Page - 50 - in Intelligent Environments 2019 - Workshop Proceedings of the 15th International Conference on Intelligent Environments

raw data; the data structures and storage, including correlated data; the dimension re- duction issues; the need of visualizing all stages the extraction stages (particularly in exploratory data analyses and conveying the results). Tohave positive results, it is nec- essary to extract the relevant information and avoid the information not useful. For this reason,we are applying some textmining techniques to identify and extract structured representationsfromrawunstructureddatasources.This involvesanapproach tomining documentswith semi-structured information and natural language, aswell as expertise inmachine learningmodels for time-series, clustering, associations, and rule inference. 3.2. KnowledgeGraphs Thekey idea is to explore thedata fromdifferent sources to extract knowledge that can helppolicyauthorities and the tourist industry stakeholders tomakecitizen live smarter, moresustainableandmoreaccessible. Inanutshell, thisdatawillbeacquired,processed and modeled as a knowledge graph [8], resulting in a Knowledge Base in which the entities are represented as nodeswhile contracts and their properties are represented as edges. There are twomain reasons for the choice of this approach. On the one hand, graphsareanatural representation for this typeofdata,whichwill facilitate thecreation of context and the finding of relevant insights. On the other hand, it also constitutes a naturalway for visualizing these data, very close to our abstractions, as opposed to tra- ditional database tables. Thiswill allowcreating intuitive and rich graphical interfaces, which evennon-technical userswill be able touse toperformdata query andvisualiza- tion. TheKnowledgeBasewill also support the building ofmodels that not only bring the patterns that are hidden in the data to light but also allow for the forecast of future needs. 3.3. DataVisualization The capability of supporting timely decisions based on visualizing big data is essential tovariousdomains,namely tourism.This isessential foradecisionmaker tounderstand not merely the events that occur in the specific location, but more importantly, where theyoccur, that is, the spatialdistributionofevents.Weintend toapplyseveralpipelines of data visualization techniques to our accumulated data. A tool is being developed to create amachine-readable specificationof infographics, deployedusingopenstandards, e.g., JSONandlinkeddata,andthatcanbeeasilysharedandlinkedtoothermedia.Also, this toolallowsplottingourdatausingexploratorydataanalysis, suchashistograms,pie- charts, and timeseries.On theotherhand,weareusingaweb-based interactivemethod- ology, to visualize and understand the temporal dynamics of the datamore deeply. By using statistical-learning analysis with geospatial analysis, applyingD3.js, and google mapAPI libraries,weintendtoenhance theunderstandingofanaccessibleandinclusive tourism-relateddata inbothabstract andgeographicalways. 4. Conclusions Wepresentedkeychallengesandasolutionapproach,usingknowledgegraphs,dataanal- ysis, andvisualization, forenablingaccessibleand inclusive tourismdecision-making in aSmartCity context. Theproposed solution is tailored for both policy authorities, aca- M.GomesandP.Novais /AData-DrivenTool toPromoteTourism forPeoplewithDisabilities50