Skip to main content

Automated information filtering and profiling


The Borges project claims to be the first real-life pilot project on Internet information filtering in Europe. An information filtering service will be developed for messages disseminated by means of Usenet News and the World Wide Web (WWW). These information sources have until now been treated as individual information sources, obliging end-users to dedicate a really great amount of time to read and discard messages and to search and browse for relevant documents at known and new servers.
Impact and expected results:

In defining the role and function of a library filtering service - and implementing it - BORGES will help libraries to exploit new information channels such as the Internet.

Using the expertise in profile design of specialised librarians and the technical support of the information filtering technology developed by the technical partners, libraries will also be able to augment their professional information broker role by offering computerised filtering services.


The prototype was made available early in the project, in order to assess user reactions to it and to modify the system accordingly.

The software, and reports directly appertaining to it, will be restricted issue - as are reports on selection of user groups, end user requirements and filtering service evaluation.

In the public domain will be reports on:

Information filtering service in libraries;
USENET and the World Wide Web as information sources;
BORGES architecture;
Profile management;
"Lessons learned" (for library audiences);
BORGES client;
Final report.
Technical approach:

Borges brings together libraries and technology providers in an attempt to design, implement and deploy on an experimental basis initially information filtering services in libraries to determine (ir)relevance of messages to help users manage dynamic information streams effectively. Users will be able to specify their information need by means of an information profile, which is matched by a filtering server against incoming streams of messages. The service will integrate and build on state of the an information retrieval software and Internet access tools.

A process of continuous evaluations effectively tries to make Borges a user-driven or functionality-driven, rather than a technology-driven project.

There are four workpackages in the project:

Filtering service design and evaluation of USENET and WWW servers, selection of user groups and requirements analysis, user feedback;
Filtering technology framework - client-server architecture and software development and implementation;
Filtering research and development with profile management and reiterative modifications;
Dissemination via reports, conference papers, journal articles, Internet publishing and a specially convened workshop.

Key issue:

The growth of the Internet is tending to by-pass traditional information providers such as libraries and publishers. The amount of data which it holds and supplies is also vast and varied, making it difficult for end-users to find relevant information.

BORGES addresses both problems by:

Defining a library filtering service and integrating the electronic data Internet into that service;
Investigating how such a service will affect the information supplier role of libraries;
Implementing the system and evaluating user feedback to reiteratively improve the service;
Analysing the potential as information sources of two Internet systems: USENET and the World Wide Web;
Providing filtering by means of managed profiles, which can be specified and updated directly by the end-user;
Making profiles dynamic, so that users can feed back judgements on the relevance or non-relevance of delivered information for the progressive improvement of the accuracy of subsequent filtering.

Documentation can be obtained from the contact below and from


Digital Equipment B.V.

3434 KK Nieuwegein

Participants (3)

CAP Gemini Innovation
Universitat Autonoma de Barcelona