Adoptingm.yushuwu.com

HowtoCombinetheBestofWeb2.
0andaSemanticWeb:ExamplesfromRevyu.
comTomHeathandEnricoMottaKnowledgeMediaInstitute,TheOpenUniversity,UKt.
heath@open.
ac.
uk,e.
motta@open.
ac.
ukItisnotuncommontoviewWeb2.
0andtheSemanticWebasmutuallyexclusive,competingpathstotheWebofthefuture,eachadvocatedbyadistinctcommunity.
Wearguethatthetwoapproachesareinfactcomplementary,andthatbothfacechallengestheothercansolve,suchashowtointegrateWeb2.
0dataonaWebscale,andhowtoenableuserstocreatesemanticallyrichannotations.
HerewewilluseexamplesfromRevyu.
comtodemonstratehowfeaturesofWeb2.
0andtheSemanticWebcanbecombinedinasingleservicethatovercomesthesechallenges.
Revyu1isaWebsitewherepeoplecanreviewandrateanythingtheychoose.
ThesiteisbuiltonSemanticWebtechnologies,butalsousescommonfeaturesofWeb2.
0,suchaskeywordtagging.
Web2.
0providesanumbrellalabelformyriadapplicationsthatelicitandreuseuser-generatedcontent,supportsocialandcollaborativeinteractionontheWeb,andprovideengaginguserinteractionsbasedonAJAX.
TheSemanticWebvisionisoneofdatapublishedinmachine-readableformats,givenformalsemanticsthroughtheuseofsharedontologies,andinterlinkedonaWebscale.
BymakingWebdatamoreopentoprocessingbymachines,theSemanticWebfundamentallyaimstobringtangiblebenefitstousers.
Despitebothapproaches'revolutionarypotential,barriersexisttothisbeingfullyrealised.
Web2.
0applicationshaveelicitedvastamountsofuser-generatedcontent,suchaswikientries,taggedphotos,andlinksjoiningpeopleinsocialnetworks.
However,atpresent,mostsuchapplicationsrepresentwalleddatagardensfromwhichinformationcannotbeeasilysetfreeandcombinedwithothersources.
Thiscanleadtotheun-Web-likesituationwheremyfriendinOrkutisastrangeronMySpace.
Overcomingthisrequiresservicestopublishdatainformatseasilyprocessedbythirdparties.
TheSemanticWeboffersaplatformonwhichtodojustthis.
PublishingdatainRDFlowersthebarrierstoitsreusebyothers.
Twoapplicationsmaychoosetodescribetheirdatausingthesameschema,ortheymaynot.
Thisisimmaterial,providedmappingscanbedefinedbetweenthetwodatamodels.
However,usingelementsfromexistingontologiesdoessignificantlystreamlinethedataintegrationprocess.
KeytothepowerofRDFistheabilitywithinonedocumenttomixstatementsthatuseelementsfromanynumberofontologies,withoutthedocumentitselfneedingtovalidateagainstafixedschema.
Forexample,inadditiontoXHTML,RevyuexposesreviewsinRDFusingelementsfromtheReviewvocabulary2andtheFOAFontology3(fordescribingpeople).
AdoptingthesepopularontologiesmakeRevyu1http://revyu.
com2http://purl.
org/stuff/rev#3http://xmlns.
com/foaf/0.
1/datainstantlyinteroperablewiththatfromothersources.
CreatingaRevyu-specificontologythatwasthenmappedtootherswouldhavebeenanequallyvalid,albeitmorecomplexprocess.
Acknowledgingthatanysemanticsarebetterthannone,RevyualsoexposesreviewsusingthehReviewmicroformat4embeddedinXHTMLpages.
ServicesthatofferdataAPIs,suchasAmazonorFlickr,goalongwaytodismantlingthewallsofthedatagarden.
However,barrierstothereuseofthisdatastillexist.
Datastructuresmaybetrulynovel,requiringnewmappingsbetweeneveryadditionaldatasourcethatisintegrated.
Evenmorecritically,wherevanillaXMLisused,integratingdatafromdifferentsourcesmayinvolverewritingexistingschemasifnewinformationistobeincorporatedandrepublishedinthesamedocument.
RDFdoesnotsufferthesamelimitation,allowingstatementstobearbitrarilycombinedinonedocument,madeindifferentlocationsbutreferencingthesameidentifier,ormadeindifferentlocationsusingdifferentidentifiersbutstatingthatbothidentifiersrefertothesamething.
ThislastscenarioisnicelyillustratedbyRevyu'sRDFdescriptionsofreviewers.
NewURIsaremintedforeachreviewer,whomayalreadyhaveseveralothers.
WhereareviewermaintainstheirownRDFdescriptioninanotherlocation,thisisprocessedtoretrievetheiradditionalURIs.
StatementsarethenaddedtotheirRevyuRDF,statingthatallURIsidentifythesameperson.
Humanusersalsobenefit,asadditionalinformationaboutthereviewer(e.
g.
photosorhomepagelinks)canberetrievedfromexternalRDFfilesandusedtoenhancetheirprofilepage(asillustratedbelow),withoutitbeingduplicatedinRevyu.
WebServicesthatpublishvanillaXMLpresentapplicationdeveloperswiththeadditionalchallengeofparsingXMLtreestoretrievethedesireddata.
Whilstmostprogramminglanguagesmakethistasktrivial,dataprocessingremainstiedtotheunderlyingsyntacticratherthansemanticstructureofthedata.
CreatingWeb2.
0mashupsconsequentlyrequiresthewritingofcustomhandlerstointeractwitheachAPI.
Nocommonlanguageisavailableforqueryingandintegratingsuchdatasources.
Thisissuemustbeovercomeifmashupsaretobecreatedonatrulyflexible,Webscale.
TheSPARQLquerylanguagefortheSemanticWebenablesstandardisedaccesstodistributeddatasources.
4http://microformats.
org/wiki/hreviewSQL-likequeriescanbeexecutedasHTTPGETrequestsagainstremote"endpoints",returningdatathatcanbeprocessedusingstandardcode,irrespectiveoftheendpointsunderlyingimplementation.
DevelopersmustsimplyknowthestructureoftheRDFgraphbehindtheendpointinordertowritetheappropriatequery.
Revyuexposesreview,people,andtaggingdataviaitsSPARQLendpoint.
Atpresent,additionalinformationaboutbooksreviewedonRevyuisretrievedviatheAmazonWebServicesAPIanddisplayedtousersofthesite,asshowninthefigurebelow.
However,creationoftheRDFBookMashup5meansAmazon-specificcodewillshortlybereplacedbyasingleSPARQLquery.
UnderpinningsuchdataintegrationandWeb-scalemashupexercisesistheneedforwidespreadmintingofURIsidentifyingnon-Webresources,toenablelinkingofdatafromdisparatelocations.
Asforreviewers,URIsareassignedtoeveryitemreviewedonRevyu,thereviewitself,andalltagsused.
Consequently,manythingsintheofflineworldnowhaveURIsagainstwhichfurtherRDFstatementscanbemade,onRevyuorelsewhere.
ThingssuchasrestaurantsorpubsareunlikelytominttheirownURIsinthenearfuture.
ConsequentlyRevyuprovidesvaluableinfrastructureonwhichtobuildnextgenerationmashups.
RevyuURIscanallbedereferenced,respondingwithHTTP303redirects,accordingtotheW3CTAG'sfindingonthehttpRange-14issue6.
HavingdiscussedhowSemanticWebapproachesmayaddresschallengesfacingWeb2.
0,wewillnowexaminetheinverserelationship.
WearguethattheSemanticWebvisionfaceschallengesofequalsignificanceifitistoreachwidespreadadoption.
Inourviewthesearefocusedontwoissues:availabilityofdataandinteractiondesign.
Initiativessuchasdbpedia7arebootstrappingtheSemanticWebbyRDF-isingexistingdatasets.
However,instarkcontrasttotheconventionalWebandWeb2.
0,fewmechanismscurrentlyexistallowingnon-specialistuserstocontributetotheSemanticWeb.
EarlygrowthoftheWebiswidelyattributedtoindividualscreatingpersonalsitesbycopying/pastingHTMLcodeorusingvisualeditors.
SuchapproachesmaynotbeappropriatetoaSemanticWeb.
However,therearemanyWeb2.
0applicationsenablingregularuserstocontributecontentwithoutspecialistskills.
Withfewexceptions,similartoolsenablinggrassrootspublishingontheSemanticWebarenotcurrentlyavailable.
Revyuisoneexception.
ByadheringtothewellestablishedinteractionpatternofcompletingformsinaWebbrowser,RevyuallowsuserstocreatecontentthatisimmediatelyusableontheSemanticWeb.
ThisoccurswithoutanyuserknowledgeofRDF,ontologies,oreventheprinciplesoftheSemanticWeb.
Inourview,specific,focusedapplicationsthatguide5http://sites.
wiwiss.
fu-berlin.
de/suhl/bizer/bookmashup/6http://www.
w3.
org/2001/tag/issues.
html#httpRange-147http://dbpedia.
org/userinputthroughtheuseofformsrepresentthemostpromisingwaytoelicitsemanticannotationsfromregularWebusers.
Inaddition,Revyuliftstheburdenofclassifyingrevieweditemsaccordingtocomplex,existing,orrelativelyfixedtaxonomies,throughtheuseofkeywordtagging.
Thiscreatesgreaterflexibilityinwhatcanbereviewed,andweargue,lowersthebarriertocontributionofreviews.
Alessdesirableconsequenceofthisfeatureisthatmachine-readablestatementsregardingthenatureofrevieweditemscannotbemadewithanyconfidence.
Furtherinvestigationisrequiredastohowthismaybeachieved.
Assumingthatsufficientdatacanbeproduced,theSemanticWebfacesafurtherchallenge:creatinginterfacesallowingnon-specialistuserstoexploitit.
Humanshavethousandsofyearsofexperiencecreatingandusingtextualdocuments,anddecadesofexperiencewithhypertextsystems.
TheSemanticWebisnotconstrainedtothenotionofadocumentinthesamewayastheconventionalWebofHTMLpages.
Thequestionremainsofhowwedesigncompelling,coherent,andusableinteractionsbasedondatafrommultiplesources,insuchawaythatitssource,trustworthiness,andvaluecanbedetermined.
Map-basedWeb2.
0mashupsprovidesomecluesinthisdirection,bypresentingalternativevisualisationsforcomplexdatasets.
HowsuchapproachesmayscaleonaWebofinfinitelyinterlinkeddataisnotclearatpresent.
ThisremainsanopenandpressingresearchquestionfortheWebcommunity.
Conclusions:ThispaperhassoughttohighlightdistinctchallengesfacingtheWeb2.
0andSemanticWebcommunities,identifypotentialsolutionseachcommunitymayoffertheother,andillustratewithexamplesfromRevyuhowthesemayberealised.
Inconclusionwemakethefollowingrecommendationstoeachcommunity.
Firstly,thattheWeb2.
0community:givesseriousconsiderationtopublishingdatainformsthataremoreeasilyreusable,suchasRDF;investigatestheuseofSPARQLforremotedataaccessratherthancustomAPIs;andmintsURIsforofflineitemsthataredistinctfromtheURIsofdocumentsdescribingthem.
Secondly,wearguethattheSemanticWebcommunitymustgiveurgentattentiontocreatinginterfacesallowingregularWebuserstocontributetotheSemanticWeb.
ThisshouldnottaketheformofmoreusableeditorsforontologiesorRDFinstancedata(whilstthesewouldundoubtedlybeuseful),butseektoexploitfamiliarinteractionpatterns.
Revyu'sform-basedapproachisnodoubtjustoneofmanyoptions.
Intandem,significanteffortmustbegiventodevelopingcompellinginterfacesabletodisplaystructured,linkeddatafromacrosstheWeb.
Mashupshavesetthestandardforsuchinterfacesandinteractions.
ThenextgenerationmustdemonstratetheuniquebenefitsofaWebofinterlinkeddata.
Lastly,whilstacknowledgingthatwehavemadedistinctionsofthisnaturehere,wesuggestthatviewingWeb2.
0andSemanticWebapproachesasmutuallyexclusiveisdetrimentaltoallthoseinvolved,andtothedevelopmentoftheWebasawhole.