Transcription
MotivationDEKDIVMethodsArchitectureDEKDIV: A Linked-Data-Driven Web Portal forLearning AnalyticsData Enrichment, Interactive Visualization,and Knowledge DiscoveryYingjie Hu, Grant McKenzie, Jiue-An Yang, Song Gao,Amin Abdalla, and Krzysztof JanowiczThe STKO LabDepartment of GeographyUniversity of California Santa BarbaraMar. 25, 2014Demo
MotivationDEKDIVMethodsOutline1Motivation2DEKDIV: A Quick Flyby3Concepts and Methods behind the Scene4Architecture and Design Principles5DemoArchitectureDemo
MotivationDEKDIVMethodsOutline1Motivation2DEKDIV: A Quick Flyby3Concepts and Methods behind the Scene4Architecture and Design Principles5DemoArchitectureDemo
MotivationDEKDIVMethodsArchitectureMotivationThe LAK dataset provides a rich amount of information aboutresearchers, papers, and conferences in the field of learninganalytics.This dataset has already been structured in a machinereadable format (RDF).Demo
ion:How can we design effective tools to gain moreinsights from the LAK dataset?Demo
MotivationDEKDIVMethodsOutline1Motivation2DEKDIV: A Quick Flyby3Concepts and Methods behind the Scene4Architecture and Design Principles5DemoArchitectureDemo
MotivationDEKDIVMethodsArchitectureDEKDIV: A Quick FlybyDEKDIV is a unified Web portal consists of a number offunctional modules for interactive visualization and knowledgediscovery.Demo
MotivationDEKDIVMethodsArchitectureDEKDIV: A Quick FlybyDataset:LAK dataset in RDFConcepts and topics extracted from papers’ full textGeospatial locations for research institutes and conferencesExternal citation data from Microsoft Academic Search as wellas expertise information from ArnetMinerDemo
MotivationDEKDIVMethodsArchitectureDEKDIV: A Quick FlybyInteractive Visualization:Where do the conference participants come from?What are the major concepts in a paper?Who are the collaborators of a researcher?Which researchers and institutes are citing your papers?.Demo
MotivationDEKDIVMethodsArchitectureDEKDIV: A Quick FlybyKnowledge Discovery:Who are the most active scholars in a conference?How similar are two scholars in their research topic space?Who can be suitable reviewers for a newly submitted paper?Who can be your potential collaborators?.Demo
MotivationDEKDIVMethodsOutline1Motivation2DEKDIV: A Quick Flyby3Concepts and Methods behind the Scene4Architecture and Design Principles5DemoArchitectureDemo
MotivationDEKDIVMethodsArchitectureScientometrics and Spatiotemporal ScientometricsScientometrics refers to the science of analyzing andmeasuring a discipline’s research topics, authors, publicationsand so forth.In previous works, we have proposed a framework forspatiotemporal scientometrics which focuses on discoveringthe spatiotemporal patterns of a research discipline (Hu etc.,ISWC 2013; Gao etc., ACMGIS 2013).DEKDIV is an Online scientometrics workbench for the LAKcommunity.Demo
MotivationDEKDIVMethodsArchitectureLatent Dirichlet allocation (LDA)LDA is an unsupervised, generative probabilistic model usedto infer topics from a textual corpus.LDA represents each topic using a mixture of keywords. Eachpaper is described as a number of topics with associatedprobability values.LDA has also been applied to describe the interested topicsfor each researcher.Demo
l Scaling (MDS)MDS is a method to visualize the similarity among individualcases in a dataset.MDS can take multiple attributes of individual cases, and putthese cases into N-dimensional space so that thebetween-object distances can represent the correspondingsimilarity.In this work, MDS has been used to represent the similaritybetween researchers based on their research topics extractedusing LDA.Demo
MotivationDEKDIVMethodsArchitectureGraph and Shortest Path DistanceAn academic graph can be constructed by consideringresearchers as nodes and co-authorships as edges.From each node in the graph, we calculate its shortest pathdistance to other nodes (researchers) using Dijkstra’sAlgorithm.The shortest path distance has been used to represent how far(or how close) that two researchers are connected.Application: finding potential collaborators for researchers.potential sim(a1 , a2 ) (1 1/d)Demo
MotivationDEKDIVMethodsOutline1Motivation2DEKDIV: A Quick Flyby3Concepts and Methods behind the Scene4Architecture and Design Principles5DemoArchitectureDemo
MotivationDEKDIVMethodsArchitectureA Self-contained Modular Design ParadigmThe Web portal is designed as a general framework, and eachfunction is implemented as a self-contained module.Demo
MotivationDEKDIVMethodsArchitectureFunctional ScalabilityUsers can ”plug” or ”unplug” functional modules by adding orremoving the import statement of the correspondingJavaScript files.Demo
MotivationDEKDIVMethodsArchitectureData ScalabilityA customized script can be designed to synchronize data andensure the data in the triple store is up to date.This strategy has already been used in the Linked Data portalof the Semantic Web ).Demo
MotivationDEKDIVMethodsArchitectureUsability DesignSimple and intuitive user interface: Four access points(conferences, researchers, papers, and analytics); ”follow yournose” data exploration.Configurable layout of modules: Each module is displayedin a separate window whose position can be configured basedon users’ preference.Rich help document: Each module is accompanied with ahelp description, and a Youtube tutorial video has also beenprovided to facilitate the understanding of the system.Demo
MotivationDEKDIVMethodsArchitectureTechnology StackClient side: JQuery, ExtJS, HTML5, CSS, D3, Leaflet, .Server side: Java Servlet, Jena API, SPARQL queries, JenaFuseki, .Client-server interaction: AJAX requests and responsesData models/formats: RDF, GeoJSON/TopoJSON, .Performance strategies: dynamic information caching forfrequently requested authors and papers, .Demo
MotivationDEKDIVMethodsOutline1Motivation2DEKDIV: A Quick Flyby3Concepts and Methods behind the Scene4Architecture and Design Principles5DemoArchitectureDemo
MotivationDemoDEKDIVMethodsArchitectureDemo
researchers, papers, and conferences in the eld of learning analytics. This dataset has already been structured in a machine readable format (RDF). Motivation DEKDIV Methods Architecture Demo . Client side: JQuery, ExtJS, HTML5, CSS, D3, Lea et, . Server side: Java Servlet, Jena API, SPARQL queries, Jena Fuseki, . Client-server .
