Information Access in Complex, Poorly Structured Information Spaces.

摘要

This research extends our previous efforts on information access (centered around the HELGON system) by choosing an information space which is less structured than the information store represented by HELGON. The domain studied is the News system available on computers. The amount of information distributed and made available through News creates a serious information overload. The conceptual framework behind this research effort explores (a) the relationship between situation models and system model (specifically the question how the system model can be restructured over time to get closer to an individual user's situation model) (b) the role of structure in dealing with large information spaces (specifically: where does it come from, who is willing to generate it, whose structure is it). The innovative system building effort (instantiating the conceptual framework as well as extending it) is centered around the INFOSCOPE system which focuses on the following issues: (a) it allows users to construct virtual newsgroups to reduce the size of the information space (b) it supports the restructuring of the information space at read time according to individual semantics (c) it makes no assumption that senders of a message do any extensive structuring and allows users to impose their own semantics (d) it incorporates agents which assist users (based on information accumulated in a user model) in suggesting better ways to deal with the information space and in restructuring it. The approach taken by INFOSCOPE differs from other approaches which require more upfront structuring. We believe that INFOSCOPE has major advantages in realistic working environments where people are unwilling to spend time and effort on tasks which are of no benefit to them. INFOSCOPE is an operational system and will be used as a general tool in our research group -- providing us with the opportunity to conduct longitudinal studies in a realistic setting.