• Personal tools
  • Log in / create account
• Navigation
  • Main Page
  • Projects
  • Resources
  • Feedback
  • Recent changes
• Toolbox
  • What links here
  • Related changes
  • Upload file
  • Special pages

Ungava

From CISTI-ICIST LAB WIKI

Principal Investigator: Glen Newton, glen.newton@nrc-cnrc.gc.ca

Project Ungava is a test-bed for innovative search, indexing, navigation and visualization of library catalog and scholarly journal article metadata and full-text, as well as external semantic content, in the context of knowledge discovery and exploratory search.

Ungava1 includes:

• Union indexing
• Search and browse interfaces
• Visualization and other information enhancing user interfaces
• Results clustering (forthcoming)
• Base for additional CISTI Research activities, e.g. Citation analysis, Recommender services, etc.
• Semantic Web experiments (Biomed, Chemistry, others)

Contents 1 Try Ungava Demo 2 Presentations on Ungava 3 About 4 History 4.1 Ungava v0.9 4.1.1 Features

Try Ungava Demo

• Using NRC Research Press Journal metadata and full-text.
• Using Colorado State University Libraries Catalog metadata.
  • This is a collaboration with Naomi Dushay, Coordinator for Research and Development, Colorado State University Libraries.

NB: The Simile-based parts of Ungava seem to not run well (or not at all) on web browsers that either do not implement, or do so poorly, the Open standards for XHTML, ECMAScript, DOM, SVG, etc. Ungava works best under Firefox >=2.0. You can get Firefox here. Other more Open standards-aware browsers such as Opera might also be considered viable options.

Presentations on Ungava

• Coalition for Network Information (CNI) Spring Task Force Meeting April 2007 PDF

About

• Features
• Ungava uses Apache Lucene as a back-end
• Ungava also experiments with Project Simile's Exhibit and Timeline.

History

Ungava v0.9

Release date: 2007 July 15

Features

• Implements COinS. If you are using FireFox and have installed, say Openly's OpenURL Referrer, you should be able to find documents using an OpenURL resolver.
• User interface

• Drill clouds for iterative search refinement for author, keyword, journal, year. Example.
• Simple removal of accumulated search query terms
• Rank by relevance (default) or date
• Find similar documents.
• Refine search. Add to the existing query. Example: Add fulltext:canola to existing search author: Glick, Bernard R. Results.

• Keyword drill-down
• Keyword drill-down toggle expansion into full-text, title, abstract
• Author drill-down. In the results record, clicking on an author will search for that author.Example: Glick, Bernard R

• View abstract toggle
• Project Simile

• Simile Timeline view of top 100 search results
• Simile Exhibit view of top 100 search results, with author, year, keyword, journal metadata available.

• In Journal navigation:

• Keyword drill clouds for: Journal, volume and issue. This shows an overall keyword cloud for each of these, allowing to get an keyword-based overview of either the entire journal, volume or issue.
• Author drill clouds for: Journal, volume and issue. This shows an overall author

• Indexing

Issues

• Loader did not use Unicode in connection to MySQL DB for article metadata and abstracts, resulting in garbage characters in some fields. These usually manifest as question marks (?) but some show as other wrong characters.

• Stemming was not done on keywords, so the keyword tag cloud shows words which should have been stemmed down to one word, like receptor receptors and rat rats in this example.