On Using Semantic Web Query Languages for Semantic Web Services Provisioning

On Using Seman ic Web Que y Languages o Seman ic Web Se ices P o isioning José Ma ía Ga cía, Ca los Ri e o, Da id Ruiz and An onio Ruiz-Co és Dep . Lenguajes y Sis emas In o má icos, Uni e si y o Se ille ETS Ingenie ía In o má ica, A . Reina Me cedes s/n, 41012 Se illa, Spain Abs ac —Al hough he e a e se e al app oaches o dis- co e Seman ic Web Se ices based on Desc ip ion Logics easoning, he use o s anda d Seman ic Web que y lan- guages o his ask is no so widely sp ead, pa ly because se ice disco e y in ol es some issues ha hese languages do no usually deal wi h, such as complex ma ching, esul s anking o in e ope abili y. In his wo k we analyze he sui - abili y o exis ing que y languages o pe o m p o isioning asks (namely disco e y, anking and selec ion) wi hin a Se- man ic Web Se ices scena io. Addi ionally, he equi emen s a Seman ic Web que y language has o ul ill in o de o be used wi hin a p o isioning scena io a e enume a ed, gi ing some insigh s in o how o ex end cu en que y languages o do so. Fu he mo e, an analysis o cu en p o isioning p oposals achie emen o hose equi emen s is p esen ed. Keywo ds: Se ice Disco e y, Seman ic Web Se ices, On ology Languages, Que y Languages, Seman ic Web. 1. In oduc ion Once a se ice has been published and made a ailable om a eposi o y, po en ial use s can e ch o desi ed se ices. This e ching, e e enced as se ice p o isioning o p ocu emen [1], in ol es di e en sequen ial asks, namely disco e y, anking and selec ion. Fi s ly, se ices ha ul ill he use equi emen s a e disco e ed. Secondly, hose se - ices a e anked wi h espec o use p e e ences. Finally, he bes anked se ice is selec ed so i can be execu ed la e on. Usually, seman ic disco e y is conside ed as a unc ional il e , because a his s age he use is looking o a se ice ha p o ides a eques ed unc ionali y. Cu en disco e y app oaches p esen ma chmaking algo i hms ha a e highly coupled wi h he se ice ep esen a ion o malism, o en based on Desc ip ions Logics [2]–[7]. These p oposals de- ines ma ching deg ees ha measu e he simila i y be ween he use equi emen s and he a ailable se ice desc ip ions. Fu he mo e, cu en p oposals ely on se ice desc ip ions and use p e e ences de ined using OWL-S[8] and WSMO [9] on ologies. Ranking and selec ion o en in ol es non- unc ional p op- e ies de ined o e se ices, e.g. cos o a ailabili y. These p ope ies a e used o ob ain a anking o disco e ed se - ices, so he bes se ice, in e ms o use p e e ences which a e based on said p ope ies, can be selec ed. Cu en p opos- als p o ide on ologies o exp ess non- unc ional p ope ies abou se ices ha a e used wi hin ad-hoc anking algo- i hms [10]–[13]. As wi h disco e y app oaches, anking and selec ion p oposals ha e a high coupling be ween p e e ence desc ip ion o malisms and algo i hms used o pe o m hese asks. The ques ion ha a ise in his p o isioning scena io is: why cu en p oposals a e no using a Seman ic Web que y language o pe o m disco e y, anking and selec ion? One eason can be ound a he le el o ma u i y o hese que y languages so ha , un il ecen ly he e has no been a s anda d que y language o he Seman ic Web. Ne e heless, he some imes complex easoning needed o ma ch se ices and use p e e ences con o ms a key ea u e ha cu en que y languages do no comple ely suppo , especially RDF-based languages. In his pape we depic he equi emen s a que y language has o ul ill in o de o be used o disco e and ank Seman ic Web Se ices (SWS), ho oughly analyzing he sui abili y o cu en que y languages o p o isioning asks, and discussing ex ensions ha make hese languages complian wi h he enume a ed equi emen s, a leas in pa . The es o he pape is s uc u ed as ollows. In Sec. 2 ex- is ing que y languages o he Seman ic Web a e desc ibed. Then, in Sec. 3 an analysis o he equi emen s o hese que y languages o suppo SWS p o isioning is p esen ed, along wi h a discussion abou cu en p oposals on he opic. Finally, in Sec. 4 we sum up ou con ibu ions, and discuss ou conclusions. 2. Que y Languages o he Seman ic Web One o he mos impo an ea u es o he Seman ic Web is ha i sepa a es he da a in o ma ion om he schema model o be applied o his in o ma ion [14]. In he Seman ic Web ield, he W3C ecommends RDF (Resou ce Desc ip- ion F amewo k) [15] as he da a model, and RDFS (RDF Schema) [16] o OWL (Web On ology Language) [17] as he schema model. OWL is di ided in h ee inc easingly exp essi e sub- languages: Li e, DL and Full. DL s ands o Desc ip ion Logics, which is a logical o malism ha p o ides he heo e ical ounda ions o Seman ic Web on ologies [18], TBox Se iceP o ile P o ile subclass o RBox hasPa ame e hasInpu hasOu pu subp ope y o ABox <<P o ile>> :amazonSe ice <<Inpu >> :keyWo dAu ho <<Ou pu >> :bookResul s hasInpu hasOu pu Fig. 1: DL on ology componen s [19]. A DL on ology has h ee concep ual componen s: asse ions abou classes (TBox), asse ions abou p ope ies and p ope y hie a chies (RBox), and p ope y asse ions be ween indi iduals and membe ship asse ions (ABox). An example o his h ee concep ual componen s is p esen ed in Fig. 1, whe e some pa s o he p o ile on ology o OWL- Sspeci ica ion [8] a e shown. Thus, in he TBox, he e a e wo classes, P o ile and Se iceP o ile and a subclass o asse ion be ween hem. In he RBox, wo subp ope y o asse ions a e shown, be ween hasInpu ,hasOu pu and hasPa ame e p ope ies. Then, in he ABox example, he e a e h ee membe ship asse ions (keyWo dAu ho ypeO Inpu ,amazonSe ice ypeO P o ile, and bookResul s ypeO Ou pu , ep esen ed by he wo d in double angle b acke s), and wo p ope y asse ions (hasInpu (amazonSe ice, key- Wo dAu ho ) and hasOu pu (amazonSe ice, bookResul s)). The e exis s wo main app oaches in Seman ic Web que y languages: RDF-based and DL-based que y languages [20], [21]. On he one hand, RDF-based que y languages allow o e ch RDF iples based on ma ching iple pa e ns wi h RDF g aphs. On he o he hand, DL-based que y languages allow o que y OWL-DL on ologies wi h TBox que ies, RBox que ies, ABox que ies o any combina ion o hem. Conce ning RDF-based que y languages, he e a e se e al app oaches wi h di e en ea u es. In [20], Baile e al. su ey 26 di e en que y languages such as SeRQL [22], RQL [23] o RDQL [24]. A p esen , he g ea majo i y o hese languages ha e nei he any implemen a ion no an upda ed implemen a ion. This is caused by he ac ha SPARQL is he only language ha is a W3C ecommen- da ion [25]. In ac , SPARQL is ully suppo ed in se e al implemen a ions1. O he su eys o p e-SPARQL languages can be ound a [22], [26]. SPARQL has ou di e en ypes o que ies: SELECT, CONSTRUCT,DESCRIBE and ASK. Each ype se es o a di e en pu pose: SELECT que ies e u n a iables and hei 1h p://www.w3.o g/2001/sw/Da aAccess/ es s/implemen a ions bindings di ec ly; CONSTRUCT que ies build an RDF g aph based on a empla e de ined in he que y; ASK que ies es whe he o no a pa e n has any solu ion; and DESCRIBE que ies e u n an RDF g aph no based on a empla e in he que y (as in CONSTRUCT que ies) bu on a p e-con igu ed g aph. The SPARQL Wo king G oup has al eady de ec ed some ex ensions o be applied o he cu en speci ica ion2. Some o hese ex ensions a e: inse /upda e/dele e que ies, access o collec ion membe s, o agg ega e unc ions (COUNT,SUM, GROUP BY, e c). Besides hese ex ensions, o he s ha e al eady been p o- posed in he Seman ic Web esea ch ield. Fo ins ance, Kie e e al. p esen iSPARQL [27] which suppo s cus- omized simila i y unc ions o que y RDF g aphs. These unc ions a e used in di e en Seman ic Web esea ch ields such as seman ic da a in eg a ion o on ology ma ching. PSPARQL [28] ex ends he o iginal ecommenda ion allow- ing egula exp essions in he p edica es o he g aph pa - e ns, while CPSPARQL [29] is an ex ension o PSPARQL ha in oduces cons ain s on pa hs. Mo eo e , SPARQL2L [30] and SPARQLeR [31] a e e y ela ed o PSPARQL and suppo he disco e y o seman ic associa ions which a e undi ec ed pa hs connec ing wo en i ies o an on ology. The SPARQL s anda d is designed o que y RDF da a only, no including RDFS ocabula y. Al hough RDF is a da a o ma ep esen ing a di ec ed labeled g aph, SPARQL only p o ides limi ed na iga ional unc ionali ies. nSPARQL [32] is ano he ex ension o SPARQL which allows o que y RDF da a acco ding o he seman ics o RDFS. This ex ension uses ecu si e g aph pa hs o achie e i s goal. Finally, Sibe ski e al. [33] p esen an ex ension which suppo s he exp ession o p e e ences and anking ha is u he discussed in Sec. 3.2. Rega ding DL-based que y languages, SPARQL-DL [21] is aligned wi h SPARQL o imp o e he in e ope abili y o applica ions on he Seman ic Web, and can be implemen ed on op o exis ing OWL-DL easone s because o i s sim- plici y. A p elimina y p o o ype o SPARQL-DL has been implemen ed on op o he OWL-DL Pelle easone [34]. OWL-QL [35] is a language and p o ocol o que y/answe dialogues in which Seman ic Web compu a ional agen s a e in ol ed. These agen s use OWL on ologies o make he dialogues possible. OWL-QL is designed o be easily adap able o o he decla a i e o mal logic ep esen a ions such as RDF o RDFS. O he DL-based que y languages a e OWL SAIQL [36] o SWQL [37]. Al hough DL-based que y languages p o ide mo e ea- soning mechanisms han RDF-based ones, he o me a e no ma u e enough and hey a e in ea ly s ages o de elopmen [20]. In his ield, SPARQL-DL is he mos p omising one 2h p://www.w3.o g/2009/01/spa ql-cha e because, as s a ed be o e, i is included in he well-known Pelle easone , which is in con inuous de elopmen . F om he p esen ed su ey we conclude ha , hough he e a e se e al Seman ic Web que y languages, SPARQL is he mos widely used, pa ly because i is a W3C ecommenda- ion. In ac , i seems o be he chosen que y language o be applied o SWS p o isioning p ocesses. Howe e , i s lack o easoning mechanisms pe se makes necessa y o ex end i in o de o allow lexible ma chmaking o se ices. The needed ex ensions o suppo his p o isioning scena io a e in oduced in he nex sec ion. 3. SWS P o isioning Using Que y Lan- guages As in oduced in Sec. 1, SWS a e usually disco e ed in e ms o a eques ed unc ionali y. This p ocess basically applies a unc ional il e o a se ice eposi o y, so a se o complian se ices a e e u ned o he use . Al hough se ice disco e y could make use o Seman ic Web que y languages o de ine he il e applied, only a ew p oposals ac ually use hem [33], [38], [39]. Addi ionally, once se ices ha e been disco e ed, hey ha e o be anked so he use can selec he bes one in e m o s a ed p e e ences. These asks could be also pe o med using a que y language, sepa a ely om disco e y [40], o in he same que y [38]. In he ollowing, we analyze wha a e he equi emen s a Seman ic Web que y language has o sa is y in o de o suppo SWS p o isioning asks, and hen discuss o wha ex en cu en p oposals ha use que y languages o pe o m hese asks ul ill he iden i ied equi emen s. 3.1 Requi emen s Analysis Conside ing he SWS p o isioning scena io, whe e a use wan s o e ch he bes se ice om a eposi o y in e ms o his o he p e e ences, he que y language chosen o suppo i has o be able o desc ibe que ies o se ice disco e y and anking. In o de o do so, we ha e iden i ied se en equi emen s ha a e enume a ed in he ollowing: (R1) Based on s anda ds. Cu en ly, SPARQL is he p o- posed s anda d que y language o he Seman ic Web, so any p o isioning app oach ha wan s o use que y languages in i s p ocess should be based on SPARQL, possibly ex ending i o using one o i s cu en ly published ex ensions. (R2) Compa ible wi h SWS amewo ks. The e a e h ee main amewo ks and on ologies o de ine SWS: OWL- S[8], WSMO [9], and SAWSDL [41]. Any que y language used in a p o isioning scena io has o be capable o handle SWS desc ip ions om any o he enume a ed amewo ks. (R3) Suppo o complex ma ching and simila i y deg ees. Se ices a e no always desc ibed using exac ly he same domain as use eques s, so he e is a need o some easoning abou equi alences and simila i y deg ees be ween concep s, especially use ul in disco - e y whe e so ma ching is needed [7]. Fo ins ance, iSPARQL suppo s his kind o lexible ma ching [27]. (R4) Reasoning mechanisms. Rela ed o he p e ious e- qui emen , easoning is a key ea u e o suppo in e - ope abili y and so ma ching be ween concep s being used in que ies. DL-based que y languages o e some acili ies o ul ill his equi emen , such as SPARQL- DL which is e alua ed by Pelle easone . Some p o- posals pe o m he easoning be o e he que y execu- ion, upda ing he knowledge base and hen execu ing he que y. (R5) E alua ion mechanisms. Especially in he anking p o- cess, e alua ion mechanisms a e needed in o de o compu e p e e ence alues used o ank disco e ed se ices. Again, DL-based que y languages o e lim- i ed suppo o his equi emen , bu some p e e ences can be compu ed easie using di e en e alua ion mechanisms wi hin a hyb id app oach [40], especially when con inuous domains a e in ol ed. (R6) Facili ies o o de esul s by compu ed alues. A e disco e ing, se ices ha e o be anked in e ms o use p e e ences, so a que y language should p o ide acili ies no only o e alua e hose p e e ences, bu o o de he esul ing alues using di e en o de ing policies. S anda d SPARQL o e s a basic o de ing suppo in i s ORDER BY clause. (R7) Decoupled om o malism. Que ies ha e o be gene ic and no coupled wi h he ac ual echniques used o e alua e hem. Thus, di e en implemen a ions o he easoning and e alua ion mechanisms can be used and changed dynamically, depending on he exp essi eness o use p e e ences. The discussed equi emen s lis make a con enien ame- wo k o compa e di e en p oposals which use que y lan- guages o pe o m p o isioning ask. This lis is used in he nex sec ion o ha pu pose. 3.2 Discussion o Cu en P oposals The e a e some p oposals ha use a Seman ic Web que y language o pe o m disco e y, anking and selec ion o se ices. They choose SPARQL as hei base language, hough some ex ensions ha e o be added o ully suppo p o isioning asks, i.e. o ul ill some o he equi emen s we ha e iden i ied be o e. Thus, Lampa e e al. [38] p o ide an on ology o ep e- sen se ice o e s and eques s ha con o ms he ounda- ions o a disco e y and selec ion p ocess pe o med using ules in SWRL [42] and SPARQL que ies. These que ies includes p edica es ha ha e o be e alua ed a un- ime, so hey include an ex ension o SPARQL ha is implemen ed using di e en p oposed algo i hms. Thus, a gene ic que y Table 1: Requi emen s sa is ied by discussed p oposals Lampa e e al. [38] Iqbal e al. [39] Sibe ski e al. [33] R1 √ √ √ R2 √ √ ∼ R3 ∼ × × R4 ∼ × × R5 √× ∼ R6 √×√ R7 × × × o a use eques is p o ided, hough his que y depends on ules ha change he ma chmaking policy, e.g. allowing ma ching deg ees as in [7]. Ano he disco e y app oach ha uses SPARQL o ac ually pe o m seman ic se ice disco e y is p oposed by Iqbal e al. in [39]. In his case, he au ho s embed seman ic in o ma ion abou se ices using SAWSDL, which is an ex ension o add seman ics o WSDL desc ip ions [41]. Thus, hey de ine p e and pos -condi ions o se ices using SPARQL CONSTRUCT que ies so ha depending on each se ice unc ionali y, hey add co esponding RDF uples ep esen ing ha unc ionali y o he knowledge base. Then, hei disco e y algo i hm use an ASK que y o check whe he a se ice ul ills a use eques o no , e u ning he esul s. Finally, conce ning anking, he e is ano he app oach p esen ed in [33], whe e Sibe ski e al. p opose an ex en- sion o SPARQL so ha p e e ences a e desc ibed di ec ly using he que y language, wi hou basing on exis ing p e e - ences and non- unc ional p ope ies on ologies, as in o he seman ic anking app oaches [12], [43]. They p o ide a PREFERRING clause ha s a es p e e ences among alues o a iables, simila o FILTER exp essions. Howe e , his app oach does no ha e he lexibili y and easoning acili ies ha p o ides a solu ion based on an ex e nal on ology. Table 1 shows how well p e iously discussed p oposals ma ch he equi emen s enume a ed in Sec. 3.1. In his able, √means a ull suppo o he equi emen ; ∼indica es ha he p oposal p o ides a pa ial o incomple e ma e ializa ion o he co esponding equi emen ; and ×is used when he equi emen is no su icien ly suppo ed. F om his compa ison, se e al conclusions can be ob- ained. Fi s ly, he mos comple e p oposal is he one p e- sen ed by Lampa e e al. [38]. I s main d awback is ha i s ma ching (R3) and easoning mechanisms (R4) depends on logic ules ha he use mus p o ide. In addi ion, hough i is able o ank se ices in e ms o complex p e e ences ha a e e alua ed a un- ime, he o malism and algo i hms used o ha e alua ion a e explici ly exp essed using ules, causing a no desi ed coupling (R7).In he case o Iqbal e al. [39], hey use s anda d SPARQL wi hou ex ensions in a SAWSDL desc ip ion, so only he i s and second equi emen s a e me . Finally, Sibe ski e al. [33] o e an in e es ing app oach o ul ill equi emen R6 by ex ending SPARQL bo h syn ac ically and seman ically, bu he es o he equi emen a e no comple ely suppo ed. In gene al, we conclude ha he main limi a ions o cu en app oaches a e, on he one hand, hei lack o mechanisms o pe o m complex ma chings and easoning asks ( equi emen s R3, R4, and o a lesse ex en R5), and on he o he hand, hei high coupling be ween desc ip ion o malisms and algo i hms used o e alua e he que ies (R7). 4. Conclusions Seman ic Web que y languages ha e no been used o SWS p o isioning un il ecen ly. Howe e , some p oposals a e eme ging in he ield, which a e mainly based on SPARQL. The e a e also se e al ex ensions o SPARQL ha can be adop ed by SWS p o isioning p oposals which ha e been discussed ho oughly in his pape . Fu he mo e, we ha e p o ided a lis o equi emen s ha que y languages and hei ex ensions ha e o mee in o de o be use ul wi hin a p o isioning scena io. This equi emen s analysis also p o ides a con enien amewo k o compa e cu en and ongoing esea ches on que y languages o disco e and ank SWS. Addi ionally, in his wo k we ha e discussed some p opos- als, concluding ha hey pa ly ul ill hose equi emen s, bu he e a e some a eas ha need u he esea ch. In pa icula , ma ching, easoning and e alua ion mechanisms ha e o be wo ked ou , bu aking ca e o he le el o coupling hese mechanisms ha e wi h espec o de ini ion o malisms. Acknowledgmen This wo k has been pa ially suppo ed by he Eu opean Commission (FEDER) and Spanish Go e nmen unde CI- CYT p ojec Web-Fac o ies (TIN2006-00472) and by he Andalusian Go e nmen unde p ojec ISABEL (TIC-2533). Re e ences [1] A. Ruiz-Co és, O. Ma ín-Díaz, A. Du án-To o, and M. To o, “Im- p o ing he au oma ic p ocu emen o web se ices using cons ain p og amming,” In . J. Coope a i e In . Sys , ol. 14, no. 4, pp. 439– 468, 2005. [2] J. González-Cas illo, D. T as ou , and C. Ba olini, “Desc ip ion logics o ma chmaking o se ices,” Hewle Packa d Labs, Tech. Rep. HPL- 2001-265, 2001. [3] L. Li and I. Ho ocks, “A so wa e amewo k o ma chmaking based on seman ic web echnology,” in In . Wo ld Wide Web Con e ence, 2003, pp. 331–339. [4] C. Lu z and U. Sa le , “A p oposal o desc ibing se ices wi h DLs,” in In . Wo kshop on Desc ip ion Logics, 2002. [5] E. Mo a, J. Domingue, L. Cab al, and M. Gaspa i, “IRS-II: A amewo k and in as uc u e o seman ic web se ices,” in In . Seman ic Web Con e ence, 2003, pp. 306–318. [6] N. S ini asan, M. Paolucci, and K. Syca a, “Seman ic web se ice disco e y in he OWL-S IDE.” in Hawaii In e na ional Con e ence on Sys ems Science, 2006. [7] K. Syca a, M. Paolucci, A. Ankoleka , and N. S ini asan, “Au oma ed disco e y, in e ac ion and composi ion o seman ic web se ices.” J. Web Sem., ol. 1, no. 1, pp. 27–46, 2003. [8] D. Ma in, M. Bu s ein, J. Hobbs, O. Lassila, D. Mcde mo , e al., “OWL-S: Seman ic ma kup o web se ices,” DAML, Tech. Rep. 1.1, 2004. [9] D. Roman, H. Lausen, and U. Kelle , “Web se ice modeling on ology (WSMO),” WSMO, Tech. Rep. D2 1.3 Final D a , 2006. [10] J. Pa hak, N. Koul, D. Ca agea, and V. G. Hona a , “A amewo k o seman ic web se ices disco e y,” in WIDM ’05: P oceedings o he 7 h annual ACM in e na ional wo kshop on Web in o ma ion and da a managemen . New Yo k, NY, USA: ACM P ess, 2005, pp. 45–50. [11] E. M. Maximilien and M. P. Singh, “A amewo k and on ology o dynamic web se ices selec ion,” In e ne Compu ing, IEEE, ol. 8, no. 5, pp. 84–93, 2004. [12] X. Wang, T. Vi a , M. Ke igan, and I. Toma, “A QoS-awa e selec ion model o seman ic web se ices.” in ICSOC 2006, se . LNCS, A. Dan and W. Lame sdo , Eds., ol. 4294. Sp inge , 2006, pp. 390–401. [13] C. Zhou, L. Chia, and B. Lee, “DAML-QoS on ology o web se ices,” in IEEE In e na ional Con e ence on Web Se ices, 2004, pp. 472–479. [14] G. An oniou and F. anHa melen, A Seman ic Web P ime , 2nd ed. Camb idge, MA, USA: MIT P ess, 2008. [15] D. Becke , “RDF/XML Syn ax Speci ica ion,” W3C, Tech. Rep., 2004. [Online]. A ailable: h p://www.w3.o g/TR/ d -syn ax- g amma / [16] D. B ickley and R. Guha, “RDF Vocabula y Desc ip ion Language 1.0: RDF Schema,” W3C, Tech. Rep., 2004. [Online]. A ailable: h p://www.w3.o g/TR/ d -schema/ [17] D. L. McGuinness and F. an Ha melen, “OWL Web On ology Language,” W3C, Tech. Rep., 2004. [Online]. A ailable: h p://www.w3.o g/TR/owl- ea u es/ [18] A. Fokoue, A. Ke shenbaum, L. Ma, E. Schonbe g, and K. S ini as, “The Summa y Abox: Cu ing On ologies Down o Size,” in In e na- ional Seman ic Web Con e ence, 2006, pp. 343–356. [19] G. D. Giacomo and M. Lenze ini, “TBox and ABox Reasoning in Exp essi e Desc ip ion Logics,” in Desc ip ion Logics, 1996, pp. 37– 48. [20] J. Bailey, F. B y, T. Fu che, and S. Scha e , “Web and Seman ic Web Que y Languages: A Su ey,” in Reasoning Web, 2005, pp. 35–133. [21] E. Si in and B. Pa sia, “SPARQL-DL: SPARQL Que y o OWL-DL,” in OWLED, 2007. [22] P. Haase, J. B oeks a, A. Ebe ha , and R. Volz, “A Compa ison o RDF Que y Languages,” in In e na ional Seman ic Web Con e ence, 2004, pp. 502–517. [23] G. Ka ouna akis, S. Alexaki, V. Ch is ophides, D. Plexousakis, and M. Scholl, “RQL: a decla a i e que y language o RDF,” in WWW, 2002, pp. 592–603. [24] A. Seabo ne, “RDQL - A Que y Language o RDF,” HP Labs B is ol, Tech. Rep., 2004. [Online]. A ailable: h p://www.w3.o g/Submission/2004/SUBM-RDQL-20040109/ [25] E. P ud’hommeaux and A. Seabo ne, “SPARQL Que y Language o RDF,” W3C, Tech. Rep., 2006. [Online]. A ailable: h p://www.w3.o g/TR/ d -spa ql-que y/ [26] R. Angles and C. Gu ié ez, “Que ying RDF Da a om a G aph Da abase Pe spec i e,” in ESWC, 2005, pp. 346–360. [27] C. Kie e , A. Be ns ein, and M. S ocke , “The Fundamen als o iSPARQL: A Vi ual T iple App oach o Simila i y-Based Seman ic Web Tasks,” in ISWC/ASWC, 2007, pp. 295–309. [28] J.-F. Bage , F. Alkha eeb, and J. Euzena , “RDF wi h egula exp essions,” INRIA, Tech. Rep., 2007. [Online]. A ailable: h p://hal.in ia. /docs/00/14/85/17/PDF/RR-6191.pd [29] F. Alkha eeb, J.-F. Bage , and J. Euzena , “Cons ained Regula Exp essions in SPARQL,” in The 2008 In e na ional Con e ence on Seman ic Web and Web Se ices (SWWS). Las Vegas, NV: CSREA P ess, Jul 2008, pp. 91–99. [30] K. Anyanwu, A. Maduko, and A. P. She h, “SPARQ2L: owa ds suppo o subg aph ex ac ion que ies in d da abases,” in WWW, 2007, pp. 797–806. [31] K. Kochu and M. Janik, “SPARQLeR: Ex ended Spa ql o Seman ic Associa ion Disco e y,” in ESWC, 2007, pp. 145–159. [32] J. Pé ez, M. A enas, and C. Gu ie ez, “nSPARQL: A Na iga ional Language o RDF,” in In e na ional Seman ic Web Con e ence, 2008, pp. 66–81. [33] W. Sibe ski, J. Z. Pan, and U. Thaden, “Que ying he Seman ic Web wi h P e e ences,” in In e na ional Seman ic Web Con e ence, 2006, pp. 612–624. [34] E. Si in, B. Pa sia, B. C. G au, A. Kalyanpu , and Y. Ka z, “Pelle : A p ac ical OWL-DL easone ,” J. Web Sem., ol. 5, no. 2, pp. 51–53, 2007. [35] R. Fikes, P. J. Hayes, and I. Ho ocks, “OWL-QL - a language o deduc i e que y answe ing on he Seman ic Web,” J. Web Sem., ol. 2, no. 1, pp. 19–29, 2004. [36] A. Kubias, S. Schenk, S. S aab, and J. Z. Pan, “OWL SAIQL - An OWL DL Que y Language o On ology Ex ac ion,” in OWLED, 2007. [37] P. Leh i and P. Fankhause , “SWQL - A Que y Language o Da a In eg a ion Based on OWL,” in OTM Wo kshops, 2005, pp. 926–935. [38] S. Lampa e , A. Ankoleka , R. S ude , and S. G imm, “P e e ence- based selec ion o highly con igu able web se ices,” in WWW ’07: P oceedings o he 16 h in e na ional con e ence on Wo ld Wide Web. New Yo k, NY, USA: ACM, 2007, pp. 1013–1022. [39] K. Iqbal, M. L. Sbodio, V. Pe is e as, and G. Giuliani, “Seman ic se ice disco e y using SAWSDL and SPARQL,” in Seman ics, Knowledge and G id, 2008. SKG ’08. Fou h In e na ional Con e ence on, 2008, pp. 205–212. [40] J. M. Ga cía, D. Ruiz, and A. Ruiz-Co és, “Seman ic disco e y and selec ion: A qos-awa e, hyb id model,” in The 2008 In e na ional Con e ence on Seman ic Web and Web Se ices (SWWS). Las Vegas, NV: CSREA P ess, Jul 2008, pp. 3–9. [41] J. Fa ell and H. Lausen, “Seman ic anno a ions o WSDL and XML Schema,” W3C Recommenda ion, Wo ld Wide Web Conso ium, Tech. Rep., Augus 2007. [Online]. A ailable: h p://www.w3.o g/TR/sawsdl/ [42] I. Ho ocks, P. F. Pa el-Schneide , H. Boley, S. Tabe , B. G oso , and M. Dean, “SWRL: A seman ic web ule language combining OWL and RuleML,” W3C Membe Submission, Tech. Rep., 2004. [43] J. M. Ga cía, I. Toma, D. Ruiz, and A. Ruiz-Co és, “A se ice anke based on logic ules e alua ion and cons ain p og amming,” in 2nd ECOWS Non-Func ional P ope ies and Se ice Le el Ag eemen s in Se ice O ien ed Compu ing Wo kshop, se . CEUR Wo kshop P oceedings, ol. 411, Dublin, I eland, No 2008.