Fox's Statement -
Revised based on workshop participant statements

Challenges

Individualize:

Characterize, empower, and provide individualized support for (all WWW) users --- handling differences in:

• age,
• cognitive style,
• culture,
• degree of interest/involvement,
• education,
• experience,
• group involvement,
• language,
• physical capabilities,
• preferences,
• social context.

Support complex needs:

Study, characterize, generalize and formalize ways to handle:

• anomalous states of knowledge,
• problems,
• user information needs,

while

• going beyond queries,
• managing uncertainty / imprecision / ambiguity / incomplete information, / varying quality, dealing with various types of structuring / lack thereof,

with

• clusters/classifications,
• constraints,
• grammars,
• patterns/templates/cases,
• schema.

Model:

Develop a model or model framework for supporting users' information needs which:

• abstracts/generalizes current models,
• allows dramatic spreads in objects involved (from word to concept to composite object to encyclopedia to special collection),
• allows users to filter/download/organize/build link sets,
• characterizes fully the many relevant situations,
• considers context,
• covers full-text/multimedia/hypermedia information,
• facilitates and benefits from collaboration (e.g., http://www.cs.umn.edu/Research/GroupLens),
• handles time variation from static to dynamic,
• spans from anonymous transactions to sessions to life-long research support,
• spans from passive (e.g., routing) to active (e.g., end-user searching) involvement,
• spans from personal information space management to global digital libraries,
• includes catalogs and finding aids,
• incorporates learning by the user and system,
• supports multiple overlapping/conflicting sources.

Opportunities

DL-Eval Effort:

• Undertake a large-scale Digital Library Evaluation effort (e.g., TREC+HCI+social use)--- involving usability studies, log analysis, and other methods --- to determine requirements to support user searching and to characterize and specify user tasks that implicitly require search support.
• Note that this must be done with new search and digital library systems, so we go beyond old studies of Boolean and bibliographic systems, to cover: vector and probabilistic searching, relevance feedback, full-text and multimedia access, realistic size collections, innovative interfaces, ...
• It might select a small number of in-use DLs after reviewing / building a taxonomy of those extant.
• It must identify multiple evaluation criteria (user available time, amount of social interaction, amount of learning, recall/precision, cost) and encourage formative evaluation and iterative improvement.

WWW:

Capitalize on the interest in Internet and WWW to connect searching work with closely related areas:

• browsing/visualization,
• CSCW (including newsgroups, listservs),
• database management,
• filtering,
• hypertext/hypermedia,
• knowledge bases,
• link semantics,
• resource discovery in distributed environments,
• thesauri / concept spaces.

Call for a new NSF Office / program to solve WWW problems with publishing, search, navigation, browsing, interfaces, collaboration, and scale-up.

Improved Architectures/Protocols:

Leverage lessons learned from Tipster, TREC, DLI and related efforts --- to specify and maybe even standardize on:

• an architecture,
• suite of agents,
• set of components or APIs,
• set of protocols related to user searching,

that will enable various groups each to focus on building an agent or module that will later fit into the larger client/server search framework (with interoperability and interchange designed in), probably with a focus on hybrid / combined approaches.