A Web Portal for Integrating and Sharing Geographic

i1.nd37 37 gi113.indd Wei-Jen Chung and pointoutseveral future works. advantages anddisadvantages ofMAKOCI to test MAKOCI. Finally, we conclude the in ontologies. Aprototype wasimplemented communicated basedonthesamesemantics modulesandbe be dividedinto various multi-agents, theprocesses inMAKOCI can interoperability. With theassistanceof onaconceptuallevel for semantic services are definedto annotate anddiscover GI ontologies,By formal meaningsofconcepts semantically.share, anddiscover GIservices onestop to manage, publish, GI service as whichacts cyberinfrastructure), oriented called MAKOCI (multi-agentknowledge ontology-based multi-agentsplatform, interoperability. This paperproposes an data,andchallengesofsemantic observation ofearth in many duplicationcollection hasresulted (GIservices) information services geographiccommunication ofvarious 3 2 Chin-Te Jung 7 6 5 4 106, Tel: +886-2-23931122,[email protected] 106, Tel: +886-2-23931122,[email protected] 106, Tel: +886-2-23931122,[email protected] [email protected] New Taipei city, Taiwan, No. 152,Sec. 3,BeishenRoad, 222, Tel: +886-2-86625945, No. 6,Sec. 1,Roosevelt Road, 106, Tel: +886-2-23931122,[email protected] Corresponding author * Taiwan, No. 1,Sec. 4,Roosevelt Road, 106, Tel: +886-2-33665829,[email protected] No. 1,Sec. 4,Roosevelt Road, 106, Tel: +886-2-33665829,[email protected] 1 ABSTRACT. INFORMATION SERVICES AND SHARING GEOGRAPHIC A WEB PORTAL FOR INTEGRATING MAKOCI: Engineer, Taiwan GISCenter, Taipei city, Taiwan, 6F, No. 6,Sec. 1,Roosevelt Road, Engineer, Taiwan GISCenter, Taipei city, Taiwan, 6F, No. 6,Sec. 1,Roosevelt Road, Engineer, Taiwan GISCenter, Taipei city, Taiwan, 6F, No. 6,Sec. 1,Roosevelt Road, Professor, ofInformation Management, Department Tungnan University, Director of Technologies Department, Taiwan GISCenter, Taipei city, Taiwan, 6F, Professor, ofGeography, Department National Taiwan University, Taipei city, Ph.D., ofGeography, Department Na The lackofintegration and 1 , Chih-HongSun 5 , Hong-Yang Lin 2 , Min-Fang Lien 6 , Ping-Ying Tsai tional TaiwanUniversity, Taipei city, Taiwan, monitors, stress gaugesonbridges, mobile floodgauges,of alltypes: airpollution sensorsandsensorsystemsconnected Web enablement(SWE),for exploiting Web- ofopenstandards,framework calledsensor recently hasbuiltauniqueandrevolutionary (OGC) The OpenGeospatialConsortium for andcommunication. sharing services by in-situsensors)canbeencodedinto Web satellite imagesorphenomenonmeasured data(e.g., observed observations, earth In exchanging information [Kvaløyetal., 2005]. tools,software for processing dataand use of Web instead ofconventional services, years hasledto significant advancesinthe The explosive growth inrecent oftheInternet structure Ontologies, Multi-agentSystem, Cyber infra- INTRODUCTION KEYWORDS: 3 , Chih-Shyang Chang 7 Geographic Information Service, Geographic Service, Information 4 , 20.0311:00:24 11:00:24 22.03.2013

37 GEOGRAPHY ggi113.indd 38 i 1 1 3 . i n d d

38 GEOGRAPHY 3 8 Geospatial One-Stop theotherhand, acentralplatform,In suchas the terminologies. without ashared understandingon services will bedifficultto integrate GI thosetwo ATemp whereas theotheronemay use service, ofairtemperature inaGI an observation one may use communicate [Kuhn,2003].For example, across thedomainsto interoperate and it lacksacommonandshared understanding same phenomenonbydifferent domains and mayterminology the beusedto describe isbecausethatdifferentSheth, 2002].It [Fonsecafor integrating GIservices and has beenrecognized asthemainhamper Semantic interoperability ofgeospatialdata 2003] appearmore critical. [Egenhofer,these GIservices 2002;Kuhn, andsemanticallyinteroperate GI services acentralplatform forservices, search required andintegratingsharing heterogeneous GI [Peng and Tsou, while 2003].Nevertheless, well asinformation integration andexchange geospatial dataacquisitionandprocesses as have dominated the Web(WPS), to perform and Web processing (WFS), service service Web feature (WMS), Web mappingservice such asgeography language(GML), markup are more standard formats for GIservices, (GI services). With theassistanceofOGC,there ofgeographic informationas onetype services names orcoordinates, andcanbeconsidered geographical information, suchasplace these sensorsare accompaniedwith often etal., 2005]. [Jaeger databy The observed sensors andsensorsystems via Web services imaging devicesandcountlessother earth monitors, heart Webcams, satellite-borne 2 1 do not know where theyare. addition, do notknow In other usersto use the ifthey Web services over the Web. However, itmay challengesfor datainto standard observation Web services software, suchasGeoServer Data providers caneasilyuseGIS-Server to beregistered, discovered, andmanaged. http://geoserver.org/display/GEOS/Welcome http://geo.data.gov to express the same observation. It It to express thesameobservation. AirTemperature 1 , is lack for GI services , islackfor GIservices 2 , to publishtheir

to describe and theirrelationships in adomainare of concepts meanings ofboththetypes a domain” [Uschold, 1998] whereas theformal conceptualization isaway of about “thinking conceptualization” [Gruber, 1993],where a defined as “an explicitspecificationofa information science,In anontology is Ontologies out somefuture works. disadvantages oftheplatform andpoints 5concludes the advantagesand Section 4showsanimplemented prototype.Section presents theproposed ofMAKOCI. framework ontologies, RDF, andmultiagents. 3 Section introducesbriefly thebackground about This paperisorganized asfollows. 2 Section 2008], onthesemanticlevel. for RDF[Prud'Hommeaux andSeaborne, but alsobyusingSPARQL, language aquery by browsing orusingkeywords categories notonly can search appropriate GIservices Carroll, 2004]sothat consumers Web service and (RDF)[Klyne framework description metadata isencodedintheresource semantic interoperability. The registered to annotate theregistered metadatafor use definedformal meaningsinontologies aimed to beregistered into MAKOCI and 2007]. is The metadataofeachGIservice thatanalyze GISdata)[Lutz etal.,services (Web GIS data)andgeoprocessing services thatprovide services (Web accessing services into data- categories: two divide GIservices (MAKOCI). cyberinfrastructure oriented We platform iscalledmulti-agentknowledge intheplatform. GIservices The discovery providers andconsumersto register and and usesmultiagentsto assist Web service understandings for semanticinteroperability, which facilitates ontologies to definecommon multi-agentplatform,ontology-enabled Therefore, thisstudyaimsto develop an intheplatform. GI services when theyusedifferent to keywords search hampered amongusersaswell, especially abovesemantic heterogeneity described BACKGROUND 222.03.2013 11:00:24 2 . 0 3 . 2 0 1 3

1 1 : 0 0 : 2 4 ggi113.indd 39 i 1 1 3 . i n d d

3 Zhang and Tsou, 2009]. an SDI[Bernard etal., 2003;Lutz, 2005,2006; in et al., GIservices 2007], anddiscovering improving geospatialdatasearches [Purves Sheth, 2002;Kuhn,2003;Lemmens, 2006], et al., 1999;Egenhofer, 2002;Fonseca and semantic interoperability [Bishr, 1998;Harvey be thefoundation for achieving geospatial semantic Web. Geospatialontologies can by Egenhofer [2002] for ageospatial suggested factor specifications, whichisakey geospatial conceptsbasedonontological andrelationsdistinctions between geospatial domainsto identifysemantic isto develop ontologiesfor in GIservices A solutionto thesemanticinteroperability etal., 2004]. [Knublauch string) (e.g.,XML schemadatatype aninteger or linkanindividualto an properties datatype linkindividuals, whereas properties Object and [McGuinness Van 2004]. Harmelen, properties anddatatype range into object Properties canbeclassifiedonthebasisofthis to another(calledtherangeofproperty). one class(calledthedomainofproperty) individualsfrom directional linksthatconnect and Propertiesof thesametype. are binary ofindividuals that canbeseenascollections domain, whereas classesrepresent concepts representIndividuals real instancesina (i.e., etal., relationships) 2004]. [Knublauch include individuals, classes, andproperties Van 2004].Basicelements ofOWL Harmelen, recommended by and W3C [McGuinness standard languageofontologies andis The Web isthe ontology language (OWL) methods [Dietal., 2006;Lutz etal., 2007]. 2008],andanalytical etal.,Purves 2007;Stock, search [Bernard etal., algorithms 2003; processing [Peachavanish2007], and Karimi, intocan beincorporated dataintegration 2003]. The formal meaningsinanontology andMusen, people andmachines[Crubézy semantics) canbecommunicated among shared andcommonunderstandings(i.e., [Benjamins etal., Ontologies, 1998].By inferenceinteroperability andknowledge by theexplicitspecificationsfor semantic language delineated inamachine-readable 9 Carroll, 2004;Bizer predicate, RDF encodestheinformation inof andinfer implicit information. to query The be readable and accessiblebymachines in the Web byformal semantics, whichcan 2004]. The RDFaimsto represent information andCarroll, [Klyne Web (W3C) Consortium by World Widerecommended semantic Web The RDFisoneofstandard for frameworks RDF its environment, communicate withother of autonomous behavior whichcansense Agents are computer programs capable system Multi-agent more specificationsaboutthesubject. Data ontology (i.e., owl#Taiwan_ River, subject, in Fig. 1,machinescanusetheURIof information viaURIs. Forfurther instance, butalso canaccess inanRDFtriple object and the relationship the subject between Moreover, machinesnotonlycanunderstand Geometry.owl# Polyline the predicate, and www.tgic.org.tw/GIS.owl#hasGeometry owl#Taiwan_River such as to identifyeachresource intheRDFtriple, the predicate, the the uniform resource locator (URL),whichuse by URIs[Bizer etal., 2009].URIissimilarto a andisalsorepresented andobject, subject specifies therelationships the between whereasXML schemadatatype, thepredicate concepts andindividualsinontologies) or (URIs)to identifyresourcesidentifiers (e.g., are ofatriple uniformand object resource and object of the triple are ofthetriple and object geometry, asshowninFig. 1. The subject a GIservice, For example, canstate that anRDFtriple protocol andCarroll, 2004]. [Klyne the HTTP means to identifyresources over the Web by website. However, URIprovide amore generic http:// polyline and

scheme to describe anaddressscheme to describe ofa http://www.tgic.org.tw/GISData.

http://www.tgic.org.tw/GISData. geometry whichisrelated by geometry Taiwan_River, hasGeometry. object for thesubject, whichisreferred to GIS

et al GISData.owl http://www.tgic.org.tw/ triplets [Klyne and [Klyne triplets for theobject. . , 2009]. The subject hasthe Taiwan_River URIsare used ) to access polyline subject, http:// and for for 222.03.2013 11:00:24 2 .

0 3 . 2 0 1 3

1 39 GEOGRAPHY : 0 0 : 2 4 ggi113.indd 40 i 1 1 3 . i n d d

40 GEOGRAPHY 4 0 to collect information and making decision informationto andmaking collect of rationalbehaviors, and2)assistingpeople stimuliusingcomputationalmodels external measure behaviors ofpeopleresponse to be classified:1)simulatingtheperceived or ofagentscan ofagents,purpose types two andSimonis,[Moodley 2006].For the from multipleagentswithaninfrastructure referred interactions to asystem thatsupport Wooldridge, 1999].Multi-agentsystem are andJennings, 1995; react [Wooldridge agents, andpossessrationalbehavior to Fig. 1. An RDF triple states that a GI service, Taiwan_River, has the polyline geometry polyline the has Taiwan_River, service, aGI that states triple 1. RDF Fig. An Fig. 2. MASII architecture [Moodley and Kinyua, 2006] Kinyua, and [Moodley architecture MASII 2. Fig. ontologies supports, agentsareontologies challenged supports, without rationalbehaviors orrules. Without semantic heterogeneous ofinformation programs andcannotautomatically handle However, agentsare onlycomputer information for theusers. behaviors, related andautomatically collect environment to andanalyze detect users’ agents are thelatter acted oneinasoftware [Sengupta andSieber, thisstudy, 2007].In in hardware environments andsoftware 222.03.2013 11:00:25 2 . 0 3 . 2 0 1 3

1 1 : 0 0 : 2 5 ggi113.indd 41 i 1 1 3 . i n d d

4 adopts multi-agentsand ontologies to technologies [Yang data,andof geospatialknowledge, based environment for the integration an geospatial cyberinfrastructure, Web- MAKOCI aimsto develop aplatform for The framework of MAKOCI MULTI-AGENTS PLATFORM (MAKOCI) about theproposed platform. Please refer 3for to Section more detail applicationforwith apractical complement. multi-agentsplatform an ontology-enabled on theMASII architecture andproposed Simonis, 2006]. Thus, in thispaper, we based and 2006;Moodley andKinyua, in [Moodley isnotpresentedinfer sensorweb services howto registerNevertheless, andsemantic interoperability amongagents. and semanticinontologies for semantic agent whichmaintainscommonknowledge agent,4)adapter agent andtheservice to beconsumedbytheuser web services whichexpresses conventional agent service be consumedbytheuseragent,3)non- facilitator andsearched inthedirectory provider whichcanbe agent isaservice communicate withotheragents2)service user agentwhichrepresent anend-userto shown inFig. 2. The agentsincludes1) the shared semanticinontologies, as agents to communicate information under The MASII architecture containsseveral applications for multipledomains. data withontologies thatfacilitate developing aimed to integrate heterogeneous sensor 2006], andKinyua, [Moodley by Moodley applications(MASII)for Internet distributed is basedonthemulti-agentinfrastructure and Simonis, 2006]proposed aplatform, which andSimonis[Moodley finding; andMoodley perceiving, cognizing, for andacting theway for thecommunicationofagentswhile ontologies to modelcommonknowledge For 2001]hasused [Rauble, instances, Rauble and shared understanding[Hadzic exchange information basedonacommon to AN ONTOLOGY-ENABLED 1 etal., 2010].MAKOCI

et al., 2009]. 2009]. will berecorded andstored for theanalysis addition, theusers’ In services). activities standards andcallthe into service-specific agents(to translate requests and Service references offormal conceptsinontologies) searching available Ontology agents(for the the usersbyanalyzingusers’ and activities agent isfound, the User agentwillserve Facilitator.Directory Once aavailable User and search anavailable Useragentinthe oftheusersto examinetheauthorities in MAKOCI, aSafeguard willbeinitialized 2006]. While usersloginto theapplications andSimonis,be discoverable [Moodley register theircapabilitiessothatagentscan Facilities isanagent’s where registry agents Facilitator.and aDirectory The Directory agent, User agent,Ontology agent, Service (2) The agentlevel involves Safeguard agent, inONTOCAT.services consumerstoWeb usediscovered GI service which suppliesanintegrated applicationfor system), (intelligent spatialdecisionsupport into ontologies; and(c)iSDSS knowledge domain to construct domain experts editor), whichoffers anenvironment for and multi-agents;(b)ONTOEDIT (ontologies assisted byontologies discover GIservices providers andconsumersto register and forwhich provides aregistry Web service ONTOCAT (ontology catalog)application, (1) The applicationlevel includes(a) levels,two asshowninFig. 3: MAKOCI containsthefollowing framework each agentsfor exchanging information. The and canbeautonomously communicated in ontologies inresponse ofusers’ queries, users to automatically discover GIservices And multi-agentscanbeintelligent to assist while registering orsearching GIservices. sharing interoperability andknowledge Ontologies areGI services. usedfor semantic semantically search andusetheirrequired and GI services, consumerstoWeb service to publish andsemanticallyannotate their domain knowledge, providersWeb service to their contribute of domainexperts for thecollaboration geospatial knowledge (asgeospatialdata)with entail GIservices 222.03.2013 11:00:25 2 . 0 3 . 2 0 1 3

1 41 GEOGRAPHY : 0 0 : 2 5 ggi113.indd 42 i 1 1 3 . i n d d

42 GEOGRAPHY 4 2 [Moodley and Kinyua, 2006]. andKinyua, [Moodley Three OGC polylines, andpolygons) andattributes (e.g., ofgeometries with types points, are ofreal-world abstractions phenomena and definitionsofgeospatial features, which The GISontology delineates therelationships ontology GIS [2005]. of Kolas domains, whichare basedonthesuggestions are developed inthe7related geospatial located atthecenter and oftheframework for semanticinteroperability. Ontologies are meaning ofconceptsandtheirrelationships andrefervocabularies to theformal is basedontheontologies to control The communicationamongtheseagents Ontologies Developed recommendations ofrelevant GIservices. of users’ behaviors andfor theautomatic Fig. 3. The framework of MAKOCI of framework The 3. Fig. the classesunder Spatial Referencing byCoordinates” extends The Specification “OGC Abstract Topic 2: relationships theseclasses. between the to describe properties set asobject and Attribute, Geometry, SpatialReference System, 2006]. For example, inFig. 4, andKinyua, system [Moodley andattributes polyline, orpolygon) withaspatialreference (e.g., point,should containageometry Feature” stipulates thatageospatialfeature The Specification “OGC Abstract Topic 5: the GISontology. specifications canbefollowed to develop hasAttribute, andhasSpatialReference in theGISontology, whereas spatial references) are created asclasses with thesameattributes, geometries, and Feature Collection (i.e., asetoffeatures Spatial Reference hasGeometry, Feature, 222.03.2013 11:00:25 are 2 . 0 3 . 2 0

1 3

1 1 : 0 0 : 2 5 ggi113.indd 43 i 1 1 3 . i n d d

4 [Babitski classification contains34spa tial (INSPIRE) to establishthisont in the European Information Community forclassification from Infrastructure Spatial classification ofGISdata. This studyfollows the The GISData Theme ontology managesthe GIS Data Theme ontology OGC Web Service are created asindividualsbelonging to the [Goodwin andRussomanno, 2006],which [OGC, 2005], WMS [OGC,2006b], and WPS on theOGC standards,Web e.g., Service WFS standards Here, for GIservices. we focus only The ontologyWeb specified Service Web ontology Service for future work. geometry points, polylines, andpolygons, leaving raster suchas geometries consider onlyvector-type class [OGC,2006a].However, we here expanded asclassesunderthe common geospatialgeometries, whichcanbe mon Architecture” provides of categorization The “OGC SimpleFeatureCom- Access Part 1: Systems, 3 suchasfor

et al., 2009]. class. Datum and Project OGCGeometry Fig. 4. The developed ontologies data themes, ology. The inFig. 4

the under GISfunction structure-independent category, and operation classifiedunderthe Alternatively, theclassification. as anewclassto extend Under the operation categorized by For example, inFig. 5, customize theGISFunction Theme ontology. canbecreatedGIS functions asclassesto classification, otherstructure-independent reclassification) [Albrecht, 1998].Underthe (e.g., spatialsearch, thematicsearch, and from 144GIS independent ofdatastructures, andare classified operations are GISoperations, whichare classification.Universal structure GISfunction operations pr and20universalthe 6maincategories The GISFunction Theme ontology describes GIS Function Themeontology ofthe as subcategories example, data componentsandspatialsets. For customizedata theme, spatial we canfurther which are grouped in3annexes. For each Measurement Lake Overlay Measurement and oposed by Albrecht [1998] to oposed byAlbrecht [1998]to functions into 20 categories into 20categories functions Sum operation. operation,we canadd Watercourse can be added as a canbeaddedas Overlay Overlay Hydrography. Locational Analysis. is a universal GIS isauniversal GIS is a universal GIS is auniversal GIS canbeadded Measurements

Clip 222.03.2013 11:00:26 2 .

0 3 . 2 0 1 3

1 43 GEOGRAPHY : 0 0 : 2 6 ggi113.indd 44 i 1 1 3 . i n d d

44 GEOGRAPHY 4 4 process) class[Janowicz of as atype canbespecified Each geoprocessing service hasPrecondtion, orhasResult al is recommended bythe et W3C [Martin OWL (OWL-S)Web service ontologies which The Process ontology isoneofthe Model Process ontology Model semantic annotations. the following Process ontology for Model can alsobereferred to formal classesin results (theIOPR)ofgeoprocessing services the inputs, outputs, preconditions, and (e.g., th services ontology to ofgeoprocessing state thetype ws_buffer are considered asindividuals(e.g.,services Besides thecategories, geoprocessing and properties (e.g.,and properties (e.g., byusingthecontrolled classes services relationships ontheIOPRofgeoprocessing Fig. 5. The relationships between GIS Function Theme, Process Model, Units of Measure, Web Service, Service, Web Measure, of Units Model, Process Theme, Function GIS between relationships 5.The Fig. . , 2004]. It aimsto imposemeaningsand , 2004].It Atomic Process, Input, inFig. 5)intheGISFunction Theme Atomic Process e Buffer hasInput, hasOutput, hasInput, hasOutput, class).Furthermore,

et al ). (i.e., a one-step (i.e., a one-step . , 2011]. It can , 2011].It or Output and GIS ontologies GIS and ) properties properties Input to the and outputsofageoprocessing service not onlysemanticallyannotate theinputs can firstbereferred to the For example, inFig. 5,the bythe express theirtype class byusingthe class for individuals, suchasthe property. The additionalclassesofthese be stated asanoutputofthe unit individuals inthe hasInput the inputsof declared asaninstanceof polygon) andattributes, becauseithas been out_buffer the additionalclasses. For instance, the follow of thedefinitionsandrestrictions Consequently, theseindividualshave to be alsostated usingthe (e.g.,datatype for xsd:Double) Measure , and and

class for property, where in_buffe

Output must have a geometry type (e.g., type must have ageometry distance hasInput r units and classesbythepredefined input RDF:Type ws_buffer ) canbeannotated as and , andtheXMLschema out_buffer, class(e.g., hasOutput RDF:Type RDF:Type Feature Collection Feature Collection ws_buffer property, sothe service bythe service out_buffer Atomic Process distance the hasOutput hasOutput property. property. butalso in_buffer, service service Length can 222.03.2013 11:00:26 , can 2 . 0 3 . 2 0

1 3

1 1 : 0 0 : 2 6 ggi113.indd 45 i 1 1 3 . i n d d

4 element setversion 1.1 We followed theDublincore metadata CoreDublin ontology the of measurement, suchas 2001]. This ontology isusedto definetheunits andPease, [Niles terms of general-purpose an upper-level ontology providing definitions the suggested uppermerge ontology (SUMO), The UnitsofMeasure ontology isadopted from Units ofMeasure ontology ofourfuture work. listed aspart annotations ofpreconditions andresults are inthisstudy.a geoprocessing service Further we focus onlyontheinputsandoutputsof execution (result). However, for simplicity, may begenerated successful after geometry execution (precondition), whereas polygon distance successful execution. For example, the formula expressing whatwillbetrueupon before execution, whereas aresult isalogical stating thataconditionshouldbesatisfied 2010]. Aprecondition isalogical formula using specificexpressions [Janowicz the canbeannotated by geoprocessing service Moreover, thepreconditions andresults ofa FeatureCollection onthe class andhasto satisfytherestrictions 4 3 standardize 2)For thedomainknowledge. software 1)Forframework. theontologies, Protégé implement aprototype for theproposed We usedthefollowing techniques to interoperability ofGIservices. developed ontologies are usedfor semantic of MAKOCI isimplemented, andthe thefollowing aprototype section, In keywords, withformal meanings. suchastitle, descriptions, and of GIservices, Core ontology to metadata annotate generic PROTOTYPE PROTOTYPE 5 http://protege.stanford.edu/ http://dublincore.org/documents/dces/ Length hasPrecondition individualmusthave valuesbefore 4 wasusedasanontology editor to

Measurement class, asshownin Fig. 5. and 3 , to develop theDublin class(Fig. 5). hasResult meters and properties properties miles

et al. in in , instance, from the ontologies, asshowninFig. 7.For providers corresponding values to select listsforalso provides several drop-down formal meaningsinontologies. ONTOCAT are to beannotated critical with services the inputsandoutputsofgeoprocessing forNevertheless, geoprocessing services, values.selected and relationship theGI service between willbeusedto annotate theproperties The and from theindividualsof listsare drop-down system andgeometry For example, thevaluesincoordinate of individualsandclassesfrom ontologies. down liststo ensure selection thatcorrect suchasdrop- metadata viatheuserinterface, toselected annotate semanticsofthe definedinontologies) canbe properties (e.g.,knowledge classes, individuals, and shown inFig. 6. While submitting, formal into ONTOCAT,systems, orgeometry) as (e.g., title,of theservices URL,coordinate First, theproviders submitthemetadata providers to publishtheirGIservices. consumers. Three steps are designed for the by management anddiscovery Web service for semantic to thepublishedGIservices formal meaningsinontologies canbegiven intheONTOCAT.geoprocessing services, The including dataaccessingand services, providersWeb canpublishGI service Registering inMAKOCI GIservices in MAKOCI. the prototype andillustrate theprocesses to demonstrate how to search GIservices) (i.e. howto and aspects register GIservices to builda multi-agentsystem. We usedtwo developed framework (JADE) as abasemap. 3)For theagents, Java agent API JavaScript the GoogleMaps was usedasaprogramming and interface, the ONTOCAT, theProtégé JAVA OWL API 7 6 5 http://jade.tilab.com/ http://code.google.com/apis/maps/ http://protege.stanford.edu/plugins/owl/api/ hasSpatialReference Geometry in_buffer classintheGISontology. is aninputparameter of Spatial Reference and 7 wasfollowed hasGeometry 6 wasused class 222.03.2013 11:00:27 2 5 .

0 3 . 2 0 1 3

1 45 GEOGRAPHY : 0 0 : 2 7 ggi113.indd 46 i 1 1 3 . i n d d

46 GEOGRAPHY 4 6 geometry andthelatitudelongitude geometry in the ws_buffer (i.e., out_buffer) of ws_buffer geoprocessing of(i.e., ws_buffer out_buffer) Fig. 7. A screenshot of annotating the inputs (i.e., in_buffer and distance) and an output output an and distance) and in_buffer (i.e., whichhave to beimplemented WFS service in ONTOCAT in service standard withthe Fig. 6. Screenshots of register data accessing service in ONTOCAT in service accessing data register of Screenshots 6. Fig. Point

services, to classify GI services. Classes in these Classesinthese to classifyGIservices. services, Function Theme ontology for geoprocessing andGIS ontology for dataaccessingservices class from ontologies, two GISData Theme providers have atleastone toSecond, select hasSpatialReference coordinate system (i.e., as such bycorresponding properties, described these metadataoftheparameter willbe hasWebServiceType, hasGeometry, for classifying a geoprocessing service, service, ageoprocessing classifying for from GIS Function Theme ontology Fig. 8. Select appropriate classes classes appropriate Select 8. Fig. ws_buffer , intheontologies. WGS84 ). Moreover, and 222.03.2013 11:00:27 2 . 0 3 . 2 0 1 3

1 1 : 0 0 : 2 7 ggi113.indd 47 i 1 1 3 . i n d d

4 service; or service; selected forselected class intheGISData Theme ontology canbe asshown inFiselection, ontologies willbelisted intheONTOCAT for which GI services can be served. Providers canbeserved. can which GIservices to express thegeographical area GI services Third, canbegiven ageospatialextent for geoprocessing service. ontology canbealsochosenfor 7 Fig. 9. A rectangular area in Taipei city, Taiwan, is given to express the geospatial extent of ws_buffer of extent geospatial the express to given is Taiwan, city, Taipei in area 9. Arectangular Fig. and GISData are namespaces of corresponding ontologies, such as Dublin Core, Web Service, GIS, GIS, Service, Web Core, Dublin as such ontologies, corresponding of namespaces are GISData and Fig. 10. The registered metadata of Taiwan_River is encoded in RDF, where DC, WS, GIS, GIS, WS, DC, RDF, in where encoded is Taiwan_River of metadata 10.registered The Fig. Buffer Taiwan classinGISFunction Theme

River g. 8.F , adataaccessing and GIS Data Theme ontologies, respectively ontologies, Theme Data GIS and or example, ws_buffer Flood , a

triplets. can infer resources more external viathese presented sothatmachines inRDFtriplets willbesemanticallyformalized and services ws_buffer. in Fig. 10for willbeencodedinRDF,service asshown Once registered, themetadataofGI area asshowninFig. for aGIservice, service 9. delimit arectangulararea inONTOCAT asa The registered metadataofGI Taiwan_River and Fig. 11for 222.03.2013 11:00:28 2 . 0 3 . 2 0 1 3

1 47 GEOGRAPHY : 0 0 : 2 8 ggi113.indd 48 i 1 1 3 . i n d d

48 GEOGRAPHY 4 8 is similar to the one of structured query is similarto theoneofstructuredquery and Seaborne, 2008]. The syntaxofSPARQL language for RDF[Prud'Hommeauxquery the conceptuallevel byusing SPARQL, a on the consumerscansearch GIservices Function Theme ontologies). Alternatively, ontologiestwo (i.e., GISData Theme andGIS by browsing theirclassifiedclassesfrom the consumerscansearchWeb GIservices service inMAKOCI Searching GIservices Theme ontology with WGS84 coordinate system system coordinate WGS84 with ontology Theme which are under the Flood class in the GIS Data Data GIS the in class Flood the under are which and GISFunction are namespaces for corresponding ontologies, such as Dublin Core, Web Service, Service, Web Core, Dublin as such ontologies, corresponding for namespaces are GISFunction and Fig. 12. A SPARQL to query GI services (?x) Fig. 11. The registered metadata of ws_buffer is encoded in RDF, where DC, WS, GIS, Process, Process, GIS, WS, DC, RDF, in where encoded is ws_buffer of metadata 11.Fig. registered The and Polyline geometry GIS, Process, and GIS Function Theme ontologies, respectively ontologies, Theme Function GIS and Process, GIS, historical dataaboutourenvironment.historical How potential to provide real and timemonitoring have the observation intheearth GI services CONCLUSION ANDFUTURE WORK required GIservices. all registered andto search GIservices, theRDF intheMAKOCI, whichstoresquery concepts andvocabulariesto semantically the consumerscanuseformal meaningsof in theendpoint,asshownFig. 12. Therefore, and the whichare under want to search allGIservices inRDF.services For example, iftheconsumers theregisteredfor theconsumersto query GI ONTOCAT,In aSPARQL endpointisprovided SPARQL RDF[Yu, 2011]. aimsto query where language (SQL),whichuses Flood Polyline syntax to query databases. However, syntaxto query classwith geometry, aSPARQL canbegiven WGS84

coordinate system select-from- 222.03.2013 11:00:30 2 . 0 3 . 2 0 1 3

1 1 : 0 0 : 3 0 ggi113.indd 49 i 1 1 3 . i n d d

4 . Bernard, L.,Einspanier, U., S.,Hubner, Haubrock, S.,Kuhn, W., Lessing, R.,Lutz, M.,and Visser, 4. Benjamins, V.R., Fensel, D., andPerez, managementthrough on- A.G.(1998)Knowledge 3. G.,Bergweiler, Babitski, S.,Hoffmann,J., Schön,D., Stasch,C.,and A.(2009) Walkowski, 2. Albrecht, J. systematisation (1998)Universal of AnalyticalGISOperations–Atask-oriented 1. Data Theme ontology. More comprehensive ofINSPIREarecategories usedfor theGIS of classesintheontologies, suchasthe34 classification to buildhierarchical relations andonly usemain meanings for GIservices, (as showninFig. formal 4)to describe This studydesigns onlysixontologies More comprehensive ontologies shown inthefollowing issues. that needto beimproved inthefuture, as However, there are several areas ofMAKOCI integrating GIservices. MAKOCI couldbeasolutionfor and sharing Forsearch observations, GIservices. earth in ontologies to register, manage, and they canuseformal meaningsofconcepts butheterogeneous ineachGIservice, do notneedto considerthesemantic andontologies. MAKOCI, By services users theGI assistances to search andquery multi-agents are adopted for intelligently whereasinteroperability ofGIservices, managementandsemantic for knowledge Ontologies are GIservices. used and sharing agents to develop aplatform for integrating MAKOCI facilitates ontologies andmulti- research (i.e., MAKOCI) effort to thischallenge. ahead ofus. This studypresents aninitial isamajorchallenge volume ofGIservices to integrate andshare allthesetremendous REFERENCES 9 structures. Photogrammetric-Fernerkundung-GeoInformation (PFG),N.6,pp. 451–462. U. (2003)Ontologies for intelligent search andsemantic translationinspatialdatainfra- atBasel, Management, Switzerland. ofKnowledge Aspects tologies. Paper read atProceedings Conference of theSecondInernational onPractical at GeoSpatialSemantics2009,LNCS5892,Berlin. in disasterOntology-based management. Paper integration ofsensorweb services read atlantic perspectives, eds. M.CragliaandH.Onsrud, 577–591.London: Taylor &Francis. Geographic Research: GISfunctionality. Information trans- In data-structure-independent information of GI services can be interlinked. canbeinterlinked. information ofGIservices allrelevant datasetstolinks from make external expanded for providersWeb to input service addition,ONTOCATestablish links. In shouldbe for datasetsto external one-stop GI service asa OpenDataandact join into theLinked the dataexplicitlydefined. MAKOCI should relevant themeaningsof datasetsto make data into to theRDFformat external andlinked their OpenDataandconverted the Linked as andGeoNames, haveWikipedia joinedinto sets [Bizer etal., 2009].Several websites, such data to otherexternal the datainter-linkable readable dataonthe Web (i.e., RDF)to make OpenDataaimsto publishmachine- Linked Connection with Linked Open Data [Iorio inontologiesand share domainknowledge to encode directly allowing domainexperts editor would beasignificant enhancement, ontology Therefore,and Wiki-styled a Web is achallengefor many domainexperts. using Protégé to encode the knowledge by usingProtégé software. Nevertheless, and relationships ofontologies are encoded theprototype,In theconcepts, individuals, Friendly ontologies editor relationships for GIservices future completely to conceptsand describe ontologies shouldbedeveloped inthe

et al., 2006]. 222.03.2013 11:00:30 2 . 0 3 . 2 0 1 3

1 49 GEOGRAPHY : 0 0 : 3 0 ggi113.indd 50 i 1 1 3 . i n d d

50 GEOGRAPHY 5 0 18. Janowicz, K., Bröring, A.,Stasch,C.,Schade, Janowicz,K., S.,Everding, T., and Llaves, A.(2011)ARESTful 18. Bröring, A.,Stasch,C.,andEverding, Janowicz,K., T. (2010) Towards URIsfor Meaningful 17. Jaeger, E.,Altintas, I.,Zhang, J., Ludäscher, B., Pennington, D., andMichener, 16. W. (2005)A Iorio, A.D., Presutti, V., and Vitali, F. (2005) AnOntology-Based Applicationfor 15. WikiFactory: Harvey, F., Kuhn, W., Pundt, H.,Bishr, 14. Y., C.(1999)Semanticinteroperabi- andRiedemann, Hadzic, M., Wongthongtham, P.,13. Dilon, T., andChang, E.(2009)Ontology-Based Multi- Gruber, T.R. ontology specifications. (1993)Atranslationapproach Knowledge to portable 12. Goodwin,C.,andRussomanno, D.J. prototype (2006)Anontology-based sensornetwork 11. Fonseca, F., andSheth,A.(2002) The GeospatialSemantic Web. Washington,10. D.C.: Univer- Egenhofer, M.(2002) Towards theSemanticGeospatial Web. Paper 9. read atProceedings Di,L.,Zhao, P., Yang, W., 8. and Yue, P. Automatic (2006)Ontology-driven Geospatial-Proces- ofproblem Hand- solving. Crubézy,M.,andMusen,M.A.(2003)Ontologies In insupport 7. Bizer, C.,Heath, T., andBerners-Lee, T. sofar. data-the story J. (2009)Linked Int. Semantic 6. Bishr, Y. to GISinteroperbility. (1998)Overcoming thesemanticandotherbarriers Interna- 5. Manuscript accepted forManuscript publication. N., pp.Proxy ofDigital Journal Earth, for SensorData.International andDataModel Linked Germany.Berlin, Search, Discover andShare GeospatialData2010” atFuture Symposium2010,at Internet SensorData.PaperLinked read atProceedings ofthe Workshop “Towards Digital Earth: CA. Scientific andstatisticaldatabasemanagement,atSantaBarbara, vices. Paper read atSSDBM’2005 Proceedings conference ofthe17thinternational on approachscientific workflow to geospatialdataprocessing distributed using web ser- rence, ESWC2006, atBudva,Montenegro. Creating Domain-Oriented PaperWikis. read at The 3rd European Semantic Web Confe- 33, pp. 213–232. acentralissuefor geographic sharing information. lity: The annalsofregional science, N. Agent Systems. Heidelberg: Berlin Springer. Acquisition, N.5, V. 2,pp. 199–220. tion Processing atNashville, inSensorNetworks, TN. environment. Paper read atProceedings Conference ofthe5thInternational onInforma- ofGeographic Science Information Consortium sity White Paper. Systems, atMcLean, Virginia. of the Tenth SymposiumonAdvances ACM inGeographic International Information Science Earth Technology Conference, College Park, MD. basedon sing Modeling Chaining, ProceedingsWeb-service AnnualNASA oftheSixth book onontologies, eds. S.StaabandR.Studer, Springer. 321–341.Berlin: Web Inf. Syst.,N.5, V. 3,pp. 1–22. ofGeographic Science,tional Journal Information N.12, V. 4,pp. 299–314. 222.03.2013 11:00:30 2 . 0 3 . 2 0 1 3

1 1 : 0 0 : 3 0 ggi113.indd 51 i 1 1 3 . i n d d

5 32. Niles, I.,andPease, Niles, A.(2001) Towards a Standard UpperOntology. Paper read atthe2nd 32. Moodley, D., andSimonis, I.(2006) Anewarchitecture for thesensorweb: theSWAP 31. Moodley, D., J. andKinyua, (2006) Towards amulti-agent infrastructure for In- distributed 30. McGuinness, D. L., and Van F. Harmelen, (2004)OWL Web Ontology Languageoverview 29. D., Burstein, Martin, M.,Hobbs, J., Lassia,O., D., S.,Narayanan, S.,Pa- McDermott, McIlraith, 28. Lutz, M.,Lucchi, R.,Friis-Christensen, A.,andOstländer, Descrip- N.(2007)ARule-Based 27. andcompositionofgeographic Lutz, information M.(2006)Ontology-based discovery 26. Lutz, M.(2005)Overcoming Differences – inSpatialDataInfrastructures ofMeaning 25. Lemmens, International R.(2006) Semanticinteroperability geo-services, ofdistributed 24. Kvaløy, T. A.,Rongen, E., A.,andSloot,P.Tirado-Ramos, (2005)Automatic compositionand 23. Kuhn, W. (2003)Semanticreference ofGeographic systems. Journal Infor- International 22. Kolas, D., Hebeler, J., M.(2005)Geospatial Semantic andDean, Web: Architecture ofOnto- 21. H.,Rector, Knublauch, A.,Stevens, R.,and Wroe, guideto building C.(2004)Apractical 20. G.,andCarroll, Klyne, J. J. (2004)Resource (RDF):Concepts framework and description 19. 1 Ogunquit, Maine. Ogunquit, ConferenceInternational onFormal Systems Ontology (FOIS-2001),at inInformation framework, Semantic The 5thInternational Web Conference Athens. (IWSC), Africa. applications,ternet 8thAnnualConference Applications, on Bloemfontein, South WWW 2004 [cited 7June2012].Available from http://www.w3.org/TR/owl-features/. sion/OWL-S. for Markup Web Services. W3C 2004[cited. Available from http://www.w3.org/Submis- olucci, M.,Parsia, B., Payne, T., (2004)OWL-S: N.,andSycara, E.,Srinivasan, K. Sirin, Semantic Semantics, eds. F. T. Fonseca, 114–127:Springer. andS.Levashkin, M.A.Rodriguez tion Framework GeoSpatial for In theComposition ofGeographic Services. Information for Institute Geoinformatics,services, ofMünster, University Münster. Enschede, Infrastructures, theNetherlands. andInformation Spatial DataInfrastructures Achievement andChallenges, Position Paper for Workshop onCross-learning between Technology, Enschede, The Netherlands. of University (ITC), Delft Observation ScienceandEarth forInstitute Geo-Information Germany.LNCS 3470,atBerlin, Paper ofsemanticweb services. selection read atAdvances Computing-EGC 2005, inGrid mation Science, N.17, V. 5,pp. 405–409. logies. Paper read City, atGeoS2005,LNCS3799,Mexico Mexico. (ISWC2005), Galway, Ireland. andDirections, Semantic Experiences Fourthon OWL: International Web Conference tools ontologies pluginedition1.0, OWL andCO-ODE usingtheProtege-OWL Workshop Syntax,ed.Abstract B. McBride. 222.03.2013 11:00:31 2 . 0 3 . 2 0 1 3

1 51 GEOGRAPHY : 0 0 : 3 1 ggi113.indd 52 i 1 1 3 . i n d d

52 GEOGRAPHY 5 2 4 . Zhang, T., and Tsou, spatial agrid-enabled M.H.(2009)Developing Web for Inter- portal 48. Yu, L.(2011)ADeveloper’s Guideto theSemantic Web. Springer. Berlin: 47. Yang, R.,Goodchild, M.(2010)GeospatialCyberinfrastruc- M.,andGahegan, C.,Raskin, 46. Wooldridge, M.,andJennings, agents: N.R.(1995)Intelligent Know- andpractice. Theory 45. Wooldridge, Agents. M.(1999)Intelligent Cambridge, MA: MITPress. 44. level Uschold, modeling:conceptsandterminology. M.(1998)Knowledge The Know- 43. ofBehaviour UsingNaturalSemantic SemanticSimilarity (2008) Determining K. Stock, 42. Sengupta,R.,andSieber, R.(2007)GeospatialAgents, Agents Everywhere... Transactions 41. M.(2001)Ontology Rauble, andepistemology for agent-based wayfinding simulation. 40. Purves, R.S.,Clough,P., Jones, C.B., Arampatzis, A.,Bucher, B., finch,D., Fu, G.,Joho, H., Syed,39. Prud’Hommeaux, E.,andSeaborne, A.(2008)SPARQL languagefor RDF. query In 38. W3C Peng, Z.,and forTsou, Services Geographic GIS:Distributed Inofrmation M.(2003)Internet 37. Peachavanish, H. (2007)Ontological Engineering for Geospa- Interpreting R.,andKarimi, 36. OGC.(2006b)OpenGIS® Specification,version 1.3.0.In Implementation Web Server Map 35. Specificationfor OGC.(2006a)OpenGIS Implementation Geographic information –Sim- 34. OGC.(2005) Web Feature Specification: Implementation Technical Service OGC report 33. Information Science,Information N.23, V. 5,pp. 605–630. ofGeographical Journal International andgeospatial cyberinfrastructure. net GIServices 264–277. ture: Past, present andfuture. Computers, Environment Systems, andUrban N.34,pp. ledge engineering review, N.10, V. 2,pp. 115–152. ledge Engineering Review, N.13, V. 1,pp. 5–29. N. 12,pp. 733–755. to to UserObjectives Match AvailableMetalanguage Web Services. Transactions inGIS, in GIS,N.11, V. 4,pp. 483–506. ofGeographic Science, Journal Information N.15,pp.International 653–667. Geographic Science, Information N.21, V. 7,pp. 717–745. aware search engine for of Journal information retrieval International ontheinternet. A.K., Vaid, S.,and Yang, B. (2007) The design and implementationofSPIRIT: aspatially draft. working and the Internet Wireless Ne Networks. tial Queries. Transactions inGIS,N.11, V. 1,pp. 115–130. Technical OGC06-042,ed. O. report G.Consortium. ple feature access–Part 1:Common architecture Techincal 06-103r4. report 04-094. w Jersey: John w Jersey: Wiley &Sons. 222.03.2013 11:00:31 2 . 0 3 . 2 0 1 3

1 1 : 0 0 : 3 1 ggi113.indd 53 i 1 1 3 . i n d d

5 3 mitigation, andsustainabledevelopment. specialties are system, ingeographic hazards information system, decisionsupport National department, Taiwan from University research August 1994tillJuly1997. His March, 1998to June, 2000.Prior to that,Dr. ofthegeography aschairman Sunalsoserved of theCommission Research, onSustainable Development NationalScienceCouncil from Research Center, National Taiwan from University 1998to 2004andtheexecutive secretary Chih-HongSun Chin-Te Jung Min-Fang Lien Georgia, USA (1986). He served asdirectorGeorgia, oftheGlobalChange USA(1986).Heserved (1977) andhisPh.D. degree ingeography from of theUniversity undergraduate educationattheNational Taiwan University System AssociationInformation in2010.Dr. Sunreceived his Yuan. Dr. aspresident Sunwasalsoelected for Asia Geographic Council ofEconomicPlanning Executive andDevelopment, System Center, which isanon-profit organization fundedbythe of appointed aschairman Taiwan Geographic Information 2008hewas (2005–2006).In Geographic Society Information National Taiwan andPast University President of Taiwan systems. decision support ontologies,information web services, datamining, andspatial are focused ongeographic information science, geographic from National Taiwan research in2012.His University interests Programme (UNDP).Hereceived Ph.D. degree ingeography GIS inagriculture. management. Herrecent researches are wireless sensornetwork, Minfang’s specialtiesare GIS, system analysisandproject ofgeography,student ofDepartment National Taiwan University. of technical divisionin Taiwan GIS center andalsoaPhD. inIT-relatedbeen working companies. Nowsheisthedirector and management.Sincegraduated from university, shehas isaGISdeveloper ofUnited NationsDeveloped was born in wasborn Taiwan. Shemajored ingeography isaprofessor atGeography Department, 222.03.2013 11:00:31 2 . 0 3 . 2 0 1 3

1 53 GEOGRAPHY : 0 0 : 3 1 ggi113.indd 54 i 1 1 3 . i n d d

54 GEOGRAPHY 5 4 Hong-Yang Lin Chih-Shyang Chang Ping-Ying Tsai Wei-Jen Chung the Taiwan GISCenter. graduated in2000.SinceJune2010heisasystem engineer of National Taiwan ofScienceand University Technology and onit. working different, theworld some systems thatcanmake andheisstill He alsoloves somethingnew. to learn wishisto establish His focus ofhisresearch RDF). liesoninformation technology (JADE ofsemanticweb architecture. whichwere part services, The how to combineontology withamulti-agentsystem by Web Science in2004. The mainideaofhismaster’sis dissertation mining, engineering andweb mining. software Taipei City, Taiwan, R.O.C.. research His interests includedata Scienceat ofInformation Department Tungnan University, New R.O.C. in1996.Heiscurrently anAssistant Professor inthe science from National Tsing atHsinchu, HuaUniversity Taiwan, interconnection networks. are graph theory, discrete mathematics, algorithms, and Sinica from August research 2009to July2011.His specialties postdoctoral fellow ofMathematics, Academia atInstitute September 2008to that,Dr. July2009.After Tsai as a served of Engineering, HsiaInstitute Information Hwa Technology from ofComputerassistant professor Scienceand atDepartment from National Taiwan asan in2008.Heserved University System Center. Hereceived Ph.D. degree incomputer science isanengineer of Taiwan Geographic Information studied ElectronicandComputer Engineering at obtained hisMaster’s degree ofComputer received thePh.D. degree incomputer 222.03.2013 11:00:31 2 . 0 3 . 2 0 1 3

1 1 : 0 0 : 3 1