URN

Participants

• CSDL: Robert Brewer
• Affliates:  NSF

Summary

URN is an Egret-based system designed to address the "information explosion" that occurs through use of world-wide information resources such as the Internet USENET newsgroups, gopher, and so forth. The problems in filtering relevant and useful knowledge from the avalanche of unrelated, low quality, or outdated information posted and accessible is well-known.

URN attempts to exploit collaboration among people with similar interests to support incremental restructuring, classification, and filtering of these information sources. A prototype collaborative newsreader has been implemented and evaluated to test these mechanisms.

Effective utilization of Usenet can be improved through explicitly collaborative efforts among small groups of people with similar interests who work together to retrieve, filter, and ultimately restructure information produced by Usenet into a form amenable to their own needs.

URN is designed to explore the representations and processes needed to provide a model of the interests of individuals within the group that can be used to predict the relevancy of future Usenet contributions. URN users collaboratively and incrementally create a shared, global representation of the content of each Usenet posting, but individually assess its relevancy to their own personal interests. The collaboration minimizes the overhead to any individual of this annotation, while improving the quality of the data used for relevancy assessment.

Significantly, URN does not require users to agree upon a common single measure of relevancy - URN maintains a separate model of each user's interests. Instead, URN users focus their collaborative efforts on building a shared representation of each article's content and structure.

URN was evaluated internally by members of CSDL through a case study involving newsgroup reading. Metrics were collected on user voting behavior in order to show that URN improves the Usenet reading experience.

Software

URN never advanced beyond the prototype stage, and as such the URN software is not available for public release.

Publications 

Status

Keywords

collaborative filtering, classification, evaluation, usenet, adaptive interfaces, large databases.