A Visual Dashboard for Linked Data - Semantic Web Journal

More documents

Recommendations

Info

2 S. Mazumdar, D. Petrelli and F.Ciravegna / A Visual Dashboard for Linked Data: An Exploration of User and System Requirementsto explore linked data are based on a variation of anaggregated list. In some cases this is the best optionas the abstraction of the data itself make it difficult todisplay it otherwise, e.g. information about a personand their work is properly displayed as a (web)pageand applications like Sig.ma 3 are a good example ofuser-centred display. However, although this is a solutioneasy to generalise, it cannot be considered thebest one as the aggregation of data into a list flattenthe depth of the linked data itself and could hide itsmost important characteristics. For example numericdata may be more meaningful and revealing when visualizedas graphs or plots or pie charts than as a tableor list. [5] describes that finding the right visualizationfor a given set of data is not a trivial task: “One mustdetermine which questions to ask, identify the appropriatedata, and select effective visual encoding to mapdata values to graphical features such as position, size,shape, and colour.”. To facilitate this investigation specifictools are needed that support looking at the samedata from different angles and, eventually, take a decisionon which visualization is the most effective forthe task in hand.We propose a tool that, in a dashboard metaphor,provides consumers of linked data different visualizationsto be used simultaneously. We define “consumersof linked data” as both potential users and applicationdevelopers interested in understanding what a specificlinked data set is about. To be effective, such a tool hasto be easy to use but also easy to plug into any linkeddata set made available. This paper discusses the implementationof a generic dashboard-based visualizationframework, Points of View (.views.) and reportson the exploration of two fundamental aspects: Userneeds and System requirements. On the bases of thelessons learnt, we provide basic guidelines for the designof generic linked data applications. The paper isorganized as follows: Section 2 discusses related workand section 3 the design rationale. Section 4 discussesthe user interactions involved in the system. Section5 provides an insight into the implementation. Sections6 and 7 report our findings of user and systemformative evaluation. Section 8 discusses the proposedguidelines. An outline of the future work concludes thepaper.3 Sig.ma Semantic Information Mash-up http://sig.ma/ [accessed18/03/2011]2. Related WorkVisualization of linked data has focussed so farmainly on providing browsers for visualizing RDF.Applications like mSpace [14], user-composed facetsbrowsing for a classical music database, or MuseumFinland[10], a web-based pre-defined facet browsingfor museum collections, provide browsing functionalitieson specific data sets. Some researchers haveaddressed the visualization of data in a generic way:Longwell [2] provides faceted browsing for arbitrarydata, but requires a domain-expert to configure the differentfacets, while Welkin 4 loads RDF models andprovides a graphical representation of the data alongwith lists of predicates and resources. Both RDF Gravity5 and IsaViz 6 are graph-based visualizations of RDFOWL graphs or ontologies: the latter uses graph visualizationsto help authoring RDF data, the former providesa visualization of existing RDF data. Experimentationwith large scale RDF is progressing with severaltools available to try graph 7 .In order to be generic, all these examples map thedata onto highly abstract visualisation structures likegraphs, missing out on the advantage of visual displayfor understanding and reasoning [5]. To providedomain-independent tools that are at the same time familiar,some approaches started from the standard webcontent and enrich it with semantic. PowerMagpie 8adds semantics to web browsing sessions by analyzingthe text of a web-page and maps the extracted conceptsto existing semantic web ontologies. Similarly, Piggy-Bank ([8]) extracts concepts from browsed pages andstores them in a local database for later facet browsingor search, e.g. sorted lists of subsequent links, dates,times etc.Closer to our intended goal of providing a generic,informative and intuitive visualizations on large linkeddatasets are Tabulator [11], /facet [7] and SemanticWonder Cloud [12]. Tabulator [11] visualizes RDFdata on geographical maps, timelines and calendarviews; /facet [7] is a generic semantic browser that al-4 Welkin, http://simile.mit.edu/welkin/ [Accessed 17/03/2011]5 RDF Gravity, http://semweb.salzburgresearch.at/apps/rdfgravity/[Accessed 17/03/2011]6 IsaViz, http://www.w3.org/2001/11/IsaViz/ [Accessed17/03/2011]7 Large-scale RDF Graph Visualization Tools,http://www.mkbergman.com/414/large-scale-rdf-graphvisualization-tools[accessed 17/03/2011]8 PowerMagpie, http://powermagpie.open.ac.uk/ [Accessed17/03/2011]
S. Mazumdar, D. Petrelli and F.Ciravegna / A Visual Dashboard for Linked Data: An Exploration of User and System Requirements 3lows the user to select the facet sequence, results arevisualized as list or on timeline enriched by concepts;and the Semantic Wonder Cloud [12] provides a mindmaplike visualization on DBpedia with large centralconcepts and satellite ones. Although closer to our intent,these three systems differ in core design decisions.Tabulator provides multiple views in differenttabs on a single page thus loading the user with thecognitive effort of remembering the content of a visualizationfrom one tab to the next when exploring thedata, while we intend to support visual comparison byproviding simultaneous views.As /facet and Semantic Wonder Cloud, we intendto provide visualizations of different facets, but ratherthan just displaying a preconceived view, (e.g. a timelinein /facet) or as related concepts in a simplifiedgraph (Semantic Wonder Cloud), we go a step furthermapping the data on several, specific and intuitive visualframeworks (as we attempted in [13]). However,the possibilities for visualizing data are multiple [5],many more of those we explored in [13], therefore alarger range of views is considered here. Worth noticingas it points out in the opposite direction, is the workdone in Exhibit [9]. Instead of a generic visualizationframework, Exhibit offers to the owner a simple environmentfor publishing data visually that a genericuser could look at and interact with. Although this approachis particularly relevant with the current trend ofWeb 2.0 tools and amateur web authorship, it fails insituations where the data owner is just interested in releasingit, but cannot spend effort on (or does not knowhow to) making that data visually accessible. This isthe case for the government data that rely on the goodwill of others to make it graphically accessible.3. The Dashboard Design RationaleA tool that provides the user with a flexible wayto look at the data from many perspectives needs tobe customizable as the most effective type of visualizationhighly depends on the data type and the taskin hand [5]. This design decision on effective customisationled to the adoption of a dashboard layout[3]. A dashboard provides simultaneous visual summariesof large sets of information in a limited amountof space (here, a single web page). Effective dashboardsshould be able to provide all the informationin a meaningful, correct and intuitive way [3]. Whilewidely used in business information systems since the80s, dashboard-like user interfaces are becoming increasinglycommon in other domains only now. Popularwebsites like igoogle 9 and BBC 10 use a design inspiredby dashboard layout, by providing contextualwidgets, each tuned to display a specific set of information.The design rationale embedded in our visualisationsystem Points of View, abbreviated as .views., isto create a dashboard for generic linked data by makingvisualizations available in customizable widgets asshown in Figure 1.As linked data is given for public consumption, itis not predictable which visualization users will findmore useful given their task. For example, governmentdata on schools performance would be better visualizedas individual items on a map for parents tryingto decide the best choice for their children, but wouldbe more meaningful to public servants who want tocompare schools performing trends across the countryif it was aggregated in tables. Therefore multipleviews over the same data seem to be indispensable tosupport the understanding of the value of linked dataand facilitate its use and consumption. Although weacknowledge that not all visualizations are equal andthat a specific view can show or hinder interesting phenomenain the data [5,3], we think it is important to explorethe issue of visualizing linked data as broadly aspossible, leaving the introduction of visualization constraints(i.e. which data type should be visualized, howand for which purpose) for a later stage when the basicframework has been understood. So in this work wefocus on:1. Understanding how a dashboard approach couldsupport the user in exploring unknown datasetsand appreciate the multi-faceted nature of the underlyinglinked data in a short span of time.2. Understand which are the technical implicationsand constraints to provide a generic visualizationservice over linked data, both stored locally or remotelyaccessible via endpoints, and which technicalcontraints affect the user interaction.In summary, our aim is to facilitate the visualizationof a generic linked data set in a way that is familiar,easy to understand and customise.4. User InteractionFigure 1 and Figure 2 show the same interface appliedon different data sets. The different visualisa-9 igoogle interface, http://www.google.com/ig10 BBC website, http://www.bbc.co.uk/
Page 1: Undefined 1 (2009) 1-5 1IOS PressA
Page 5 and 6: S. Mazumdar, D. Petrelli and F.Cira

A Visual Dashboard for Linked Data - Semantic Web Journal

You also want an ePaper? Increase the reach of your titles

Delete template?

Save as template?