Knowledge and Distributed Intelligence

Workshop
RECOMMENDATIONS

I. Science Issues and Barriers to Inter- and Multidisciplinary Collaboration

A. SCIENTIFIC OPPORTUNITIES

(What will be useful from KDI/What are examples of good problems?)

Enhanced capabilities for the description, analysis, simulation and understanding of complex systems, including those associated with:

global climate
natural ecosystems
the brain
genome
proteome
human organizations
astrophysics
complex materials

Creation of interactive decision making - problem solving environments, enabling:

design of systems/solutions in conjunction with the "customer"
facilitated access to distributed tools
interoperability of data and knowledge bases

Refinement of approaches to the use and creation of distributed knowledge by study of the art and science of collaboration and dissemination of information about new approaches to foster more successful sharing of tools and information.

Principles of Instrumentation to support collaboration including:

shared research facilities/instruments/tools
telecommunications tools adapted to facilitate interactions and sharing in multisite/multimedia scientific collaborations

Principles and practices for experiments involving:

different numbers of collaborators
different numbers of data acquisition devices
different numbers of databases

MOLECULAR TO MACROSCOPIC BIOLOGY
An example of an extremely complex multi-scale multi-dimensional problem area that transcends scales of dimensions and time from molecular dynamics to sequence to pan dimensional biological structure and function to integrative biology.

Immediate infrastructure deficits identified here include:

the inability to interoperate across data and knowledge
a generalized lack of integration of currently available and rapidly accruing data
an inability to effectively interact with local or potentially remotely accessible instrumentation for data acquisition and the linkage of this streaming data with knowledge bases
a lack of easy and generalized access to tools for these tasks

Possible KDI questions in the above include:

Propose methods to integrate knowledge: develop technological tools (e.g.. need to deal with representation issues, visualization, remote control of instrumentation, tools to cross semantic barriers)
build easy interfaces for the community which facilitate education and communication across hierarchy of biologists and dimensional scales of biological structures

ATOMIC TO MACROSCOPIC MATERIALS SCIENCE

Key issues:

Dealing with complex materials
Universal control of instrumentation

Needs:

integrate tools with interfaces
integrate knowledge into databases
easy access to tools
rational design of optimal experimentation
integrating instrument specific knowledge
increased connectivity to instrumentation
user education

What to expect from KDI:

new phenomenon
new materials
cost minimization

B. BARRIERS TO COLLABORATION AND POSSIBLE SOLUTIONS

Culture of communities does not foster interdisciplinary work. Suggestion: Reward brave cross-disciplinary workers. Recruit visionary leaders with proven abilities to lead multidisciplinary teams.
Language and abstractions are not common. Suggestion: Develop semantic interoperability tools
Transfer and examination of large heterogeneous data sets is too difficult. Suggestion: Explore new strategies to overcome this problem such as decorrelation and compression techniques.
There are cultural/operational barriers to multidisciplinary work at NSF and other federal agencies. Suggestion: Make competitions across directorates at NSF
Inability to perform complex queries across diverse data sources and software systems. Suggestion: Language translators and wrappers for design and optimization
There are too few interdisciplinary educational opportunities. Suggestion: Support cross-disciplinary curriculum development and hands on training
Instrument specific (diverse) interfaces to shared or sharable instruments are needed to facilitate use. Suggestion: Development of "transparent" user interface tools and training tools. (Interfaces with common look and feel despite connection to various instruments used for similar purposes; e.g., electron microscopes, spectrometers, etc. )
There is not sufficient or non uniform trust in the validity of knowledge accruing in different scientific domains. Suggestion: Enlist scientific societies and communities to help establish merits/validity of data deposited.
We are lacking tools and strategies for navigation and exploration in large multidimensional databases or diverse/distributed databases. Suggestion: Creative cueing,.....

C. THE KDI PROCESS

(What kind of grants - programs should be considered and might be most effective?) Suggestions included research programs which would:

create or leverage on existing collaborative and interactive environments.
are interdisciplinary as required by problem/topic (e.g.. materials + biology + computer)
demonstrate vision regarding how connectivity and interactivity are going to impact one or more fields or major science or engineering problem.
have very strong educational component which aims to create multitalented scientists and engineers.
present strongest possible scientific rationale
present strongest possible technological rationale
create multidisciplinary groups to work on targeted problems with very strong project/program leaders.

II. Technology Issues, Existing Enabling Infrastructures, Targeted Areas for Enhancements

A. Providing/ expanding meaningful (to user) bandwidth

Making databases extensible, searchable with new types of (as yet unanticipated) queries.
Developing search engines which adapt to individual userÕs hardware and software and network sophistication
Adapt to the userÕs sophistication level.
Provide access to distributed resources in real time.

B. Develop methods for negotiation/cooperation/ education with 2 or more people

Develop methods for effective visual, audio, tactile, and other sensory feedback-- avatars, transducers, etc.
Make progress on the human scheduling problem--getting people together, removing interruptions, motivate concentration and attention, etc.
Develop methods for scheduling people, instruments, networks, computation, etc.
Provide ways for asynchronous use; recording, archiving, editing, distribution of multi-sensory, multi-modality streams.
Implement adaptable translators

C. Study user interface to advanced instruments, including simulations, data mining.

Navigation, every citizen access, and very high resolution displays are important research areas.
Development of very high level programming languages, capabilities of self-documentation and self-training support materials.
Database organization and access (e.g., time-dependent data access, adaptive mesh refinement)

D. Provide networking programming--beyond scheduling

Achieve guarantees of performance
Quality of service
Control and prioritization of resources

III. Ethical, Legal and Social Implications, Unintended Consequences

Opportunities exist for research on the social, economic, legal, and ethical impact of KDI. The following list prioritizes the relative importance of each research area.

The likelihood of gains to scientists from the development of KDI technology will be dramatically increased if the development of technology is accompanied by systematic case-based analyses of KDI experiments, as these experiments unfold in active scientific communities.
The KDI should adopt a "portfolio" approach where, as much as possible, key problems are addressed across directorates to compel research that builds links between existing communities.
KDI technologies are likely to transform the practice and organization of science. Research should address the ongoing adaptation of social structures within scientific communities, with particular attention to structural innovation that maximizes the benefits from use of KDI technologies.
The increasing scope and importance of community data systems merits research on data validation and authentication, particularly as use expands beyond the original communities that generated the data (i.e., "anonymous collaboration").
In order to assess the impact of KDI efforts, it is necessary to develop measures of "knowledge capital" in order to determine the effect of KDI on the use and creation of scientific information.
As Internet access to scientific instrumentation increases, the importance of resource allocation mechanisms increases. Specifically, research is needed to assess the consequences of alternative allocation strategies, i.e., rationing VS market.
The function of "information brokering," that is scientists helping each other locate needed resources and data, is a crucial component of scientific progress. Research should be conducted to explore the feasibility of new structures and tools, made possible through KDI, that will improve the availability and quality of information brokering.
The expanded availability of scientific data via KDI will create wider public access to research results. The consequences of this wider access are unknown, but the likelihood of unintended effects is high. Specifically, research should be conducted to determine the potential for the misuse or the misinterpretation of data available via KDI technologies.
The evolution of the Internet is confusing traditional notions about intellectual property. KDI is likely to accelerate this trend, particularly as multiple institutions collaborate to produce technology that addresses community needs. Research should be conducted to understand the relationship between the level of patent or copyright control and the dissemination of innovations.
The development of effective KDI tools may require funding mechanisms outside traditional NSF patterns. Specifically, the effort to produce reliable and easily used tools may require centers devoted to tool development. These centers should be viewed as opportunities for the NSF to make investments in high risk/high payoff technology (ala Mosaic).

last updated: June 13, 1997