ASIST2006: Search Result Visualization Search Result Visualization
Monday 8 am.

3D Information Visualization: an Introduction and Practical Applications
Brad Eden
UCSB
(he said to google for his slides, this link goes to his LITA slides which appear to be very similar: http://www.ala.org/ala/lita/litaevents/litanationalforum2005sanjoseca/30_Eden.ppt

“the use of computer-supported, interactive, visual representations of abstract data to amplify cognition” (quote from standard text which he appropriately cited but I didn’t catch)

Students and future university students are doing 3d visualization right now and every day in gaming. Scientists are using 3d viz in biomedicine, etc.

Discussed dimensions and ways of doing visualization linked to http://otal.umd.edu/Olive/.

Demoed 2-d topic map and AquaBrowser (the Lexington Library implementation), mentioned Stanford’s grokker implementation.

Live plasma http://www.liveplasma.com/ – music 2d search

CubicEye – 3d, 5sided web browser

See his ALA tech report(s) https://publications.techsource.ala.org/search/index.pl?q=Brad%20Eden&queryType=pdfAuthor&results=10

Also recent report on visualization in the humanities

Information Visualization in Large Scale Digital Repositories
Linn Marks Collins
LANL

Context national lab with national security mission and security concerns. Special library setting. She is working IAW the NSF Cyberinfrastucture vision for the 21st Discovery (7/20/06, v.7.1)

They locally load WoS and other research databases so that usage information isn’t tracked by the vendor.

Users are applied scientists and engineers. Problem solvers and higher-order thinkers. (finally a presenter who understands scientists and engineers and their sophisticated interaction with data – a woman after my own heart!)

ActiveGraph
Interactive scatterplot, papers by citations, color indicates subject. Can see outliers, and apply log transformation and see the papers that were cited 3 or fewer times. Their library produces this information quarterly to lab management (publish information impt to the field not just their career). Click on a point and get metadata. They map the articles to another taxonomy bcs the isi subj cats don’t represent their work.

She’s got a group in her library that is trying to improve on citation analysis by weighing (?) a general metric using some indication of reading of the paper.
(read-only version, newer version with read-write ability to add data points, edit metadata, etc.)

(for more, see their paper in the International Journal on Digital Libraries last year, special issue on infoviz)

ScienceSifter
Ah-ha – for their search, they’ve found that their scientists don’t like the “easy search” they want very detailed, precise search . But why should their users have to search at all when the library knows about the projects?
Hyperbolic tree model using RSS feeds of locally loaded tables of contents. (urchin from Nature publishing group) Aggregate and filter feeds.

Multiple viewing options
- channel editors (reference librarians who figure what users need – extended and repeated reference interviews to find information profile of a project group)
- save items to view later (bcs know that researchers only look for info at certain times in their projects)
- visual overview or list
- hyperbolic treemap – can zoom to get info
? no author identification in this interaction – is that a problem?

Channel editors can highlight items that are particularly important, save important things for later.

Now just bibliographic records but will be working on datasets the same way.

See also her paper from IEEE International Conf on e-science and Grid computing, Melbourne 2005. (http://www.gridbus.org/escience/escience2005/index.html)

Future
- escience data deluge
- need people who understand digital libraries and metadata who also understand the science and the natures of the datasets
- grid, projects like sky survey and sensor arrays

Collective Intelligence & Holistic Sense making
Chaomei Chen
Drexel
Collective intelligence, sum of the parts is greater than the whole, an emergent property of a group of people (society, invisible college, etc)

Questions you might ask
1. hot topics?
2. how are hot topics related?
3. how do these topics evolve over time, space?
4. how do we access the emergent insights?

Example, terrorism-
Looking for turning points and insights using visual science, color, and structure.
Looks at connections between two thematic clusters.

Geospatial patterns of terrorist events using google earth, or geospatial patterns of relevant research (does proximity matter for co-citation?)

Glea S et al (2002) How do the citations to this paper spread geographically? (yeah, but, doesn’t it seem like affiliation data would be more useful—like if they all attended the same meeting? Or belong to the same professional society or discussion list? In 2006 geography is only important when travel budgets are limited so perhaps for well-funded research programs will cite the papers first because they will attend more meetings… maybe that’s an interesting research question)

Amazon customer reviews The Da Vinci Code
Took reviews, looked at positive and negative (I believe by computer analysis), then pulled out word frequency. In context “really good book” could actually be referring to other books mentioned in review (IOW, “this book was awful, if you want to read a really good book, read…”).

Decision tree method to determine in terrorism-related abstracts from WoS when something is new and emerging as a theme.

Retrieval vs. Visual Analytics
Retrieval – recall, discrete search, part, formal
Visual – recognition, continuous foraging, whole, intuitive, overview and then zoom

http://www.pages.drexel.edu/~cc345


asist2006

Labels: asist2006

  ¶ 10:48 AM| |cites (technorati) |