Producing Linked Data for Smart Cities: the case of Catania

Undefined 1 (2014) 1ndash5 1IOS Press

The Role of Semantics in Smart Cities

Producing Linked Data for Smart Cities thecase of CataniaSergio Consoli lowast Misael Mongiovigrave a Diego Reforgiato Recupero a Silvio Peroni ac Aldo Gangemi ab

Andrea Giovanni Nuzzolese a and Valentina Presutti aa STLab Institute of Cognitive Sciences and Technologies National Research Council via Gaifami 18 95126Catania Italy E-mail sergioconsoliistccnritb LIPN Universiteacute Paris 13 - Sorbonne Citeacute - CNRSc Department of Computer Science University of Bologna ItalyE-mail misaelmongiovi diegoreforgiato andreanuzzoleseistccnrit silvioperoniuniboitaldogangemilipnuniv-paris13fr valentinapresutticnrit

Abstract

Semantic Web technologies and in particular Linked Open Data provide a means for sharing knowledge about cities as physi-cal social and technical systems so enabling the development of smart city applications This paper presents the case of Cataniawith the aim of sharing the lessons learnt which can be reused as reference practices in other cases with similar requirementsThe importance of achieving syntactic as well as semantic interoperability - as a result of transforming heterogeneous sourcesinto Linked Data - is discussed semantic interoperability must be solved at data level in order to ease the development of smartcity applications This claim is supported by showing how this issue impacts on the design of two smart city applications Asmain contributions the paper describes (i) methods procedures and tools used for transforming heterogeneous sources intoLinked Data (ii) an ontology design pattern for modelling urban public transportation routes (iii) methods procedures and toolsfor ensuring semantic interoperability during the transformation process (iv) the design of two smart city applications based onLinked Data All produced data models and prototypes are publicly accessible online

Keywords Semantics and open data for cities Ontologies for smart cities Semantics and eGovernment RDF stream processingfor smart city applications

1 Introduction

Intelligent or smart cities are characterised by thecombination and use of emergent physical infrastruc-tures information and communication technologies(ICT) and institutional settings for knowledge sharingand innovation The aim of smart cities consists of in-creasing common problem-solving capabilities for thebenefit of citizens and Public Administrations (PAs)

Corresponding author E-mail sergioconsoliistccnrit

advancing the information and knowledge capabilitiesof the community and opening a new cycle of in-novation and e-services [17] By injecting advancedinformation technologies into the social system andby increasing its innovation capabilities cities becomemore open innovative efficient and manageable Inaddition the smart city paradigm has strong implica-tions in the Public Administration management in theway of doing politics and in the relationship amongpoliticians public servants and citizens Open Govern-mentrsquos principles [27] like transparency participation

0000-000014$0000 ccopy 2014 ndash IOS Press and the authors All rights reserved

2 Semantic platform to build smart government data model and applications

and collaboration are central keys for the integration ofcitizens within the smart city paradigm

The development of a smart city involves a multi-tude of technologies and processes [16] The Internet-of-Things networks of sensors and smart devicesembedded systems the Internet of users and peopleCloud Computing are all determining a deep revolu-tion on transport environment business and govern-ment by introducing new kinds of informational andcognitive processes such as information collection andprocessing real-time alerting and forecasting collec-tive and crowdsourced intelligence and cooperativedistributed problem-solving and learning [50] How-ever the interaction among all actors and these het-erogeneous solutions still remains a challenge Trans-forming our cities into the smart cities of the futureencompasses incorporating technologies and key digi-tal advances and link them with machine-to-machinesolutions and real-time data analytics Collecting dataand transforming them into tangible insights is crucialfor modern innovative smart cities

In this context the application of Semantic Webtechnologies on smart cities data has an extremely highpotential and practical impact [9] They facilitate dataintegration from multiple heterogeneous sources en-able the development of information filtering systemsand support knowledge discovery tasks In particularin the last years the Linked Open Data (LOD) initiativereached significant adoption and is considered the ref-erence practice for sharing and publishing structureddata on the Web [811] LOD offers the possibility ofusing data across different domains for purposes likestatistics analysis maps and publications By linkingthis knowledge interrelations and correlations can bequickly understood and new conclusions arise

Since cities have large amounts of data heteroge-neous in nature and with different quality and securityrequirements research on the opening process datareengineering linking formalisation and consumptionis of primary interest in smart cities [3] The hetero-geneity problem has to be tackled at different levelsOn the one hand syntactic interoperability is neededto unify the format of knowledge sources enablingeg distributed query [32] Syntactic interoperabilitycan be achieved by conforming to universal knowl-edge representation languages and by adopting stan-dards practices The widely adopted RDF OWL andLOD allow us to achieve such a syntactic interoper-ability On the other hand semantic interoperability isalso needed Semantic interoperability can be achievedby adopting a uniform data representation and for-

malizing all concepts into a holistic data model (con-ceptual interoperability) RDF and OWL assist us inachieving the former goal However conceptual inter-operability is domain specific and cannot be achievedonly by the adoption of standars tools and practicesThe large heterogeneous data sources in smart citiesmake the problem even harder as different semanticperspectives must be addressed in order to cope withknowledge source conceptualisations To give an ex-ample addresses of different data entities for our citydata like ldquoHospitalsrdquo ldquoChurchesrdquo ldquoPost officesrdquo ldquoPo-licerdquo ldquoSchoolsrdquo etc are aligned to the same concep-tual entity ldquoAddressrdquo (characterised by properties likeldquostreetrdquo ldquoaddress numberrdquo ldquozip coderdquo ldquocity blockrdquo) In this way it is possible to intercross data and ex-ploit them more providing application developers theopportunity to easily design their city services [49]Semantic interoperability at domain level allows mak-ing sense of distributed data and enabling their auto-matic interpretation The issue of resolving semanticinteroperability among different data sources is movedfrom the application level to the data model level De-velopers are then relieved from the burden of reconcil-ing uniforming and linking data at a conceptual leveland are able to build their solutions in a more sustain-able and efficient way The published data sources aremade discoverable and become accessible via queriesandor public facilities and integrated into higher-levelservices

In this paper we present

ndash a methodology used to collect enrich and pub-lish LOD for the Municipality of Catania a cityin Southern Italy in the context of the projectPRISMA ldquoPlatfoRms Interoperable cloud forSMArt-Governmentrdquo [43] We present the col-lected city data discuss issues around them anddescribe the process to create a semantic modelfor the city data

ndash an ontology design pattern for modelling urbanpublic transportation routes This acts as a novelgovernment data model for smart cities that en-able conceptual interoperability

ndash methods procedures and tools for ensuring se-mantic interoperability during the transforma-tion process For example information about cityfacilities coming from different sources is uni-formly represented as facility types (eg gro-cery shop) in order to generalize user require-ments when needed OWL reasoners can alsohelp in defining new facility types (eg all fa-

Semantic platform to build smart government data model and applications 3

cilities that satisfy certain conditions) Our datamodel presents these heterogeneous data in a uni-form and abstracted way which is central to theprovision of ldquosmartrdquo applications for the city

ndash two use cases that demonstrate the utility of ourmodel and represent top priorities for the city ofCatania The first is a service that suggests thebest location of a facility based on some user de-fined parameters and that may used for exam-ple to aid entrepreneurs in deciding the locationof offices for startup companies to assist an urbanplanner to locating facilities and infrastructuresor to offer an updated evaluation of a propertyto purchase The second application is related tosustainable mobility and emergency vehicle rout-ing It is aimed at supporting real-time manage-ment of road traffic and public transport inform-ing citizens on the state of roads in urban areasin particular during urban emergencies and redi-recting the road traffic by providing best alterna-tives routes to find way outs the nearest hospitalsor other locations of interest

All produced data models and prototypes are publiclyaccessible online

The paper is organized as follows Section 2 pro-vides a background of the state of the art on LODdata modelling and design for smart cities Section 3introduces the techniques and tools adopted to gatherthe data to deal with the heterogeneous data sourcesof the city and to produce the semantic linked datamodel An accurate description of the resulting ontol-ogy along with the methods adopted to publish andto query the accessible data is also included Section4 proposes some application models The paper endswith some conclusions and future directions in Sec-tion 5

2 Literature review

The smart city paradigm appeared at the beginningof this century as a fundamental component of theglobal knowledge economy It represents a model fororganizing people-driven innovation ecosystems andcity-based global innovation hubs [3144] Integrationof data and applications in smart cities has been fa-cilitated by the development of standards Some ofthem have been developed to capture city messages

and events such as the Common Alerting Protocol1the National Information Exchange Model2 and theUniversal Core3 Other standards have been developedto describe the city organization eg the MunicipalReference Model4 However most city departmentsstill deal with ad hoc cumbersome code to map inputsand outputs from their legacy applications Moreoverthe development of standard protocols does not solveall issues Users and services often want to get datafrom some specific (spatial) area and a certain periodof time In a large-scale distributed environment suchas a city having highly dynamic resources and deliver-ing a large amount of data the usual steps of discov-ering indexing and efficiently querying data are com-plex tasks

Semantic Web technologies are invaluable for in-tegrating data of a smart city Recently some efforthas been spent to extend them in this direction [9]City data provided by heterogeneous sources need tobe appropriately interpreted aggregated filtered an-notated and combined with other data sources in or-der to be queried or analyzed Here typical data in-tegration issues arise data need to be integrated withmeta-data and other data from different streams or re-sources such as static databases Semantic Web knowl-edge bases and social web APIs An appropriate se-mantic model can help to provide an interoperable rep-resentation of data [49] Open Governmentrsquos princi-ples [27] like transparency participation and collabo-ration are also central keys for the integration of cit-izens within the smart city paradigm Open Data andaccess to information are essential in the process ofincreasing positive interactions between citizens andthe city administration [3] One of the aims of usingICTs in smart cities is to enhance the communicationand interaction among citizens and public administra-tion as shown by the LOD approach suggested in [5]The proposed model describes some basic commonattributes on the characteristics of data for smart ICTsystems Their method delegates details about specificstreams to linked-data models which provide on de-mand and service-specific external domain knowledge

1Oasis Common Alerting Protocol (CAP) v 12httpdocsoasis-openorgemergencycapv12CAP-v12-oshtml

2National Information Exchange Model (NIEM) v 30httpswwwniemgov

3Universal Core (UCore) Common Data Model httpisegovuniversal-core-ucore

4Municipal Reference Model httpwwwmisa-asimca


Linked Sensor Middleware [41] is an attempt to builda platform that bridges the real-world city data with theSemantic Web thanks to wrappers for real-time datagathering and publishing data annotation and visual-ization and a SPARQL endpoint for LOD querying

To facilitate future services in smart cities the Dig-ital Administration Code incorporates many interna-tional experiences on the combination of multiple pub-lic administration data sources and their publicationas LOD [28] The main thrust is coming from big ini-tiatives in the United States (datagov) [2021] andthe United Kingdom (datagovuk) [48] both pro-viding thousands of raw sets of LOD within their por-tals The Data-gov Wiki5 is a project which investi-gates open government datasets using Semantic Webtechnologies Datasets are being translated in RDFlinked to the linked data cloud and interesting appli-cations and demos on linked government data are be-ing developed In the rest of the world there are othernotable initiatives In Germany one of the first LODportal was built for the state of Baden-Wuumlrttemberg(opendataservice-bwde) It is divided inthree main parts LOD applications and tools TheLOD portal supplied in Kenya (opendatagoke)is another example showing the great benefits providedin the matter of accountability Similar initiatives inItaly have been undertaken by the city hall of Flo-rence6 the Agency for Digital Italy7 the Piedmont re-gion8 and the Chamber of Deputies9 In addition theItalian National Research Council (CNR) has launchedits open data project ldquodatacnritrdquo [425] Webelieve that by following these relevant examples it ispossible to encourage in the medium-long term sim-ilar policies and strategies to other public administra-tions and smart cities initiatives worldwide In this pa-per we extend our initial works [1514] on the produc-tion of Linked Data for the Municipality of Cataniaand present the principles practices and methods usedfor the construction of the LOD-based semantic datamodel and the extracted ontologies for the smart cityproject of the Municipality of Catania Methods pre-sented in the past mostly focus on specific aspects (egdata coming mainly from sensors) Here we tacklemore the issue of reconciling large number of city data

5httpdata-govtwrpieduwiki6httpopendatacomunefiitlinked_data

html7httpwwwdigitpagovit8httpwwwdatipiemonteitrdfhtml9httpdaticamerait

of different nature eg organizations toponymy pub-lic services etc into a uniformed and integrated se-mantic city model taking more care about data het-erogeneity and semantic interoperability at the conceptlevel which highly support application developers intothe design of their city services and applications

3 Building a Government Data Model for SmartCities

The methodologies adopted in our work for the Mu-nicipality of Catania are based on W3C standards10pattern-based ontology design (as eg described andevaluated in [26][42]) and guidelines for LOD designand data publication in view of semantic interoperabil-ity coordinated and issued by the Agency for Digi-tal Italy [12] as eg implemented for the design ofLinked Open Data for the Italian Index of Public Ad-ministration11 Our project also benefited from experi-ence gained in previous projects in particular the de-velopment of datacnrit [4] the Linked OpenData portal for the Italian National Research Council

The handled data are in Italian therefore the pro-duced LOD model is also lexicalized for the Ital-ian context However the whole generation process iscompletely language-independent Figure 1 providesa diagram that summarizes the methodology adoptedto produce the semantic linked data model From theleft to the right we describe the main data sources thetools used for data transformation and the standardsemployed for knowledge representation

In the remaining part of this section we describetechniques and tools adopted to produce the seman-tic linked data model for the city Section 31 lists theheterogeneous data sources of the city Section 32 de-picts the different methods used to process the dataand convert them into a suitable semantic RDFOWLrepresentation Section 33 describes an ontology de-sign pattern for modelling urban public transportationroutes Section 34 illustrates the procedure employed

10httpwwww3orgstandardssemanticwebW3C is the reference standard organization for Semantic Webin general and for Linked Open Data Several working groupshave been established by the W3C Their specifications for RDFSemantic Web Deployment and Government Linked Data areconsidered a reference in opening interoperable public data W3Cand the European Commission have in fact built the infrastructureand the culture of Semantic Web since 1999 with the engagementof many working groups and funded RampD projects

11httpspcdatadigitpagovitdatahtml


to enrich our data with useful knowledge from DBpe-dia Section 35 describes the final refinement of thecity ontology to respect international standards andgood practices In Section 36 we give a description ofthe tools provided to access and query the data FinallySection 37 describes a visualization tool that showsgeo-referenced semantic data objects in the triple-storein a user-friendly way by means of Google Maps

31 Analysis of the scenario and data sources

During the preliminary phase of the project we col-lected different data sources by several organizationstied with the Municipality of Catania and interested inpublishing Linked Open Data Those data have beenre-engineered following the directions given by infor-mation analysts and data experts of the Municipalityof Catania with respect to the considered reference do-mains The main data sources were identified from aGeographic Information System (GIS) [40] and a datawarehouse (used for reporting and data analysis) con-sisting of several databases that integrate geo-locatedinformation about the province of Catania Seven ter-ritorial levels ndash hydrography topography buildingsinfrastructures technological networks administrativeboundaries and land ndash form the geo-located part of theinformation flow in the Municipality of Catania [40]The GIS is designed to contain the main data of theMunicipality of Catania with the purpose of main-taining in-depth knowledge of the local area The GIScontains geo-located data and alphanumeric informa-tion related to basic cartography and ortho-photos theroad graph buildings cadastral sections data from the1991 and 2001 census of the population the last mas-ter plan the gas network resident population munic-ipalities hospitals universities schools pharmaciespost offices emergency areas public safety fire de-partments public green areas public community cen-tres prisons and institutions for minors and orphan-ages These entities are provided in the form of ashape-based file [36] for each data record ie files withextensions dbf shp shx sbn sbx xml

Beside the GIS we used other city data sources ofdifferent nature In particular

ndash data on lines and stops of the public transport bussystem provided as a REST web service in Jsonformat12

12httpwwwamtctitiamtiamtjphp

ndash maintenance of the public lighting system of thecity released as an XML file

ndash maintenance of the state of roads sidewalkssigns and markings provided as a Microsoft SQLServer database dump

ndash historical data on municipal waste collection pro-vided as a Microsoft Excel file

ndash historical data on the urban fault reporting ser-vice available as a mySQL Server database in-stance

Each of the supplied information data sources has re-quired a different methodology to be analyzed ex-tracted converted into a LOD model published andintegrated with a common ontology describing the citybusiness processes Raw data are available upon re-quest

32 Data transformation into RDF

In the following we report each data source of thecity and the different modeling tools and technologieswe have employed for each of them

321 Geographic Information System (GIS)A list of data entity types contained in the GIS is

given in Table 1 with the correspondent class nameused in our ontology We used Tabels13 a softwaretool developed by the research foundation CTIC forconverting geo-referenced data provided by the GIS(in particular files with extensions dbf and shp)into RDF Tabels relies on the GeoTools libraries14

to store data records into a RDF representation andmodel the spatial geometry as a standard KeyholeMarkup Language (KML) file [22] KML is an in-ternational standard language for geographic repre-sentation Tabels generates automatically a customSPARQL-based script able to transform each row ofthe input data into a new instance of a correspondingRDF class Each value in the column of the input ta-ble was converted into a new triple where the subjectis the mentioned instance (SQL table) the predicate isa property based on the name of the column headerand the object is a rdfsLiteral whose value is the valueof the column The transformation procedure is com-pletely customisable to suit specific requirements ieto change and annotate classes names and associatedproperties We used Tabels for generating a first ontol-ogy and a set of shapes associated to geo-referenced

13httpidifundacioncticorgtabels14httpgeotoolsorg


Figure 1 Methodology adopted to produce the semantic linked data model for the Municipality of Catania

Figure 2 Example of a KML file produced for a ldquopharmacyrdquo entity

objects in KML format We converted the geometric

coordinates given originally in the Geodetic reference

system Gauss-Boaga (or Rome 40) into the standard


Table 1Processed data stream from the GIS

Data Class name (in Italian)

Road Arches Archi StradaliDensity Contour Contorno DensitaacutePublic Safety Pubblica SicurezzaLocations of Social Services Sedi Servizi SocialiAreas of Emergency Aree di EmergenzaPharmacies FarmacieFiber Optic Network Rete Fibra OtticaSections 1991 Census Sezioni Censimento 1991Sections 2001 Census Sezioni Censimento 2001Prisons CarceriBlocks IsolatiNetwork of Gas Pipes Rete GasNursing Homes Case RiposoMunicipality MunicipalitaacuteSchools Areas Scuole AreeMunicipal Offices Uffici ComunaliPollution Control Units Centraline SmogCivic Numbers Numeri CiviciSchools ScuoleUniversities UniversitaacuteChurches ChieseHospitals OspedaliTraffic Lights SemaforiWAN Users Utenti WANJurisdictions CircoscrizioniPost Offices PosteWater Tanks Serbatoi IdriciGreen Areas Aree VerdeMunicipal Boundaries Confini ComunaliGeneral Plan Piano Regolatore GeneraleSocial Services Areas Aree Servizi SocialiFirefighters Vigili del Fuoco

Geodetic system WGS84 [24]15 An example of finalKML file is given in Figure 2 The example refers toan object (specifically school ldquoAndrea Doriardquo) that canbe represented in a map as a polygon The shape isa list of points whose coordinates are specified in tagldquoltcoordinatesgtrdquo The tag ldquoltdescriptiongtrdquo contains aset of DBpedia links representing entities associated tothis object Associated links are obtained by means ofentity linking which we discuss below in Section 34We also aligned the resulting RDF triples to existing

15WGS84 Geo Positioning RDF vocabulary httpwwww3org200301geowgs84_pos

vocabularies in particular NeoGeo16 an ontology forGeoData and the Collections Ontology17 an OWL 2DL ontology for creating sets bags and lists of re-sources and for inferring collection properties even inthe presence of incomplete information

To give an example of the procedure employed forthe GIS data consider the data table ldquoTraffic Lightsrdquo(Italian ldquoSemaforirdquo) The SQL schema of this table in-cludes the fields

ndash ObjectID - unique number incremented sequen-tially

ndash Shape - of type Geometry that represents the co-ordinates defining the geometric characteristics ofthe entity

ndash Id - Identification number this is redundant andwill be removed from the model

ndash name - of type String that represents the entityname

ndash Sde_SDE_se - integer number that specifies thekind of traffic light

After parsing the shp and dbf files Tabels gen-erates the transformation program a SPARQL-basedscript that can be used to import the data (see Fig-ure 3) As already mentioned it is possible to edit thescript to suit custom requirements Once any changesin the transformation program is completed it is pos-sible to save and run it to generate the RDF triplesFigure 4(a) shows the RDFTurtle produced by Tabelsby using the described methodology for a single ldquoTraf-fic Lightrdquo entity Figure 4(b) shows the correspond-ing final ontology of this entity obtained by conversionthrough SPARQL CONSTRUCT instructions This ex-ample further shows the ability and simplicity of thedescribed methodology to convert GIS-based data en-abling a rapid analysis retrieval and conversion ofdata into a structured RDF format and their publica-tion in the form of Linked Open Data

322 Data on lines and stops of the public transportbus system

Data available in Json format have been parsed bya customised Java script and re-engineered into a setof RDFOWL triples (class Linee Bus in Italian) Wereused data and object properties already defined inour ontology to provide integration and uniformity Wealso aligned the ontology to existing Semantic Web vo-

16httpgeovocaborgdocneogeohtml17Collections Ontology (CO) version 20 httppurl

orgco


Figure 3 A view on the transformation program used by Tabels to convert the shape files to RDF for the table ldquoTraffic Lightsrdquo (Italian ldquoSe-maforirdquo)

(a)

(b)Figure 4 Top panel (a) RDFTurtle produced by the transformation program of Tabels for a single entity of the table ldquoTraffic Lightsrdquo (ItalianldquoSemaforirdquo) Bottom panel (b) Corresponding final RDFTurtle ontology obtained through SPARQL CONSTRUCT conversion to fully matchthe designed ontology

cabularies when possible Data are geo-referenced In

particular public transport lines are given as geometric

lines while stops are geometric points Coordinates are

expressed in the standard Geodetic system WGS84


and organised by following the NeoGeo specificationFor each geo-referenced data entity the correspondingKML file is also created and made publicly accessi-ble Apart being visualizable by our tool (described be-low) the KML files may also be useful for differentpurposes For example Figure 5 shows all KML filesrelated to the public transport lines uploaded to GoogleEarth18 White lines represent bus lines in the city thatare given to Google Earth in KML format which en-ables a clear and easy-to-understand view of all routesA simple user interaction permits to select or deselectone or more routes Again we used the Collections On-tology for creating and handling collections in OWL2 such as service areas routes and timetables Ourmodelling choices for the urban public transportationsystem are presented in mode detail in Section 33

323 Maintenance of the public lighting system ofthe city

Original data have been provided in XML format(class Illuminazione Pubblica in Italian) Althoughtools for transforming XML data into RDF are avail-able (eg ReDeFer19) we found more flexible and use-ful to use a customised conversion script This choiceallowed us to provide alignments to existing SemanticWeb vocabularies and to reuse data and object proper-ties already defined The datasets contain informationrelated to management and maintenance of the publiclighting system of the city such as fault messages thestate of faults and the life-cycle of faults (including de-scription coordination technicians involved and pri-ority levels and related to the opening maintenanceresolution and closure states) An example of the finalontology is described in Figure 6 Classes are colouredin yellow instances in green and datatype propertiesin orange The most relevant class is LightingServicewhich represents a maintenance service event usuallyscheduled as a consequence of a fault in the light-ing system The service is supervised by a Supervi-sor which is a subclass of Person An instance ofLightingService has also a ServiceStatus which rep-resents the current status and can be one of the fol-lowing ldquoopenedrdquo ldquoongoingrdquo ldquocompletedrdquo ServiceS-tatus can be reused for other kinds of maintenance ser-vices eg related to the urban fault reporting serviceThe figure also shows an example of a lighting faultinstance identified with the number 2060316 super-vised by Eng Rossi and having status ldquocompletedrdquo

18httpsearthgooglecom19httprhizomiknethtmlredefer

324 Maintenance of the state of roads sidewalkssigns and markings

This dataset related to management and mainte-nance of the state of roads sidewalks signs and mark-ings of the city (class Guasti Stradali in Italian) wasprovided as a Microsoft SQL Server database dumpwhich was managed by using the D2RQ platform20D2RQ is an open-source framework for accessing re-lational databases and produce ldquoRDF dumpsrdquo accord-ing to certain specifications Initially the tool creates aD2RQ mapping file [10] by analyzing the schema ofthe existing database This mapping file called the de-fault mapping maps each table to a new RDFS class(named after the tablersquos name) and each column toa property (named after the columnrsquos name) We cus-tomize the mapping to align the resulting ontology toexisting Semantic Web vocabularies and reuse data andobject properties already defined For example in theoriginal SQL Server database there was a data columncalled ldquodbo2012Municipalityrdquo referring to the mu-nicipality where the maintenance service is requiredAs we already defined municipalities for the ontologywhen we dealt with the GIS of the city we aligned theobject with the Municipality class already defined forthe ontology This was possibile thanks to the follow-ing snippet code in the D2RQ mapping program

mapdbo_2012_Municipality a d2rq21PropertyBridged2rqbelongsToClassMap mapdbo_2012d2rqproperty prisma-ont22Municipalityd2rqpropertyDefinitionLabel ldquoidentificatier municipalityrdquod2rqcolumn ldquodbo2012Municipalityrdquo

In short a mapdbo_2012_Municipality of kindd2rqPropertyBridge defining the D2RQ mappingrule is created this belongs (d2rqbelongsToClassMap)to the database mapdbo_2012 which is our SQLServer database instance and maps the database col-umn ldquodbo2012Municipalityrdquo to the entity class prisma-ontMunicipality of our ontology

A similar example is the class ServiceStatus de-fined for the maintenance of the public lighting sys-tem of the city which is aligned to the database col-umn ldquoStatusrdquo indicating the lifecycle status (ldquoopenedrdquoldquoongoingrdquo ldquocompletedrdquo) of a maintenance service Wereuse this ontology by customizing the D2RQ mappingprogram in a similar way as we explained for munici-

20D2RQ - Accessing Relational Databases as Virtual RDFGraphs Version 081 httpd2rqorgd2r-server

21httpwwwwiwissfu-berlindesuhlbizerD2RQ01

22httpwwwontologydesignpatternsorgontprisma


Figure 5 A screenshot of the KML files related to the public transport lines uploaded to Google Earth

Figure 6 Example of the ontology that models the public lighting system maintenance


palities These examples show how semantic interoper-ability at the concept level among heterogeneous datais enabled within our uniformed city ontology Afterthe D2RQ mapping file is created and customized theplatform allows us to dump the contents of the wholerelational database into a single RDF file

325 Historical data on municipal waste collectionData related to city services for the collection of mu-

nicipal waste and disposal of large-size garbage (classRifiuti Urbani in Italian) have been provided as a largeMicrosoft Excel file We first converted the Excel filein a Microsoft SQL Server database instance by meansof a feature of Microsoft Excel We then managed itby using the D2RQ platform and following a proce-dure similar to the one previously described Again wealigned the results to existing Semantic Web vocabu-laries and integrated them with our ontology to providesemantic data interoperability

326 Historical data on the urban fault reportingservice

Historical data related to signalling reporting andmanaging urban faults (class Segnalatore Urbano inItalian) were provided as a mySQL Server databaseinstance We again used D2RQ to map the relationaldatabase to customise the mapping appropriately andto produce an RDFOWL dump of the database Thedataset contains information related to fault reportsactions required status workflows localisation ad-dresses and WGS84 coordinates arranged by usingNeoGeo and the Collections Ontology

An example of the final ontology is described in Fig-ure 7 Again classes are colored in yellow instancesin green and datatype properties in orange The keyclass is FaultReport which represents a report of acitizen about a fault in the road system A fault is as-sociated with an Address and with geografic Coordi-nates ie a set of points (class Point) A fault reporthas also an associated Image ie a picture that showsthe fault and a User which is the person that reportedthe fault We reused the class ServiceStatus from thelighting system maintenance for representing the sta-tus of the maintenance service associated with the re-port The figure also gives an example of a report in-stance In the example the report is about a fault oc-curred at the address ldquoVle Africa 31rdquo and its status isldquoongoingrdquo

33 Ontology design patterns for urban publictransportation system

In this section we describe the two main ontologydesign patterns the we have used to design the ontol-ogy for urban public transportation system in CataniaWe report on these design patterns because they areparticularly relevant and helpful examples to illustrateour ontology design choices and because they are eas-ily reused for modeling public transportation systemswithin other smart cities projects

The first of these patterns is the ordered spatial com-position pattern shown as Graffoo diagram23 in Fig-ure 8 This pattern allows us to describe any spatialthing (eg the geometric object defining a transporta-tion line) as composed by a (possibly ordered) se-quence of other spatial things (such as points linesor polygons) The pattern has been developed by tak-ing inspiration from the Collections Ontology [13]that defines generic collections such as sets bags andlists and provides some properties to correctly indexthe various elements of the collection and actuallyreuses the sequence pattern24 for arranging the var-ious spatial elements in order We have developedthree different geo-referenced objects of our ontologyby using this pattern ie coordinates service zonesand routes as shown in Figure 10 In particular inour case a transportation line or a stop contains aprismaCoordinates entity which has a relationgeosfContains of a sequence of points (definedaccording to the NeoGeo and GeoSparql25 ontologies)

Points are arranged according to the sequence pat-tern that allows us to link to the next point in the se-quence through the sequencedirectlyPrecedesrelation In addition each point has associated an in-dex (xsdint) by means of the BBC ProgrammesOntology poposition relation identifying its po-sition in the sequence Start and end points withinthe sequence are also indicated by respectively theonthasStart and onthasEnd relations Coor-dinates of a single point are expressed as according tothe standard Geodetic system WGS84 [24]26

23Graphical Framework for OWL Ontologies GraffooV 10 httpwwwessepuntatoitgraffoospecificationcurrenthtml

24httpwwwontologydesignpatternsorgcpowlsequenceowl

25httpwwwopengisnetontgeosparql26WGS84 Geo Positioning RDF vocabulary httpwww

w3org200301geowgs84_pos


Figure 7 Example of the ontology that models the maintenance of roads sidewalks signs and markings

The following code shows an example of coordi-nates for a single point

prismaresourcecoordinatesbusline-101a prisma-ontCoordinates geo27sfContainsprismaresourcecoordinatesbusline-101-point-1 prismaresourcecoordinatesbusline-101-point-2 prismaresourcecoordinatesbusline-101-point-3 prisma-onthasStart

prismaresourcecoordinatesbusline-101-point-1 prisma-onthasEnd

prismaresourcecoordinatesbusline-101-point-3

prismaresourcecoordinatesbusline-101-point-1 a wgsPoint po28position 1ˆˆxsdint sequence29directlyPrecedes

prismaresourcecoordinatesbusline-101-point-2 wgs30lat 375321 wgslong 151087

prismaresourcecoordinatesbusline-101-point-2 a wgsPoint

The other pattern we have developed is the timetablepattern shown in Figure 9 Taking again inspiration

27geo httpwwwopengisnetontgeosparql28po httppurlorgontologypo29sequence httpwwwontologydesignpatterns

orgcpowlsequenceowl30wgs httpwwww3org200301geowgs84_

pos

from the Collections Ontology [13] this pattern al-lows us to describe timetables as a sequence of orderedtimes (properly indexed) to which we can associate aparticular description characterising them through thepattern description31 As shown in Figure 11 we haveused this pattern in order to model mainly the timeta-bles associated to bus routes by means of the W3CTime Ontology32 and of the sequence pattern

prismaresourcestop101-1-via-mon-d-orlandoprisma-ontweekdayTime

prismaresourceweekdaytime101-weekdaytime

prismaresourceweekdaytime101-weekdaytimea prisma-ontTime time33insideprismaresourceperiod101-weekdaytime-06-40 prismaresourceperiod101-weekdaytime-07-45 timehasBeginning

prismaresourceperiod101-weekdaytime-06-40 timehasEnd

prismaresourceperiod101-weekdaytime-06-40 a timePeriod poposition 1ˆˆxsdint timeinDateTime prismaresourcehour06-40

31Description pattern httpwwwontologydesignpatternsorgcpowldescriptionowl

32W3C Time Ontology specification httpwwww3orgTRowl-time

33time httpwwww3org2006time


sequencedirectlyPrecedesprismaresourceperiod101-weekdaytime-07-45

prismaresourcehour06-40a timeDateTimeDescription timeunitType timeunitMinute timehour 6ˆˆxsdnonNegativeInteger timeminute 40ˆˆxsdnonNegativeInteger

34 Knowledge Discovery through Entity Linking

Once the data were represented in RDF and thetriples for each data object were produced we per-formed a knowledge discovery step in order to en-rich the resulting knowledge base by linking to knowl-edge from DBpedia In particular all addresses andnames of extracted data objects have been sent to anentity linking tool TAGME34 TAGME is a tool per-forming named entity resolution given a short sen-tence it recognises named entities in the text andlink the identified text fragment to its correspondingWikipedia page The name of the extracted entitieswere compared by string similarity with the origi-nal object name (or address) New DBPedia RDF re-lations owlsameAs and dulassociatedWithhave been inserted into the data based on such stringsimilarity Specifically we inserted the owlsameAsrelation when TAGME produced an object with thesame name of the entire data object we inserted adulassociatedWith relation when TAGME ex-tracted entities whose names are different than the dataobjects it got as input

As an example let us consider Chiesa CattolicaSS Pietro e Paolo35 a very famous church in Cata-nia which is represented in our datasets Figure 12shows a screenshot of the properties for Chiesa Cat-tolica ss Pietro e Paolo where it is possible to no-tice the associatedWith relations whose value isrepresented by the entities discovered using TAGMENone of the entities extracted using TAGME matchthe overall name Chiesa Cattolica ss Pietro e Paoloand therefore each of them is included with the dulassociatedWith relation The user might want toplay with TAGME using the text Chiesa Cattolica

34httptagmediunipiit35httpwwwontologydesignpatternsorg

dataprismachiesadiscretionary-chiesa-cattolica-ss-pietro-e-paolo

ss Pietro e Paolo by including this link36 to hisherbrowser and obtaining the results from TAGME itself

One more example is related to Piazza Stesicoro37one of the main square of Catania Figure 13 shows thetriples describing the entity Piazza Stesicoro where theuser may notice the owlsameAs relation we haveincluded as Wikipedia contains a page related to thatparticular square For such an example TAGME re-turns one entity whose text matches Piazza Stesicoroand therefore we include that using the owlsameAsrelation Using this link38 the user can see live the re-sults of TAGME for the text Piazza Stesicoro

35 Ontology alignment

During the process of conversion as specified pre-viously ontology parts have been aligned among themand with standard existing ontologies to achieve con-ceptual interoperability Moreover several refinementsteps have been performed to conform to internationalstandards In this subsection we give some more detailsabout the alignment and refinement processes

The whole process followed the good practices offormal representation and naming in use in the domainof the Semantic Web and Linked Open Data [12]In particular the guidelines of the W3C Organiza-tion Ontology39 for generating publishing and con-suming LOD for organizational structures have beenensued In a first phase each entity type describedby the supplied data has been represented by a classand each entity field has been converted into a dataproperty Then both entities and properties referringto the same concepts have been unified This phaseis important since often the same concept from dif-ferent data sources is described by different enti-ties and properties For example the fields ldquoNamerdquoof the tables ldquoNursing Homesrdquo (ldquoCase Riposordquo) andldquoPharmaciesrdquo (ldquoFarmacierdquo) have been translated withtwo different data properties respectively ldquoName-of-CATANIASDO_NursingHomesrdquo and ldquoName-of-CATANIASDO_Pharmaciesrdquo

36httptagmediunipiitapitext=chiesa20cattolica20ss20pietro20e20paoloampkey=bc70153a603d9de7e79c244c41270913amplang=it

37httpwwwontologydesignpatternsorgdataprismacentralinasmogpiazza-stesicoro

38httptagmediunipiitapitext=Piazza20Stesicoroampkey=bc70153a603d9de7e79c244c41270913amplang=it

39httpwwww3orgTR2014REC-vocab-org-20140116


Figure 8 Graffoo diagram representing our modelling choice for spatial (spatialthing) objects

Figure 9 Graffoo diagram representing our modelling choice for timetable objects

Figure 10 Graffoo diagram representing how we employed spatial thing into our ontology


Figure 11 Graffoo diagram representing how we employed timetable object within our ontology

Figure 12 Screenshot of the RDF triples with their namespaces describing the entity Chiesa Cattolica ss Pietro e Paolo (a church)

Figure 13 Screenshot of the RDF triples describing the entity Piazza Stesicoro (a square)


Specifically to assure semantic interoperability andcompliance to W3C standards the following transfor-mations have been performed

ndash The names of classes were lemmatised (eg fromldquoPharmaciesrdquo to ldquoPharmacyrdquo)

ndash The names of the datatype properties were alignedwhen they were clearly showing the same seman-tics For example the propertiesldquoName-of-CATANIASDO_NursingHomesrdquo andldquoName-of-CATANIASDO_Pharmaciesrdquo was con-verted in the same property ldquoNamerdquo assignedto the entity classes ldquoNursingHomerdquo and ldquoPhar-macyrdquo

ndash Relations between entities and values (eg strings)that correspond to other entities were representedby object properties and connected to the corre-sponding entity For example the relationldquoMUNI-of-CATANIASDO_NursingHomesrdquo gen-erated by Tabels became a property ldquoMunici-palityrdquo and was used to connect individuals ofthe class ldquoNursing Homerdquo with individuals of theclass ldquoMunicipalityrdquo

ndash The data properties having values clearly as-signed to resources from external ontologies weretransformed into object properties and their val-ues were reified as individuals of specially createdclasses

The alignment was a manual process done by do-main experts Although methods for automatic align-ment exist [47] they are not as precise as human judg-ment Fortunately the number of properties in a smartcity ontology is not as huge as the number of instancestherefore manual alignment is affordable

Both the resulting ontology40 and the data41 arepublicly available The ontology is browsable by LiveOWL Documentation Environment (LODE)42 a ser-vice that visualizes classes object properties dataproperties named individuals annotation propertiesgeneral axioms and namespace declarations from anOWL ontology in human-readable HTML pages43

40httpontologydesignpatternsorgontprismaontologyowl


42Live OWL Documentation Environment (LODE) Version 12httpwwwessepuntatoitlode

43httpwwwessepuntatoitlodehttpwwwontologydesignpatternsorgontprismaontologyowld4e2763

36 SPARQL endpoint and content negotiation

The resulting dataset consist of millions of triplesand can be queried by selecting the RDF graph calledltprismagt on the dedicated SPARQL endpoint44Queries can be made by editing the text area availableinto the interface for the SPARQL query language TheSPARQL endpoint is also accessible as a REST webservice whose synopsis is the following

ndash URLrArr httpwitistccnrit8894sparql

ndash MethodrArr GETndash ParametersrArr query (mandatory)ndash MIME type supported outputrArr texthtml textrdf

+n3 applicationxml applicationjson applica-tionrdf+xml

Data are also accessible through content negoti-ation The reference namespace for the ontology isprisma-ont45 The namespace associated with the datais prisma46 These two namespaces allow content ne-gotiation related to the ontology and the associateddata The negotiation can be done either via a webbrowser (in this case the MIME type of the output is al-ways texthtml) or by making HTTP REST requests toone of the two namespaces The synopsis of the RESTrequests to the web service associated with the names-pace identified by the prefix prisma-ont is the follow-ing

ndash URLrArr httpwwwontologydesignpatternsorgontprisma

ndash MethodrArr GETndash ParametersrArr ID of the ontology object (manda-

tory the PATH parameter)ndash MIME type supported outputrArrtexthtml textrdf

+n3 textturtle textowl-functional textowl-manchester applicationowl+xml applicationrdf+xml application rdf+json

The synopsis of a REST request to the web serviceassociated with the namespace identified by the prefixprisma is the following

ndash URLrArr httpwwwontologydesignpatternsorgdataprisma

ndash MethodrArr GET

44httpwitistccnrit8894sparql45httpwwwontologydesignpatternsorg

ontprisma46httpwwwontologydesignpatternsorg

dataprisma


ndash ParametersrArr ID of the ontology object (manda-tory the PATH parameter)

ndash MIME type supported outputrArrtexthtml textrdf+n3 textturtle textowl-functional textowl-manchester applicationowl+xml applicationrdf+xml application rdf+json

37 Visualization tool for the geo-referencedsemantic data

We provide a visualization tool that shows geo-referenced objects in a map47 A screenshot of our toolis shown in Figure 14 A user can select a set of ob-ject classes in the topleft-corner listbox The listboxin the bottomleft corner shows all objects belongingto the selected classes The user can then choose a setof objects to visualize The selected objects are visu-alized on a right-hand-side map Objects of differenttypes are shown with different colours In the exam-ple in Figure 14 blue objects correspond to schoolsred objects to churches light blue regions to hospitaland light blue lines to optical fibre lines The user canclick on an object on the map for obtaining additionalinformation

Both classes and objects are retrieved from ourSPARQL endpoint All objects that are associated to ashape can be shown in the map The geometry of ob-jects is described by the NeoGeo ontology as previ-ously described Objects are passed to the map by us-ing the Google Maps javascript API48 The tool con-verts the shapes described by the NeoGeo ontology toa KML layer and sends it to the map It also integratesadditional information such as the name of the objectsand possible associations with DBpedia entries

4 Use Cases

The availability of Geo-referenced Linked OpenData enables a variety of scenarios with strong eco-nomic technologic social and ecologic impact Herewe describe two use cases that can aid in better plan-ning and maintenance of facilities and infrastructuresand effective management of emergencies The usecases are built on top of the semantic model we haveproposed in the previous sections which allows a holis-

47Publicly accessible at httpwitistccnritprismageovisualselectvisualizephp

48httpsdevelopersgooglecommapsdocumentationjavascriptreference

tic use of the data extracted from the different hetero-geneous sources and unified under a shared semanticmodel A first use case named BestLocation employsthe Geo-referenced Linked Open Data previously de-scribed to suggest suitable locations for a new facil-ity based on its proximity to existing facilities Thistool can be extremely useful in many situations Just togive some examples it can help an entrepreneur in de-ciding the location of offices for a startup company itcan be a valid instrument for a citizen that is evaluatingthe purchase of a property and it can assist an urbanplanner to locate facilities and infrastructures The sec-ond use case EmergencyRoute focuses on real-timeroad traffic and public transport management Its goalis to support sustainable mobility and especially topromptly respond to urban emergencies from smallaccidents to more serious disasters by adapting theschedules of vehicles in particular assistance vehiclesin the presence of emergencies Both the use caseshave been taken into account because within the Mu-nicipality of Catania the two described problems aretop priorities and several organizations and local insti-tutions are struggling to solve them With the semanticmodel provided within this paper we want to help withpossible solutions and opportunities in a wide spec-trum of domains

41 BestLocation where to locate a facility

An important problem in urban planning concernsdeciding the location of facilities (hospital post of-fices schools shops etc) based on some parame-ters and constraints which include their distance fromother facilities of strategic importance For example apost office located near offices deposits and shops thatneed to send large amount of mail is much more valu-able than a post office located far from the commercialcity center We propose a service that suggests the bestlocation of a facility based on user defined parame-ters This service can be made publically available andhence be a valid tool for every citizen

To introduce BestLocation consider the followingscenario A citizen that is planning to buy a housewants to know the locations of a city that are moresuitable for her considering that (1) she has school-age children (2) she often uses public transportation(3) she wants grocery shops nearby and (4) she is areligious person A good location would be close to aschool a bus stop a grocery shop and a church Forinstance in the map in Figure 15 the green pushpinpoints to a better location than the red pushpin Indeed


Figure 14 A screenshot of our visualization tool Blue objects correspond to schools red objects to churches light blue regions to hospital andlight blue lines to optical fibre lines

the former is much closer to a school a bus stop a gro-cery shop and a church A service that suggests goodlocations would be very useful in the described sce-nario Normally its implementation would requires (i)collecting data about heterogeneous entities and con-cepts (facilities of different kinds distances etc) and(ii) employing an effective and efficient optimizationalgorithm

The data model that we adopted (Section 3) solvesthe data collection problem Information about facil-ities coming from different sources is presented in auniform and organized way Facilities are assigned tofacility types This organization is crucial since theuser is interested to classes of facilities in place of spe-cific instances (eg any grocery shop in place of a spe-cific one) OWL reasoners can also help in definingnew facility types (eg all facilities that satisfy certainconditions) Note that data about facilities are initiallydistributed among different sources and stored in dif-ferent formats For example schools are taken from theGIS while bus stops are provided by means of a restservice in json format Our data model presents theseheterogeneous data in a uniform and abstract way thusrelieving the developer from collecting and aligningdata from different data sources

The problem of choosing a location that is close toa number of given points generalizes the well knownWeber problem [51] for which efficient solutions inEuclidean spaces and metric spaces have been pro-posed In Euclidean space optimization is performedover a continuous (infinite) set of points (eg all pointsin the plane) and hence geometric numerical algo-rithms can be employed [35] In metric spaces (egconsidering the travelling time as a distance) the searchis limited to a finite set of locations and hence astraightforward approach can be obtained by evalu-ating all locations one by one However a straight-forward approach may be unfeasible for large inputssince the set of possible locations (eg the set of allpossible addresses of a city) can be huge hence ap-proximated efficient solutions can help [52] Solutionsfor facility locations problems are dependent on thespecific formulation Some approaches considered inliterature are discussed in [4630]

Here we describe a novel more general model thatdiffers from previously proposed ones in several as-pects First we take into account the type of facilitiesand allow the user to specify its interest for a facil-ity type in place of a specific facility This is advan-tageous since typically a user is interested in stayingclose to at least one facility of certain type (eg at least


Figure 15 Example of ldquobetterrdquo and ldquoworserdquo locations for a living place

one post office) instead of a specific facility (eg thepost office in 345 State street) Defining the interestfor facility types is also more immediate since thereare much less types than instances Another advantageof the model described here is that it enables speci-fying a set of facility types for which it is beneficialto have as much as possible facilities close-by Thiscan be desiderable eg for a company that wishes tohave as much customers as possible nearby We alsointroduce distance constraints that enable maintaininga minimum distance from certain types of facilities

Formally we represent the input as a set F of fa-cilities (anything that has a specific location and canbe relevant for urban planning eg offices shopsschools hospitals etc) Each facility f isin F has a lo-cation l(f) (coordinates in a map) and a type t(f) (egpost office bakerrsquos shop) drown from a set of pos-sible types T

Types can be explicitly mentioned in the ontologyused in the application or they can be resolved by us-

ing either similarity measures or an OWL reasonerOn one hand we can use a semantic similarity li-brary to propose eg the type PostOffice evenif a user puts a constraint such as a place to senda letter On the other hand we can represent suffi-cient conditions as GCI (General Concept Inclusion)axioms in the ontology itself eg given a constraintsuch as Place u existcuisineFrench and theGCI existcuisineT v Restaurant we are able touse a description logic reasoner (eg HermiT49) to in-fer t =Restaurant

We also consider a set of candidate locations L (tipi-cally all buildings andor residential zoning) Everypair of locations l1 and l2 has a distance d(l1 l2) Thedistance can be measured in many ways (Euclideandistance distance in the road graph average traveltime) The average travel time is often a good choiceand can be easily obtained by existing vehicle routing

49httpwwwcsoxacukprojectsHermiT


services (Google Maps or Open Street Map) throughtheir API

We consider a set of constraints and parameters thatare used to establish the set of valid locations and toquantify the ldquogoodnessrdquo of valid locations Constraintsdefine the maximum and minimum distances fromcertain facility types More precisely for each facil-ity type t we consider the minimum distance dmin(t)from facilities of type t (possibly zero) and the maxi-mum distance dmax(t) from the closest facility of typet (possibly infinity) The minimum distance enablesspecifying facilities that the user wants to stay awayfrom For example an entrepreneur that is evaluatingwhere to locate a new shop is probably interested inhaving competitors as far as possible For each facilitytype t the user also specifies the following parameters

ndash cone(t) a value between 0 and 1 that representsthe importance of having at least one facility oftype t nearby (typically facilities that are neededto carry on the activity eg suppliers)

ndash call(t) a value between 0 and 1 that representsthe importance of having as many as possible fa-cilities of type t nearby (typically recipients of theprovided service eg customers)

Parameters and constraints are application depen-dent For a potential property buyer it is important thatgrocery shops schools bus stops and pharmacies areclose by while an entrepreneur may be more interestedto the location of potential customers andor suppliersParameters and constraints can be given directly by theuser or provided by the system as a choice from a setof predefined scenarios In the latter case parametersmay be defined by experts or learned

The goodness of a location G(l) is zero if at least oneconstraint is not satisfied ie there is at least one t isin Tsuch as d(l l(t)) lt dmin(t) or d(l l(t)) gt dmax(t)If l satisfies all constraints G(l) is defined by Eq1

To facilitate reading we summarize the notation inTable 2G(l) is the reciprocal of two sums The first sum

considers the distances from the closest facility of ev-ery facility type The second one summarizes the dis-tances of all facilities of certain types Here we aggre-gate the distances of facilities of the same type by us-ing reciprocals so as to emphasize the contribution offacilities that are close-by and diminish the contribu-tion of facilities that are far away

We are interested in spotting locations that have highG(l) value A straightforward approach would consistin computing the goodness value for all locations and

Table 2Notation of the model for BestLocation

Description Symbol

set of facilities Fset of locations Lset of facility types Tlocation of a facility l(f)

type of a facility t(f)

distance among locations d(l1 l2)

minimum distance threshold dmin(t)

maximum distance threshold dmax(t)

importance of at least one cone(t)

importance of all call(t)

presenting the results on a map with a superimposedheatmap with color hues ranging from green to redThe system could also return all locations whose good-ness is within a certain percent from the maximumHowever these approach requires the computation ofall pairwise distances between locations and facilitiesand hence O(|L| middot |F|) distances Distances in met-ric spaces can be very expensive to compute For in-stance computing the average travel time may requireinterrogating a vehicle routing service that in turn runsan expensive routing algorithm Since the number ofpossible locations is huge (even considering as loca-tions only existing buildings and residential zoningsthe number of possible location would be enormous)and an average-size city can easily contain hundredsof thousands of facilities often this approach results tobe unfeasible

Here we report an exact algorithm for metric spacesthat strongly improves efficiency by discarding pointsthat are not promising The underlying idea is that themetric distance can be lower bounded by an approx-imated Euclidean distance Since the Euclidean dis-tance is much easier to compute it can be efficientlyemployed to discard locations that provably cannot bewithin certain percent from the optimal The algorithmfirst computes an approximated goodness for every lo-cation than iteratively picks and evaluates locationsstarting from the most promising ones For every loca-tion we first employ the approximated Euclidean dis-tance to assess its eligibility as a possible optimal lo-cation The first assessment enables quickly discardinglocations that cannot be optimal Locations that passthe test are validated by computing the expensive exactgoodness

To simplicity of exposition we consider the trav-elling time as a distance metric d However our ap-


G(l) = 1sumtisinT

cone(t) middot minf |t(f)=t

d(l l(f)) +sumtisinT

call(t) middot 1sumf|t(f)=t

1d(ll(f))

(1)

proach is applicable to other metrics We consider anapproximated Euclidean distance dlb() that is a lowerbound for d Such a lower bound is based on physicalconstraints and defined as

dlb(l1 l2) =dEucl(l1 l2)

speedmax

where dEucl(l1 l2) is the Euclidean distance be-tween two locations and speedmax is the maximumspeed allowed dlb is a lower bound for d ie dlb(l1 l2) led(l1 l2) for every pair of locations

An upper bound for G(l) can be obtained by substi-tuting d with dlb in Eq 1 We call it Gub(l) The methodwe propose here computes Gub(l) for every locationl isin L and compares it with the maximum goodnessfound so far Gmax If Gub(l) lt Gmax then l cannotbe an optimal location and hence it can be discardedFor finding all locations within a certain percentage pfrom the optimal we relax this condition and discardall locations l such that Gub(l) lt (1minus p) middot G(lmax)

The detailed algorithm is shown in Figure 16 Themost expensive step is Step 6 that computes the good-ness by using the computational-heavy metric dis-tance This step is executed only for a small subset oflocations (all locations that satisfy condition Gub(l) ge(1minus p) middot Gmax)

42 EmergencyRoute vehicle routing duringemergencies

Despite large amount of research has been devotedto vehicle routing [291976] and today several re-liable vehicle routing systems are available manage-ment of mobility during emergencies from small scaleaccidents to more serious disasters is still an openproblem Indeed emergencies cause drastic changes tothe road map generating inaccessible zones generat-ing traffic jams and reducing the availability of facili-ties and assistance vehicles Response to an emergencysituation requires careful planning and professional ex-ecution of plans [45]

During these events it is essential to quickly locatethe nearest hospitals to obtain the best way out from

Require L T c call pEnsure a set of locations with goodness within percent p from the

maximum1 Compute Gub(l) for every l isin L2 GoodLocslarr empty3 Gmax larr 0

4 for all l isin L ordered by Gub(l) do5 if (Gub(l) ge (1minus p) middot Gmax) then6 Compute G(l)7 if (G(l) ge (1minus p) middot Gmax) then8 GoodLocslarr GoodLocs cup (lG(l))9 end if

10 if (G(l) gt Gmax) then11 Gmax larr G(l)12 end if13 end if14 end for15 return all (l g) isin GoodLocs such that g ge (1minus p) middot Gmax

Figure 16 Compute high-goodness locations

the emergency zones or to produce the optimal pathconnecting two suburbs for redirecting the road traf-fic Within this context two main problems emerge(1) identification of areas affected by the emergency(2) adequate routing of vehicles that take into accountinaccessible zones and corresponding traffic jams Tothis purpose in this Section we propose another usecase which represents a mobile application that imple-ments a collective real-time alerting system to informon the state of roads in urban areas A central sys-tem collects and summarizes the warnings provided byusers and identifies regions that receive a significantnumber of alerts The use of aggregated data providedby the crowd has been proved to be helpful in emer-gency situations such as the Hurricane Sandy and theHaiti Earthquake [3323] The idea is to have an auto-matic system that aggregates data and uses them to per-form ad-hoc vehicle routing and aid emergency logis-tics The collective alerting system can also be used or-dinarily to signal traffic jams accidents or other trapsThis may help people to create local driving communi-ties that work together to improve the quality of every-onersquos daily driving That might mean helping driversavoid the frustration of sitting in traffic jams or justshaving five minutes off of their regular commute byshowing them new routes they never even knew about


EmergencyRoute needs as input the road networkdata about urban traffic and reports about urban faultsIt can also make use of data from other sources In atypical city these data are spread over multiple sourcesand are available in several complex formats For ex-ample in our case study the road network was stored asa set of road segments in the GIS while reports abouturban faults were stored in a mySQL database relatedto the data on the urban fault reporting service (seeSection 32) Besides integrating these sources ourdata model enables conceptual interoperability crucialfor this application For instance our data model as-sociates reports with locations based on the availableinformation (eg address) and align them with roadsegments

Next we describe the mobile application for crowdalerting and the centralized summarization system thatidentifies affected zones Then we describe the routingsystem

421 Mobile alerting applicationThe user (ideally every citizen) is provided with a

mobile application that allows her to give warningsabout accidents traffic jams obstacles and inaccessi-ble zones The app can be integrated with a standardvehicle routing application so that the user is encour-aged to use it By using the app (in background on hercellular phone) the user contributes passively to mea-suring the traffic and obtaining other road data Theuser can also take a more active role by sharing roadreports on accidents advising on unexpected traps oralerting for any other hazards along the way By doingso she provides other users in the area with real-timeinformation about what is to come [37] More in detailthe user can send an alert by providing a textual de-scription of the alert The current location is obtainedby the GPS and automatically attached to the alert orcan be explicitly specified in the form of an address(eg if a GPS is not available) The time is also associ-ated to the message

422 Collective alertingData collected by users have to be aggregated and

summarized in order to provide a simple and clear de-scription of the zones that are affected by certain dis-aster or event The text of all alert messages is pro-cessed and alerts that are likely to refer to the sameevent are grouped together As a similarity measure be-tween alert messages we propose to use the Jaccardsimilarity J(T1 T2) = |T1 cap T2||T1 cup T2| where T1

and T2 are the sets of words contained in the two mes-sages after removing stop words Alert messages are

clustered together starting from the most similar onesuntil a certain similarity threshold is satisfied [34] Af-ter clustering all alert messages that belong to thesame group are associated to points in the road net-work based on their GPS location The road networkis represented as a graph where nodes are locations(addresses crosses etc) and edges are road segmentsSince alert messages are also associated with time theroad network associated with alert messages can beseen as a time-evolving graph with fixed structure anddynamic nodes attributes (number of alert of certaintype) Existing algorithms can be employed to extractsignificant network regions (sub-networks) and corre-sponding time intervals [393812] The output is a setof network regions that receive a number of alert mes-sages significantly higher than normally

The described system can be integrated with exist-ing disaster alert systems such as GDACS50 and inter-faced with other systems through the Common Alert-ing Protocol51

423 Emergency routingAfter emergency-affected zones have been identi-

fied routing algorithms can be made aware of thesezones in order to produce optimal paths that avoid in-accessible zones This can be done by customizableroute planning algorithms [18] In short we excludefrom the road network the zones that are marked as in-accessible based on the results from collective alert-ing Then we execute a standard routing algorithmOptimization techniques can be employed here to re-spond quickly to dynamic scenarios where the accessi-bility to certain zones changes rapidly over time Rout-ing algorithms can support the real-time managementof road traffic and public transportation by provid-ing best alternative routes to find way outs the nearestavailable hospitals or other locations of interest

5 Discussions and conclusions

In this paper we presented the process to collect en-rich and publish LOD for the Municipality of Cata-nia an ontology design pattern for modelling urbantransportation routes methods procedures and toolsfor ensure semantic interoperability and two use cases

50Global Disaster Alert and Coordination System (GDACS)available at httpwwwgdacsorg



for our work related to the project PRISMA We dis-cussed the design tradeoffs designeruser experienceand some of the pragmatic challenges related to amodel that integrates large complex heterogeneousdata sources of the city The used methodology wasimplemented by following the standards of W3C goodpractices of ontology design the guidelines issued bythe Agency for Digital Italy and the Italian Index ofPublic Administration as well as by the in-depth ex-perience of the research participants in the field Thedata are publicly accessible to users through a dedi-cated SPARQL endpoint or by means of calls to dedi-cate REST web services It is notable that the Mayor ofCatania has acknowledged that the work described inthis paper will be widely used by the Municipality ofCatania and foretells that more city data will becomeavailable to be conveyed to our semantic platform Be-sides other organizations and municipalities of Sicilyexpressed their interest in such a tool as they wouldlike to build a similar data source Therefore and tofurther spread our experience to the local communitywe have recently joined the Sicilian Open Data com-munity52 and advertised and reported our work

Prototype applications based on our published dataand related to services supporting transport publichealth urban decor and social services are alreadycurrently under development by other partners of thePRISMA project and it is foreseen that the first appli-cations will be released at the end of the 2015 We havedescribed two use cases The first is a service that sug-gests the best location of a facility based on some userdefined parameters and that may be used for exampleto aid entrepreneurs in deciding the location of officesfor startup companies to assist urban planners in locat-ing facilities and infrastructures or to offer an updatedevaluation of a property to purchase The second appli-cation is related to sustainable mobility and emergencyvehicle routing It is aimed at supporting the real-timemanagement of road traffic and public transport in-forming citizens on the state of roads in urban areasin particular during urban emergencies and redirectingthe road traffic by providing best alternatives routes tofind way outs the nearest hospitals or other locationsof interest

52OpenDataSicilia httpopendatasiciliait

6 Acknowledgments

This work has been supported by the PON RampCproject PRISMA ldquoPlatfoRms Interoperable cloud forSMArt-Governmentrdquo ref PON04a2_A Smart Citiesunder the National Operational Programme for Re-search and Competitiveness 2007-2013

References

[1] Agency for a Digital Italy Linee guida per i siti web dellePA Art 4 della Direttiva n 82009 del Ministro per lapubblica amministrazione e lrsquoinnovazione [Online] httpwwwdigitpagovitsitesdefaultfileslinee_guida_siti_web_delle_pa_2011pdf2011

[2] Agency for a Digital Italy Linee guida per lrsquointeroperabilitagravesemantica attraverso Linked Open Data Commissione dicoordinamento SPC [Online] httpwwwdigitpagovitsitesdefaultfilesallegati_tecCdC-SPC-GdL6-InteroperabilitaSemOpenData_v20_0pdf 2012

[3] H Alani D Dupplaw J Sheridan K OrsquoHara J DarlingtonN Shadbolt and C Tullo Unlocking the potential of pub-lic sector information with Semantic Web technology In Pro-ceedings of the 6th International Semantic Web Conference(ISWC 07) volume 4825 of Lecture Notes in Computer Sci-ence (including subseries Lecture Notes in Artificial Intelli-gence and Lecture Notes in Bioinformatics) pages 708ndash721Springer Berlin Heidelberg 2007

[4] C Baldassarre E Daga A Gangemi A Gliozzo A Salvatiand G Troiani Semantic scout Making sense of organiza-tional knowledge In P Cimiano and H Pinto editors Knowl-edge Engineering and Management by the Masses volume6317 of Lecture Notes in Computer Science pages 272ndash286Springer Berlin Heidelberg 2010

[5] P Barnaghi W Wang L Dong and C Wang A linked-data model for semantic sensor streams In Green Computingand Communications (GreenCom) 2013 IEEE and Internet ofThings (iThingsCPSCom) IEEE International Conference onand IEEE Cyber Physical and Social Computing pages 468ndash475 New York US 2013 IEEE

[6] H Bast D Delling A Goldberg M Muumlller-HannemannT Pajor P Sanders D Wagner and R Werneck Route plan-ning in transportation networks Technical Report MSR-TR-2014-4 Microsoft Research January 2014

[7] R Bauer and D Delling Sharc Fast and robust unidirectionalrouting Journal of Experimental Algorithmics (JEA) 1442009

[8] T Berners-Lee Y Chen L Chilton D Connolly R DhanarajJ Hollenbach A Lerer and D Sheets Tabulator Exploringand analyzing linked data on the semantic web In Proceed-ings of the 3rd International Semantic Web User InteractionWorkshop SWUI 2006 Athens USA 2006

[9] S Bischof A Karapantelakis C-S Nechifor A ShethA Mileo and P Barnaghi Semantic modelling of smart citydata In W3C Workshop on the Web of Things - Enablers andservices for an open Web of Devices Berlin Germany 2014W3C


[10] C Bizer D2R MAP - A database to RDF mapping languageIn Proceedings of the Twelfth International World Wide WebConference - Posters WWW 2003 Budapest Hungary May20-24 2003 2003

[11] C Bizer T Heath and T Berners-Lee Linked Data - TheStory So Far International Journal on Semantic Web and In-formation Systems 5(3)1ndash22 2009

[12] P Bogdanov M Mongiovigrave and A K Singh Mining HeavySubgraphs in Time-Evolving Networks In Proceedings of the2011 IEEE 11th International Conference on Data MiningICDM rsquo11 Washington DC USA 2011 IEEE Computer So-ciety

[13] P Ciccarese S Peroni and M G Hospital The CollectionsOntology Creating and handling collections in OWL 2 DLframeworks Semantic Web Journal 001ndash15 2013

[14] S Consoli A Gangemi A Nuzzolese S Peroni D Refor-giato Recupero and D Spampinato Setting the course ofemergency vehicle routing using geolinked open data for themunicipality of catania In V Presutti E Blomqvist R TroncyH Sack I Papadakis and A Tordai editors The SemanticWeb ESWC 2014 Satellite Events Lecture Notes in ComputerScience pages 42ndash53 Springer International Publishing 2014

[15] S Consoli A Gangemi A G Nuzzolese S Peroni V Pre-sutti D R Recupero and D Spampinato Geolinked opendata for the municipality of catania In Proceedings of the 4thInternational Conference on Web Intelligence Mining and Se-mantics (WIMS14) WIMS rsquo14 pages 581ndash588 New YorkNY USA 2014 ACM

[16] M Deakin Smart cities Governing modelling and analysingthe transition Routledge London 2013

[17] M Deakin From the city of bits to e-topia space citizenshipand community as global strategy International Journal of E-Adoption 6(1)16ndash33 2014

[18] D Delling A V Goldberg T Pajor and R F Werneck Cus-tomizable route planning In Proceedings of the 10th Interna-tional Symposium on Experimental Algorithms (SEArsquo11) Lec-ture Notes in Computer Science Springer Verlag May 2011

[19] D Delling P Sanders D Schultes and D Wagner Engineer-ing route planning algorithms In Algorithmics of large andcomplex networks pages 117ndash139 Springer 2009

[20] L Ding D Difranzo A Graves J Michaelis X LiD McGuinness and J Hendler Data-gov Wiki Towards Link-ing Government Data In Proceedings of the AAAI 2010 SpringSymposium on Linked Data Meets Artificial Intelligence PaloAlto CA volume SS-10-07 pages 38ndash43 AAAI Press 2010

[21] L Ding T Lebo J S Erickson D DiFranzo G T WilliamsX Li J Michaelis A Graves J G Zheng Z ShangguanJ Flores D L McGuinness and J A Hendler TWC LOGDA portal for linked open government data ecosystems Web Se-mantics Science Services and Agents on the World Wide Web9(3)325 ndash 333 2011

[22] E Dodsworth and A Nicholson Academic uses of GoogleEarth and Google Maps in a library setting Information Tech-nology and Libraries 31(2)81ndash85 2012

[23] J Dugdale B Van de Walle and C Koeppinghoff Socialmedia and sms in the haiti earthquake In Proceedings of the21st international conference companion on World Wide Webpages 713ndash714 ACM 2012

[24] EUROCONTROL WGS 84 implementation manual Instituteof Geodesy and Navigation (IfEN) University FAF MunichGermany 1998

[25] A Gangemi E Daga A Salvati G Troiani and C Baldas-sarre Linked Open Data for the Italian PA the CNR Experi-ence Informatica e Diritto 1(2) 2011

[26] A Gangemi and V Presutti Ontology design patterns InHandbook on Ontologies International Handbooks on Infor-mation Systems 2009

[27] C P Geiger and J von Lucke Open Government Data InP Parycek J M Kripp and N Edelmann editors CeDEM11Conference for E-Democracy and Open Government volume6317 of Lecture Notes in Computer Science pages 183ndash194Springer Berlin Heidelberg 2011

[28] C P Geiger and J von Lucke Open Government and (Linked)(Open) (Government) (Data) JeDEM - eJournal of eDemoc-racy and Open Government 4(2) 2012

[29] R Geisberger P Sanders D Schultes and D Delling Con-traction hierarchies Faster and simpler hierarchical routing inroad networks In Experimental Algorithms pages 319ndash333Springer 2008

[30] T S Hale and C R Moberg Location science research areview Annals of Operations Research 123(1-4)21ndash35 2003

[31] P Hall Creative cities and economic development UrbanStudies 37(4)633ndash649 2000

[32] T Heath and C Bizer Linked Data Evolving the Web into aGlobal Data Space volume 344 Morgan amp Claypool Publish-ers Synthesis Lectures on the Semantic Web edition 2011

[33] A L Hughes L A St Denis L Palen and K M AndersonOnline public communications by police amp fire services dur-ing the 2012 hurricane sandy In Proceedings of the 32nd an-nual ACM conference on Human factors in computing systemspages 1505ndash1514 ACM 2014

[34] L Kaufman and P J Rousseeuw Finding groups in data anintroduction to cluster analysis volume 344 John Wiley ampSons 2009

[35] H W Kulin and R E Kuenne An efficient algorithm for thenumerical solution of the generalized weber problem in spatialeconomics Journal of Regional Science 4(2)21ndash33 1962

[36] A Lamb and L Johnson Virtual expeditions Google EarthGIS and geovisualization technologies in teaching and learn-ing Teacher Librarian 37(3)81ndash85 2010

[37] B Liu Route finding by using knowledge about the road net-work IEEE Transactions on Systems Man and CyberneticsPart ASystems and Humans 27(4)436ndash448 1997

[38] M Mongiovigrave P Bogdanov R Ranca A K Singh E EPapalexakis and C Faloutsos Netspot Spotting significantanomalous regions on dynamic networks In SDM pages 28ndash36 SIAM 2013

[39] M Mongiovigrave P Bogdanov and A K Singh Mining evolv-ing network processes In Proceedings of the 2013 IEEE 11thInternational Conference on Data Mining ICDM rsquo13 IEEEComputer Society 2013

[40] Municipality of Catania Il Sistema Informativo Ter-ritoriale [Online] httpwwwsitrprovinciacataniait81il-sit last accessed January 2014

[41] D L Phuoc H N M Quoc J X Parreira and M HauswirthThe linked sensor middleware - Connecting the real world andthe Semantic Web [online] 2011 Semantic Web Challenge(ISWC) 2011

[42] V Presutti E Daga A Gangemi and E Blomqvist extremedesign with content ontology design patterns Proc Workshopon Ontology Patterns Washington DC USA 2009


[43] PRISMA PON RampC project PRISMA PiattafoRme cloud In-teroperabili per SMArt-government Projectrsquos portal [Online]httpwwwponsmartcities-prismait last ac-cessed Nov 2014

[44] L Qi and L Shaofu Research on digital city framework archi-tecture In Proceedings of International Conferences on Info-tech and Info-net (ICII 2001) Beijing volume 1 pages 30ndash362001

[45] D Rafael B Joshua T Ange-Lionel G Bridget N ManWoL Francesco and N Letizia Humanitarianemergency logis-tics models A state of the art overview In Proceedings of the2013 Summer Computer Simulation Conference SCSC rsquo13pages 241ndash248 Vista CA 2013 Society for Modeling ampSimulation International

[46] C S ReVelle and H A Eiselt Location analysis A synthe-sis and survey European Journal of Operational Research165(1)1ndash19 2005

[47] F Scharffe O Zamazal and D Fensel Ontology alignmentdesign patterns Knowledge and Information Systems 40(1)1ndash28 2014

[48] N Shadbolt K OrsquoHara T Berners-Lee N Gibbins H GlaserW Hall and M Schraefel Linked Open GovernmentData Lessons from datagovuk Intelligent Systems IEEE27(3)16ndash24 2012

[49] R Uceda-Sosa B Srivastava and R J Schloss Building ahighly consumable semantic model for smarter cities In Pro-ceedings of the AI for an Intelligent Planet Barcelona AIIPrsquo11 pages 1ndash8 New York NY USA 2011 ACM

[50] A Velosa and B Tratz-Ryan Hype Cycle for Smart City Tech-nologies and Solutions Gartners Inc Stamford US 2014

[51] G O Wesolowsky The weber problem History and perspec-tives Computers amp Operations Research 1993

[52] B Y Wu On approximating metric 1-median in sublineartime Inf Process Lett 114(4)163ndash166 2014


and collaboration are central keys for the integration ofcitizens within the smart city paradigm

The development of a smart city involves a multi-tude of technologies and processes [16] The Internet-of-Things networks of sensors and smart devicesembedded systems the Internet of users and peopleCloud Computing are all determining a deep revolu-tion on transport environment business and govern-ment by introducing new kinds of informational andcognitive processes such as information collection andprocessing real-time alerting and forecasting collec-tive and crowdsourced intelligence and cooperativedistributed problem-solving and learning [50] How-ever the interaction among all actors and these het-erogeneous solutions still remains a challenge Trans-forming our cities into the smart cities of the futureencompasses incorporating technologies and key digi-tal advances and link them with machine-to-machinesolutions and real-time data analytics Collecting dataand transforming them into tangible insights is crucialfor modern innovative smart cities

In this context the application of Semantic Webtechnologies on smart cities data has an extremely highpotential and practical impact [9] They facilitate dataintegration from multiple heterogeneous sources en-able the development of information filtering systemsand support knowledge discovery tasks In particularin the last years the Linked Open Data (LOD) initiativereached significant adoption and is considered the ref-erence practice for sharing and publishing structureddata on the Web [811] LOD offers the possibility ofusing data across different domains for purposes likestatistics analysis maps and publications By linkingthis knowledge interrelations and correlations can bequickly understood and new conclusions arise

Since cities have large amounts of data heteroge-neous in nature and with different quality and securityrequirements research on the opening process datareengineering linking formalisation and consumptionis of primary interest in smart cities [3] The hetero-geneity problem has to be tackled at different levelsOn the one hand syntactic interoperability is neededto unify the format of knowledge sources enablingeg distributed query [32] Syntactic interoperabilitycan be achieved by conforming to universal knowl-edge representation languages and by adopting stan-dards practices The widely adopted RDF OWL andLOD allow us to achieve such a syntactic interoper-ability On the other hand semantic interoperability isalso needed Semantic interoperability can be achievedby adopting a uniform data representation and for-

malizing all concepts into a holistic data model (con-ceptual interoperability) RDF and OWL assist us inachieving the former goal However conceptual inter-operability is domain specific and cannot be achievedonly by the adoption of standars tools and practicesThe large heterogeneous data sources in smart citiesmake the problem even harder as different semanticperspectives must be addressed in order to cope withknowledge source conceptualisations To give an ex-ample addresses of different data entities for our citydata like ldquoHospitalsrdquo ldquoChurchesrdquo ldquoPost officesrdquo ldquoPo-licerdquo ldquoSchoolsrdquo etc are aligned to the same concep-tual entity ldquoAddressrdquo (characterised by properties likeldquostreetrdquo ldquoaddress numberrdquo ldquozip coderdquo ldquocity blockrdquo) In this way it is possible to intercross data and ex-ploit them more providing application developers theopportunity to easily design their city services [49]Semantic interoperability at domain level allows mak-ing sense of distributed data and enabling their auto-matic interpretation The issue of resolving semanticinteroperability among different data sources is movedfrom the application level to the data model level De-velopers are then relieved from the burden of reconcil-ing uniforming and linking data at a conceptual leveland are able to build their solutions in a more sustain-able and efficient way The published data sources aremade discoverable and become accessible via queriesandor public facilities and integrated into higher-levelservices

In this paper we present

ndash a methodology used to collect enrich and pub-lish LOD for the Municipality of Catania a cityin Southern Italy in the context of the projectPRISMA ldquoPlatfoRms Interoperable cloud forSMArt-Governmentrdquo [43] We present the col-lected city data discuss issues around them anddescribe the process to create a semantic modelfor the city data

ndash an ontology design pattern for modelling urbanpublic transportation routes This acts as a novelgovernment data model for smart cities that en-able conceptual interoperability

ndash methods procedures and tools for ensuring se-mantic interoperability during the transforma-tion process For example information about cityfacilities coming from different sources is uni-formly represented as facility types (eg gro-cery shop) in order to generalize user require-ments when needed OWL reasoners can alsohelp in defining new facility types (eg all fa-






2 Literature review





























































orgco



(a)

















































pos






























































dataprisma







4 Use Cases

































Description Symbol












G(l) = 1sumtisinT




1d(ll(f))

(1)



speedmax

































6 Acknowledgments


References




























































2 Literature review





























































orgco



(a)

















































pos






























































dataprisma







4 Use Cases

































Description Symbol












G(l) = 1sumtisinT




1d(ll(f))

(1)



speedmax

































6 Acknowledgments


References












































































































orgco



(a)

















































pos






























































dataprisma







4 Use Cases

































Description Symbol












G(l) = 1sumtisinT




1d(ll(f))

(1)



speedmax

































6 Acknowledgments


References

























































(a)

















































pos






























































dataprisma







4 Use Cases

































Description Symbol












G(l) = 1sumtisinT




1d(ll(f))

(1)



speedmax

































6 Acknowledgments


References


































































































pos






























































dataprisma







4 Use Cases

































Description Symbol












G(l) = 1sumtisinT




1d(ll(f))

(1)



speedmax

































6 Acknowledgments


References











































































































dataprisma







4 Use Cases

































Description Symbol












G(l) = 1sumtisinT




1d(ll(f))

(1)



speedmax

































6 Acknowledgments


References





























































4 Use Cases

































Description Symbol












G(l) = 1sumtisinT




1d(ll(f))

(1)



speedmax

































6 Acknowledgments


References
















































































Description Symbol












G(l) = 1sumtisinT




1d(ll(f))

(1)



speedmax

































6 Acknowledgments


References
























































G(l) = 1sumtisinT




1d(ll(f))

(1)



speedmax

































6 Acknowledgments


References











































































6 Acknowledgments


References























































Producing Linked Data for Smart Cities: the case of Catania

Documents

Transcript of Producing Linked Data for Smart Cities: the case of Catania