Computers and networks can be used to help people help each other find things. This can be done without requiring altruism, without invading privacy, without loading users with new tasks and with considerable effectiveness judging by published research and some of our own previous work. Right now we are extending Mosaic, Netscape, WWW and proprietary browsers to further explore the possibilities. There are many experiments occurring all over the Web.

Our research has centered upon enriching the presentation of digital objects through graphical depictions of their history-of-use and making simple and easy the collaborative filtering of information. We are most interested in history-of-use and collaborative filtering techniques that add significant value but require of users little or no additional data-entry. For example, document scroll-bars that depict how much various sections have been read and spreadsheet cells colored by how often they have been edited reveal significant history-of-use information but place no additional data-entry tasks on users. We have also discovered that if users are willing to rate the quality and relevance of information items and pool the resulting ratings, then simple techniques will recommend with great accuracy for them new unfamiliar information items.

For well-known examples of community and history-of-use navigation consider Mosaic and Netscape. They enrich documents with history-of-use (i.e., the color and underlining of a link tells if a user has previously visited it) and they contain simple mechanisms for rating the quality and relevance of items (i.e., hotlists) and collaborative filtering (i.e., mailing hotlists and URLs). We are experimentally extending these facilities along lines that have proven successful in other domains. We are testing an enhanced version of Mosaic that supports fast Likert scale rating and automatically pools item-ratings in a community server that can recommend and evaluate URLs. It paints links according to how likely they are to lead to information of interest to a user. We are fielding a WWW community service that reports what people who share interests with the user find interesting.

Copyright 1994 Bell Communications Research