Seite - 48 - in Intelligent Environments 2019 - Workshop Proceedings of the 15th International Conference on Intelligent Environments

andafter travel.Mostof thisdata isexternal: forexample, in the formofTwitterorother social networking feeds.Thefirst challenge is thedifficultyof identifying the right data anddetermininghow touse it best since the proliferation of a large amount of unstruc- tured and structured tourism-related open data (see Table 1). The second challenge is to find the rightmix of theory and technology capable of interpreting the data to find meaningful tourisminsights, and the third is toovercometheobstacleofdataaccessand connectivity,which requires the rightplatforms toaggregateandmanagebigdata. Table1. Categorizationofopen tourismdataextracted from[1]. Opendata type Description Wheredatahasbeenused Geographicdata GPS-locations Mobileapplications,websites Eventdata Descriptionofevents,bandsplaying, timeta- bles, even type Mobileapplications,websites Visitor statistics Numberofovernights Mobileapplications,websites Supply statistics Number of businesses, types of businesses, number and information on attractions and museums Mobileapplications,websites Surveydata Data fromsurveystudies Mobile applications,websites, aca- demicandbusiness research Supply information Information on travel destinations, attrac- tions, restaurants andhappenings Mobile applications,websites, aca- demicandbusiness research Transit data Timetables Mobileapplications,websites Governmentaldata Taxdistributionandcollection Mobile applications,websites, aca- demicandbusiness research Allof above Smart Tourism City, augmented reality applications, services that combinedata fromseveral sources 2.2. Analysis Toturnavastamountof informationintoasemanticknowledgeinterconnectedorganiza- tionstructure,whichcanbeanalyzed indepth to identifypatternsandanomalies indata- drivenprocesses andnetworks, canbe a considerable challenge. In otherwords, it is to knowhowtoextractdataandinformationfromdifferentsources, linkingit, integrate into a domain ontology and to include it in a graph-orientedKnowledgeBase. To this end, someapproachesarebeingpursued that canbeused in thecontextof thiswork. Indeed, a setof technologiescouldbeused toconstruct this typeofgraphs, ranging fromgraph- orienteddatastorageandmanagementbuiltonrelationalandnon-relationaldatabases to semantic approachesmade on SemanticWebTechnologies. They are becoming nowa- days substantial assets for representing storing, sharing, and accessing information and domain-knowledge. Examples of graph analyzing techniques such as graph neural net- works [6]anddeep learninggraphs [7]can thusbeusefullyapplied toclassifyand learn fromgraph inputs. M.GomesandP.Novais /AData-DrivenTool toPromoteTourism forPeoplewithDisabilities48