320 likes | 470 Views
NetLens: Iterative Exploration of Content-Actor Network Data. Hyunmo Kang, Catherine Plaisant, Ben Bederson. Challenges of Network Data Visualization. by Frank van Ham. TouchGraph. Challenges of Network Data Visualization. Complex Analytic Tasks Incremental data exploration
E N D
NetLens:Iterative Exploration of Content-Actor Network Data Hyunmo Kang, Catherine Plaisant, Ben Bederson
Challenges of Network Data Visualization by Frank van Ham TouchGraph
Challenges of Network Data Visualization • Complex Analytic Tasks • Incremental data exploration • Iterative query refinement • Scalability • Common simple UI components e.g. histogram and lists • Generality • Apply to any dataset matching Content-Actor modele.g. digital library, photo collections, email collections, case law, etc.
Data Analysis (Content-Actor Model) Inter-relationship Intra-relationship Intra-relationship Entity E1 (Content) Entity E2 (Actor)
Data Analysis (Content-Actor Model) Inter-relationship Intra-relationship Intra-relationship Content (Paper) Actor (Author)
Data Analysis (Content-Actor Model) Inter-relationship Intra-relationship Intra-relationship Content (Email) Actor (People)
Data Analysis (Content-Actor Model) Inter-relationship Intra-relationship Intra-relationship Actor (Photo) Content (People or Category)
DEMOScreenshots at www.cs.umd.edu/hcil/netlensVideo at www.cs.umd.edu/hcil/netlens/VASTvideo
Task Analysis • Single step tasks • How many papers on “User Study” were published in 1998? • Who are the authors of the papers on “Virtual Reality”, which were published at the CHI 99 conference? • Which paper is the most frequently cited by the papers published at the CHI 04 conference? • Which author is most frequently cited in the “InfoVis” topic? • How many papers were published by UMD HCIL people? • Who are the authors whose nationality is Korea?
Task Analysis • Multiple step tasks • Evaluate individuals: - how many papers were self-referenced? - how frequently was each paper referenced by other papers? • Identify communities: - what are the major paper topics published by UMD HCIL? and who in this group has the most papers in that topic? - how do UMD HCIL’s research interests change over time? and who in this group made that change? • Find experts (to review papers or come to workshop): - who wrote the most papers in the InfoVis topic? and how many papers cited his papers? - whose paper in the InfoVis area is most frequently referenced by other papers? • Learning about a new topic (to find a good PhD topic): - which topic has growing publications? and who contributed most to this topic last 3 years? - what are the other topics the authors in InfoVis area also get interested in? • Where should I go on a sabbatical? - which country (or research group)’s authors most frequently reference my papers?
Design Challenges • History and Integrated Help • Sequence of interactions to accomplish a task (lost in exploration) • “How did I get here?” • “What does the current filtered dataset mean?”
Design Challenges • Multi-layered Interface • Users do not need all the windows • Complexity of data and tasks • Computation efficiency • Users’ usage levels and their preferences • Etc.
Design Challenges • Data Export • Integration of graph visualizing tool TreePlus • Exporting methods • Windows clipboard • Internal graph class object • Xml documents
NetLens Extension (Joint Institute for Knowledge Discovery) - http://jikd.umiacs.umd.edu Emails on the leftPeople on the right Overviews provided for all attributes (here for emotional tone on emails side) Filtered to show only emails related to CAenergy crisis; and the people who sentthem are shown on the right side.
NetLens Extension (Joint Institute for Knowledge Discovery) • Generality and Scalability JIKD data schema
NetLens System Architecture • NetLens • Written in C#, Piccolo toolkit • Database Server • MySQL • ADO.NET driver for MySQL • MySQL connector/NET 1.0.7 • Web Server • Mac OS X Server • Web API • CGI, JSP (e.g. email search, people’s bio, etc.)
Evaluation • Heuristic Evaluation by NIST • Possible directions: • Usability • Measure usability • Speed, performance, • Learnability • Error rates • Power • Comparing range and complexity of possible queries • SQL queries? • Generality • How easy it is to apply new datasets to NetLens
PhotoMesaBrowse, Annotate, and Search Digital Images Hyunmo Kang and Ben Bederson
PhotoMesa Image Browsing • Zoomable User Interface Zooming into a group of photos Zooming into a single photo • PhotoMesa shows all photos in a single view • Bigger preview by moving over a thumbnail • Browse photos by zooming in or out • Dynamic sorting and grouping
PhotoMesa Image Browsing • Zoomable User Interface • PhotoMesa lets you control visible photos • All photos • Unhidden photos • Representative photos • Favorite photos only Show only the representative photos for each group
PhotoMesa Image Browsing • Browse photos in “Scroll” mode with detail photo view • Zoomable User Interface Scrollable Thumbnails Panel Photo Information with EXIF Detail Photo Panel
PhotoMesa Annotating Label objects in the photo (e.g. animals, locations, etc.) Label who is in the photo Add a caption and mark photo as favorite or hidden • People Annotation Mechanisms: Checkbox Annotation, Drag-and-drop Annotation, Hotkey Annotation • Category Annotation: Create user-defined hierarchical structure of object types to annotate your photos with
PhotoMesa Annotating • Bulk Annotation: Annotate multiple photos simultaneously with the same annotation mechanisms
PhotoMesa Searching By keyword By folders By people By category By year By month
PhotoMesa Photo Sharing • Upload • Photos • Metadata e.g.) people, category, photo info, etc. • Remove • Update • Search • Web Services • Browse with web browser • Add comments
PhotoMesa SQL Query • Free Text Search (Find photos containing word “kang”) SELECT Photos.* FROM (Photos INNER JOIN (Categories INNER JOIN PhotosCategories ON Categories.categoryname = PhotosCategories.categoryname) ON Photos.url = PhotosCategories.url) INNER JOIN (People INNER JOIN PhotosPeople ON People.personname = PhotosPeople.personname) ON Photos.url = PhotosPeople.url WHERE (((PhotosPeople.personname) Like "*kang*") OR ((Photos.url) Like "*kang*") OR ((Photos.created) Like "*kang*") OR ((Photos.uploaded) Like "*kang*") OR ((Photos.description) Like "*kang*")) OR (((PhotosCategories.categoryname) Like "*kang*"));
PhotoMesa SQL Query • Add Photo INSERT INTO Photos (url, created, uploaded, description, photomark, thumbnail, width, height) Values(“url”, “2006-04-20", “2006-04-20", “hyunmo’s trip to Seoul”, 1280, 1024); • Add People INSERT INTO PhotosPeople (url, personname, x, y, time) Values(“url”, “hyunmo kang“, “0.1234”, “0.789” “2006-04-20"); INSERT INTO People (personname, lastname, firstname) Values(“hyunmo kang”, “kang”, “hyunmo”);
PhotoMesa System Architecture • PhotoMesa Client • Written in C#, Piccolo toolkit • Database Server • MySQL • ADO.NET driver for MySQL • MySQL connector/NET 1.0.7 • Web Server • Apache • Web API • PHP (photo upload, web services)
Questions? • More visualization projects are available athttp://www.cs.umd.edu/hcil • PhotoMesahttp://windsorinterfaces.com • NetLenshttp://www.cs.umd.edu/hcil/netlens • Email: kang@cs.umd.edu