University of Pittsburgh

Koshman, S. (2006) Exploring Subjective Satisfaction for Information Visualization Evaluation. Submitted.
Koshman, S., Spink, A., Jansen, B.J., Blakely, C. & Weber, J. (2006) Metasearch result visualization: an exploratory study. Accepted.
Spink, A., Jansen, B.J., Blakely, C. & Koshman, S. (2006) Overlap among major search engines. ITNG Web Technologies Track Conference, Las Vegas, Nevada. April 10-12, 2006 Forthcoming.
Jansen, B.J., Spink, A., Blakely, C. & Koshman, S. (2006) Prediction of Query Modifications During Web Searching Using Probability Transition Sequences. Submitted.
Spink, A., Jansen, B.J., Blakely, C. & Koshman, S. (2006) Web search engine overlap and uniqueness. Submitted.
Jiang, T., Koshman, S. & He, D. (2006) Heurstic Evaluation of University Websites’ Information Architecture. Submitted.
Spink, A., Jansen, J., Blakely, C. & Koshman, S. Web searcher interactions iwth multiple federate content collections. Australasian Document Computing Symposium, 2005, December 12, Syndney, Australia. 
Koshman, S., Spink, A. & Jansen, B.J. Using clusters on the Vivisimo web search engine. HCI International 2005, July 22-27, Las Vegas.
Jansen, B.J., Koshman, S. & Spink, A. Repeat visits to Vivisimo.com: implications for successive Web searching.  ASIST 2005, Oct. 28-Nov 2, Charlotte, NC.
Spink, A., Koshman, S., Jansen, B.J. Park, M., & Field, C. Multitasking web search on Vivisimo.com. IEEE ITCC  Proceedings, April 11-15, 2005, Las Vegas.

Book Chapters

Koshman, S. (2006).  Information Architecture.  Sage Handbook of Information Science and Management.  Book Chapter Forthcoming.

Technical Reports

Academic  Activities
Program Committee ITNG 2006 Web Technologies Track.
Program Committee IEEE International Conference 2005 on Information Technology. Web/IR Track.
Peer Evaluation of Teaching Committee 2006
Planning and Budgeting Committee 2006
Reviewer, International Journal of Computer Applications in Technology
Reviewer, Journal of the American Society for Information Science and Technology
Faculty Advisor, American Society for Information Science and Technology Pitt Student Chapter
Supervisor for Graduate Student Independent Research on User Testing a Circular Visualization.
Co-supervisor for Graduate Student Independent Research on Developing a Visualization Interface for Query Formulation.
Supervisor for Graduate Student Independent Research on the Peace & Security Studies Thesaurus Project.



Visualization System Reference List
My visualization research concentrates on investigating user interaction with the system for information retrieval.  As operational visualization tools move into the public domain, my research has implications for interface design, Web retrieval, information architecture, and digital libraries. This initial compilation presents a visual reference to prototype and operational systems that are primarily available on the web.  A brief introduction and a link to the system are given.  If you would like to add your visualization system to this list, please contact me at skoshman@sis.pitt.edu.

		Missing Pieces is a visualization tool developed by InfoSpace, Inc.  It shows the overlap among search engines and the meta search engine, Dogpile for the the first page of Web results.  The search engine names can be clicked on to view those items retrieved by that search engine, and it highlights the other "missing" items that would not have been retrieved by using only one search engine.
		The KwMap visualizes related keywords for Web searching.  The more relevant keywords that match the query are placed toward the center of the intertwining lines in the navigator panel.  A keyword may be clicked on to recenter the visualization around the selected term.
		Marumushi's Newsmap uses a treemap to visualize Google news.  The more important news items are given prominence in the display and headlines may be clicked on to go to the news article.
		The Google vs. Yahoo visualization is an experimental comparison tool for search results among the two search engines.  The dots can be clicked on to go directly to the Web sites.
		Mooter provides an initial graph representation of clustered search results  referred to as a "starburst".   The Mooter algorithms are based on psychological models of the user's information seeking process.
		Anacubis provides a graph visualization for depicting business information.  The company entered as the query appears as the center node and the surrounding nodes are assigned different icons to correspond with different colored lines representing information such as company personnel, address, web site, and its competitors. The icons may be clicked on to retrieve textual information.  Anacubis features a "Google-enabled visual search" for Web searching.
		Grokker uses colorful nested circles (or squares) to cluster search output and individual items in each category.  Search sets may be saved as maps and retrieved for future use.  Grokker is powered by Yahoo! Search for Web searching.
		Kartoo is a meta search engine that uses cartographic principles to display search output. Retrieved Web pages and topics are visualized on a document map.  Relationships are depicted by the use of connecting lines among the items when the item is selected by the user.  The text frames correspond to the topics and sites found on the maps.
		Webbrain uses a graph design based on the hyperbolic tree map model to visualize the search terms and resulting categories in a focus+context display.  As users click on various category labels the display shifts the term to the central point and a new category of information is shown.
		The TouchGraph visualization uses a spring-layout technique to present a similarity-based representation.  TouchGraph allows the user to zoom out or in to the graph. Nodes may be expanded to show additional linked nodes or they may be clicked on to view additional information in a pop-up window.
		The AquaBrower Library interface is designed for library online public access catalogs.  The search term is visualized in a separate frame using a focus+context technique in a word constellation.  Term attributes are represented by the color, size, and connecting lines.
		Google News is visualized by Stamen to show a representation of nouns in the daily news.  The color and size of the "color chip" represents the item’s increased or decreased coverage over time and the extent to which it is covered.  When a square is clicked on the related items from several days are also highlighted and a sparkline graph showing the extent of news coverage appears at the bottom of the display.  The topic's news article may be directly retrieved.
		The VIBE (Visualization Information Browsing Environment) was developed by researchers in Molde College, Norway and at SIS.  This image shows a screen shot from the initial prototype system which I used to conduct the first formal VIBE usability study.  Information is visualized by the retrieved set of items and the query’s keywords, which are graphically depicted on the screen as geometric icons.  The influence is visualized spatially by the proximity of the document icons to the user-defined points of interest (POIs) which represent keywords used in the query.  Information about VIBE's variants may be found at WebVibe, BIBE, and VR-VIBE.
		MapStan uses a street map metaphor to display Google similarity search results.  Web pages are clustered on different colored circular site icons called "squares"  which are linked together via "streets" that visualize similarity relationships among the retrieved items.  The wider the street, the more results contain items in the squares.  Web sites may be accessed by clicking on the items presented in the map.
		Inxight is a spin off company from Xerox Parc that features visualization-based products for information management.  This example shows Inxight's Startree visualization based on the hyperbolic tree map as implemented by LexisNexis. In this example the StarTree is used to visualize the online sources available at LexisNexis.  It is an interactive map and users can click on a node, view expanded nodes, or view the content represented by that node.  Another example is found at the National Science Digital Library to visualize their collections.

Additional Visualization Links
Descriptions of visualization interfaces and projects are found at the following sites.

TileBars
TileBars visually parses full text documents into document rectangles of varying lengths that correspond to document length.  The frequency and distribution of terms within each document rectangle are presented as rows of varying gray-colored squares.  The darker the gray square, the greater increase in term frequency.

Envision
The Envision visualization interface was applied to a digital library project and provides a grid overview of document sets in response to a query. The x-axis represents the document type and the y-axis displays the index terms.

VisIT
A visual interface for searching Internet information developed at the University of Illinois, Urbana-Champaign and the National Center for Super Computing Applications.

Visual Information Systems Center (VISC). SIS, University of Pittsburgh.

Human-Computer Interaction Lab.   University of Maryland.

Visualization and Usability Group (VUG).  National Institute of Standards and Technology (NIST).



Last updated 03/10/06.