del.icio.us WebAPI [8] PageRank 5 PageRank Youtube Digg WebAPI Mashup 6
40 11 : 1561 8 7 Fig.7 BackboneofServiceCommunityNetwork Fig.8 EvaluationResultofComposableServices 1 5 Tab.1 ComparisonResultsofFiveMethods Top-10 Top-20 Top-50 Top-100 0.654 0.683 0.674 0.665 0.432 0.459 0.449 0.421 0.501 0.518 0.506 0.495 PageRank 0.782 0.812 0.834 0.821 0.806 0.827 0.832 0.836 3 98%; 95% Mashup Mashup WebAPI WebAPI Mashup API [1] YuJianHanYanbo.Service-OrientedComputing: PrincipleandApplication[M].Beijing:TsinghuaU- niversitypress2006(. [M]. : 2006) [2] FieldingR T.ArchitecturalStylesandtheDesignof Network-basedSoftwareArchitectures[D].Irvine: UniversityofCalifornia2000 [3] PautassoCZimmermann OLeymannF.Restful WebAPI Web WebServicesvs. Big WebServices:Makingthe Right Architectural Decision [C].International World Wide WebConferenceBeijingChina2008 API [5] LewisT G.NetworkScience:TheoryandApplica- API ( Mashup) tions[m].new York:John WileySonsInc2009 Web [6] Pan WeifengLiBingShao Boetal.Service Classificationand Recommendation BasedonSoft- WebAPI warenetworks[j].chinesejournalof Computers 201134(12):2355-2369( API 3 Pro-. grammableweb [J]. 201134(12):2355-2369) API [7] HeKeqingMa YutaoLiuJingetal.Software
1562 2015 11 Networks[M].Beijing:SciencePress2008(. Web. [M]. : [J]. 200831(8):1354-1363) 2008) [14]ChenShizhanFengZhiyongWang Hui.Service [8] Yutao MKeqing HBing Letal.Empirical RelationsandIts Applicationsin Service-Oriented StudyontheCharacteristicsofComplex Networks Computing[J].Chinese Journalof Computers innetworkedsoftware[j].journalof Software 201033(11):2068-2083(. 201122(3):381-407(. icecompositionin Diverseand Large-scaleService Networks [J].IEEE Transactions on Services Computing20081(1):15-32 [J]. 201033(11):2068-2083) 201122(3):381-407) [15]SmootMOnoKRuscheinskiJetal.Cytoscape [9] OhSCLeeDKumaraR T.Efective WebServ- 2.8:New FeaturesforDataIntegrationand Net- [10]KilHOhSCElmaciogluEetal.GraphTheo- retictopologicalanalysisofwebservicenetworks [J].World Wide Web200912(3):321-343 [11]RanS.A Modelfor WebServicesDiscovery with QoS [J].ACM SIGecom Exchanges20034(1): 1-10 [12]Klusch MFriesBSycaraKP.AutomatedSemantic WebServiceDiscoverywithOWLS-MX [C].Interna- tionaljoint Conferenceon Autonomous Agentsand MultiagentSystemsHakodateJapan2006 [13]FengZaiwenHeKeqingLiBingetal.A Meth- odforsemantic Web DiscoveryBasedon Context Inference[J]. Chinese Journal of Computers 200831(8):1354-1363( [J]. workvisualization [J].Bioinformatics201127 (3):431-432 [16]PageLBrinSMotwaniRetal.ThePageRank CitationRanking:BringingOrdertothe Web [R]. TechnicalReportStanfordInfoLab1998 [17] Clauset ANewman M E J Moore C.Finding CommunityStructureinveryLargeNetworks [J]. PhysicalReview E200470(6):66-111 [18] WakitaKTsurumiT.FindingCommunityStruc- turein Mega-scaleSocialNetworks [C].Interna- tionalworld Wide WebConferenceAlbertaCana- da2007 phyofcomplex MetabolicNetworks [J].Nature 2005433(7028):895-900 [20] Kendal M.A New MeasureofRank Correlation [J].Biometrika193830 (1/2):81-89 ASoftwareNetwork-basedImportantServiceDiscoveryMethod XU Feng 12 YIN Junsong 2 HUANG Liwei 2 1 DepartmentofCommandAutomationUniversityofScienceandTechnologyNanjing210007China 2 InstituteofElectronicEquipmentSystem EngineeringCorporationBeijing100141China [19]GuimeràRAmaralL A N.FunctionalCartogra- Abstract:Servicediscoveryisahottopicinthefieldofservicescomputing.Ascomparedtoclassic methodsthatfocusonservicecontentsthispaperpresentsanewapproachfromtheviewpointofin- terconnectivitytofindimportantservicesandthecorrespondingcomposableservicesbasedonprevi- ousrecordsofservicecomposition.empiricalresultsshowthat(1)95% ofmashupsonthewebsite ProgrammableWebcontainimportantWebAPIsdiscoveredbyourapproachand(2)therateofsuc- cessforanalyzingthecomposabilityofagivensetofwebapisrunsupto98%.thusthisapproach wouldbehelpfulforefectivelysearchingandrecommendingservicesontheinternet. Keywords:softwarenetwork;servicediscovery;role;composability Firstauthor:XU FengPhDseniorengineerspecializesinthecomplexnetworkandwebservices.E-mail:xufeng 61s@163.com Correspondingauthor:YINJunsongPhD.E-mail:yjsnudt@163.com. Foundationsupport:TheNationalNaturalScienceFoundationofChinaNo.61075089.