Networked information management

Abstract

Every day millions of people trawl the Internet for information using any one of a dozen or more different search tools.Whether they find what they are looking for may depend not only on their skill, but also on their luck. In the corporate arena, organisations are making increasing amounts of information available via intranets. This paper looks at the limitations of current networked information management technology, in particular, shortcomings in the areas of retrieving, organising and sharing of information, and an information management process which would overcome these problems is described. The representation of a user's information needs and interests in a user profile is seen to be central to the process and work in this area, including a novel, non-explicit approach to the representation of profiles, is covered. Four information access systems developed at BT Laboratories are discussed and the extent to which these are currently able to support the information management process is considered.

A bandwidth friendly search engine

“…The metrics were the number of pages accepted into the correct category, the number of pages failed (not categorised that should have been), and those ambiguously categorised (accepted into more than one category, both where this double classification was valid and where one of the classifications was incorrect). Double classification is characteristic of subject classification [10], however, this is not necessarily a problem since many 'distinct' subject areas overlap. For instance, if the page subject is an icebreaking trawler then from our categories, oceans and transport could both be deemed correct classifications.…”