1. Pressing
problems in knowledge handling policy
At a time when
we are exposed to:
The complexity
of the knowledge handling system is such that conceptual ambiguity is
the rule rather than the exception. At the same time we are running short of the paper
which permits us the luxury of our incredibly ineffective, document-oriented system.
Furthermore, and more serious, the cumbersome nature of the knowledge handling
system effectively prevents the maintenance of "thinking momentum" (2) on any issue,
whether for an individual or in group interaction between researchers. Such disruption
of innovation is increasingly intolerable as well as dangerous because of our dependence
upon collective innovative and rapid responses to the many problem of society. The
scholar's relaxed acceptance of extended delays (deriving from the monastic tradition
and the priorities of the gentlemen-of-leisure who fathered many of the sciences) can no
longer set the standard for knowledge handling [1].
The US National
Science Foundation has invested heavily over the past decade in
abstracting and indexing services for a range of disciplines. It recently summarized the
current state of affairs as follows:
"The world's store of scientific and technical literature continues its
exponential
growth, with a corresponding diversification of the uses to which it can be put. We
may be nearing the limits of what can be accomplished by printing, mailing, storing,
and retrieving pieces of paper." (3)
This is not
the place to detail the evidence in support of this view. A significant practical
example, however, is the case of the United Nations. A former President of the General
Assembly remarked that "the United Nations is drowning in its own words and
suffocating in its own documentation" (4). The UN Joint Inspection Unit notes that "the
point of saturation has now been reached and indeed overstepped and that the law of
diminishing returns is taking over" (5). Their solution implemented, however, is "to set
once and for all, and strictly enforce, a reasonable but drastically reduced ceiling to the
volume of documentation its various bodies call for and its services produce" (5). It can
be argued that such a response to the problem is incredibly short-sighted in view of
mankind's need for new knowledge and the right of all to participate in the generation of
that knowledge and to receive the associated information. To reduce severely the means
of storing and disseminating such knowledge within the world's key organizational
system, without seeking a more appropriate complementary medium, can only be
counter-productive and unsatisfactory.
If some limit
is being reached then the National Science Foundation, continuing the
above quotation, considers that:
"effective communication will necessarily come to depend upon electronic
means of
handling information. In any case, for significant improvements in the accessibility
and usefulness of the information handled we must look beyond paper-based
communications to a computer- sensible literature, stored in central facilities for
instantaneous presentation at remote terminals anywhere. To create such a literature
through the conversion of printed literature would be slow, inefficient, and
formidably expensive. For this reason, a goal for publication is to capture new
literature in computer-sensible form at its source." (3)
The same document
identifies other interrelated goals:
As is noted
below, the NSF is currently funding field experiments amongst groups of
scientists. As has been noted elsewhere (6), it is difficult to convey the nature of the
communication process in this new computer- supported, paper-less environment. "Most
of our intuitions about face- to-face interaction simply do not apply to this new and
unusual form of communication... it is not surprising that computer conferencing might
actually establish an altered state of communication in which the realities of fact-to-face
communication are distorted and entirely new patterns of interaction emerge" (6). Some
impression of the significance of existing applications may be gained from the following
section. A major investment in creating and experimenting with such environments has
been made over the past decade through the ARPANET at the Center for Augmenting
Human Intellect, Stanford Research Institute (7, 8, 9).
2. Software
and Hardware
It is much to
be regretted that those who have an understanding of the existing, and
increasingly available, computer hardware and peripheral equipment are rarely able to
envisage innovative uses for that equipment outside certain specialized sectors of
engineering and fundamental defence research.
Consequently,
when such equipment is used elsewhere, the applications do not constitute
breakthroughs in the ability to respond flexibly to relationship complexity and inter-
sectoral contact, but only a greater ability to handle the increasing quantities of data
within a predefined sector. This lack is matched for those in those sectors, which could
benefit immediately from a wide variety of innovative though relatively simple
applications, but who are unaware of the possibilities.
It is impossible
to explore adequately in this context the significance of these devices and
applications for classification-related questions. Some possibilities can however be
indicated (7).
1. Computer
graphics devices (CRT displays). These are now familiar to many, if only at
airline reservation desks. What is much less well-known is the form of this device which
can handle not only lines of text but can also display highly complex relationship
networks (such as arrow diagrams) whether in two or three dimensions with many
possibilities for assisting the user in the exploration, comprehension and re-
representation of such (concept) networks for comprehension by others interested in
alternative or simplified displays. Very complex domains may be represented on displays
of several hundred colours [2].
2. Graph plotters.
Complex relationship charts of up to several square meters in size
may be drawn, with several colours, under computer control on the basis of data
selected by the user (possibly after viewing a display of successive parts of it with the
previous device). The significance of a knowledge user or institution being able to obtain
and represent the structure of a knowledge domain in this way remains to be appreciated
(7).
3. Computer-assisted
structure elucidation. Programs are now in use for the interactive
exploration by chemists of possible molecular structures in the light of inferred structural
fragments and various constraints (12, 13). The approach bears many similarities to an
application which should be available to users wishing to explore concept structures (e.g.
in association with the proposal of the Committee on Conceptual and Terminological
Analysis, and Unesco's current Interconcept program (10, 14).
4. Computer-conferencing
The US National Science Foundation is now investigating the
consequences of providing computer terminals to individuals who are members of
geographically dispersed "invisible colleges" (3). The scholars so linked, whether in one
or more countries, can exchange, store and comment on information according to an
evolving agenda. Such applications may include those mentioned above. To date
however the structures of the "agendas" governing the relationship between the items
chosen for a particular computer conference are of the simple hierarchical variety.
However, the computer does facilitate linking and sequencing keywords in the series of
interventions constituting the transcript of a conference so that such associative
networks may be explored in a manner somewhat similar to citation analysis. The
significance of being able to use such an environment to facilitate interaction in relation
to a complex evolving network of concepts, and to use the environment to explore and
experimentally re- structure collectively such a network, remains to be
appreciated—particularly with regard to interdisciplinary and intersectoral
communication [3].
3. Knowledge
representation
The previous
section indicates a breakthrough in terms of hardware in support of a new
knowledge- handling environment and the NSF initiative indicates that this is being very
seriously explored. What is still not appreciated apparently, is the significance in such an
environment of the decomposition of the "texts" of a particular author into sentences or
even words. The computer permits this (by the very nature of its operation) and
facilitates any recombination of his statements into new configurations (perhaps blended
with those of his colleagues). This is important for concept analysis (14, 15, 17).
A stage is therefore
reached in which a given text is treated by the computer as a
network of key words embedded in a field of explanatory comments. The structure of the
network bears an iconic relationship to the knowledge it represents. Knowledge
innovation is more and more closely represented by the changes to the structure and
content of that network. The explanatory and introductory comment, which constitutes
the great bulk of any text is only of secondary significance and can be stripped away,
given a much lower handling priority, or reprocessed into a more compact and
comprehensible form by communication and education specialists. Soergel, in discussing
the possibility of an automated encyclopedia, discusses this point (18) with a quotation
from Bohnert et al (19):
"An increase in accessibility without a corresponding increase in human
assimilation
rate will be self- defeating... Often one needs to know only a central idea, result,
theorem, or the methods employed, with bibliographical information for later
reference, but finds that a short course in unneeded detail is required to get to it."
Little attention
has been given to this problem of assimilation, other than a heightened
emphasis on "speed reading". It would seem that much is to be gained by looking at the
ability of the graphics devices discussed above to provide structured diagrams and
displays in which a deliberate attempt is made to use inter face programs (possibly
selected according to the presentation preference of the user) to provide a high degree of
iconicity. Relationship structures as displayed should bear a strong relationship to the
relationship between the knowledge structure which has to be absorbed as a gestalt for
learning to take place (20).
In parallel
columns below, an attempt is made to clarify the distinction between a
hypothetical knowledge-oriented system, now technically feasible, and the current
approach. The intention is not to imply that the former should replace the latter but rather to
show that the former offers various means of avoiding some of the key problems faced by the
latter - the two are however complementary. The distinction is basically between integration or
fragmentation in the handling of information.
|
PRESENT:
Document/Information System
|
FUTURE:
Knowledge-representation System
|
|
Index tends to
be based on simple hierarchy or
alphabetic listing of subject, author and title, which
can be handled on catalogue cards. Budgetary
constraints usually prevent widespread
introduction of sophisticated classification and
cross- referencing techniques.
|
"Index
" constitutes a complex network giving
a representation of entities and relationships
and the dynamics of any points under debate.
This complexity can only be handled by
multidimensional computer techniques. Cross-
references are necessarily inserted by the
author to define the location of his innovation.
Others may be inserted automatically,
optionally, or experimentally by computer.
|
|
Users want rapid
access to documents; the index
is a temporary inconvenience to gain access to a
document.
|
Users want rapid
access to the "network
index" which represents the needed items of
knowledge and their relationships; documents
are a temporary inconvenience only used if it is
necessary to re-examine data and detailed
arguments justifying the entities and
relationships incorporated (Document access is
a secondary problem for which a
documentation system may be used.)
|
|
Access to knowledge
via documents means
multiple reproduction and transfer of documents to
a variety of libraries where they may or may not
be used.
|
Access to knowledge
is direct and does not
require reproduction and transfer of
documents. (Only one copy of the document
justifying the amendment need exist on
microfiche so that copies need only be
prepared when the data and arguments must
be re- examined in detail)
|
|
Out-of-date, rejected,
low quality, false, old
documents are retained in the system and indexed
with no index indication of their status.
|
Out-of-date,
rejected, false, etc. entities or
relationships may be eliminated from the
system by listing them on paper, microfilm, or
other "documents" with the bibliographical
source from which they were obtained (ie they
are available if required but do not clog the
system).
|
|
Only the knowledge
held in the documents
physically available at that location is accessible.
The index frequently only indicates the documents
held in the documentation centre in question.
|
All knowledge
is on-line, although the
supporting documents may not be physically
accessible without delay.
|
|
Research is conducted
primarily using documents,
notes and file cards as a stimulus to creativity.
|
Research is
conducted primarily using the
knowledge-representation structure (i e. the
graphical representation) as a stimulus to
creativity. Private and tentative amendments
can be made experimentally, shared
electronically with selected colleagues, and
then destroyed, stored or released
electronically to a wider audience. The authors
"notes and file cards" can be effectively
integrated into the system to facilitate his
thinking processes.
|
|
Different styles
of documents are produced on the
same topic for research, education, public
information and propaganda, program
management, policy making, etc., purposes. The
same material is repeated, with some extensions
and some omissions, for each audience. This leads
to a "spastic" or "aphasic" response to new
situations, by different portions of society due to
delays in production of the documents for different
audiences and to significant variations in the
importance given by the authors to different items
of information.
|
The entities
and relationships entered on the
basis of research insights are also used for
other purposes. Instead of producing different
documents and reprocessing the insights,
different identified "filters" are used in
presenting or displaying the entities and
relationships to different audiences. In this
way, each new research insight is immediately
incorporated into each other form of
knowledge-representation; each portion of
society works from the same data base.
(Problems registered by non-research bodies
are immediately evident as a challenge to
research.) In this way if an element of
knowledge represented cannot be understood,
the user merely calls for a new method of
representation (of the same knowledge)
possibly using isomorphs (or even analogies)
from a domain with which he is familiar. (At
any point he can move into a programmed
learning mode and be instructed with simpler
representations or work from an area of
knowledge with which he is familiar.)
|
|
Each new document
must carry a lot of verbal
packaging to explain and define the context within
which innovative elements are introduced. Such
contextual material is repeated by each author
concerned with that domain of knowledge. There
is no guarantee that the rephrasing of earlier
arguments (necessary for status and copyright
reasons) will constitute an improvement facilitating
greater comprehension (rather than inhibiting it).
|
The author need
only enter the specific entities
or relationships which constitute his
innovation. (Since the academic's status is
bound up with his specific modifications to the
knowledge structure and not the verbalizations
held in a document, the problem of adequate
verbalization may be handled separately.
Hopefully a limited number of skilled verbal
presentations, from a minimum number of
different perspectives and literary styles, could
be constantly updated by professional writers
using the best verbal arguments by any
appropriate academic or communicator.)
|
|
Articles retain
permanently their total length and
degree of encumbrance to the document system.
|
Articles may
retain their full length only for a
period of days before being shortened or
stripped (by computer) of explanatory matter
and represented as a network of concepts - or
simply stored on microfilm or erased.
|
|
Alternative concepts
or contradictory evidence
published elsewhere can be conveniently ignored
in a document or textbook—particularly where the
counter argument comes from another discipline
(or a school of thought publishing in a different
language). The risk of explicit published criticism is
low in many fields, therefore the degree of support
for (or criticism of) any particular element in a
document remains unclear.
|
Alternative
concepts, relationships or
contradicting evidence are immediately forced
on one's attention - even in the case of
relationships linking to other disciplines. The
degree of support for (or criticism of) any
particular element is clearly evident. Members
of qualified professions may "vote" on
particular amendments to the knowledge
structure which is their concern.
|
|
Interdisciplinary
links are ignored if the author has
no interest in them. As a result there is no built-in
process within the documentation system which
encourages integrative studies to counter-balance
the further fragmentation of knowledge. Integrative
studies have low status, being equated with
educational texts, general reviews and journalism.
|
Interdisciplinary
links are already held in
position whether the author wants to ignore
them or not. Integration of isolated items of
knowledge into higher orders of synthesis is
facilitated and may be undertaken
experimentally, selectively and largely by
computer program {searches may be made for
various degrees of isomorphism between
concept structures in different domains).
Integrative innovations acquire a high status
as a means of comprehending wide domains of
knowledge and controlling the associated
information.
|
|
The documentation
system does not permit
panoramic summary of any permanent
representation of knowledge in a particular
domain.
Each verbal summary
extant at a particular
moment is under criticism and subject to reserves
from different schools of thought within the
discipline or in other disciplines. In this important
respect a document arising from a single group of
authors can never contain the totality of views in a
domain of knowledge. Only the non-concretized
interaction between a succession of documents
approximates to it. These invisible qualifiers on any
document are a feature of the "collective mentality"
of the members of the discipline. The knowledge
of the discipline at any moment is very much in
(and between) the minds of its members rather
than on paper or in a row of books.
The forum of academic
debate is concretized in a
scattering of journal articles and other documents.
There is little interaction between the journals but
the debate is somewhat summarized in the various
collections of abstracts in which the contents index
gives some indication of the interventions on
related topics.
|
Each entity
link and qualification is indicated
in the knowledge-representation system. In
effect one "layer" of the "collective mentality"
of a discipline is rendered visible. Each
modification to knowledge in the domain can
be entered on an hour-by-hour basis.
The knowledge-representation
system
constitutes a "thinking forum" in which the
juxtaposition of relevant ideas from all sources
is maximized. The researcher can expose
himself to a pattern of theoretical formulations
in the process of being continually improved,
and to which he can contribute. Concepts and
relationships can be "registered " by postcard,
but more dynamic possibilities are increasingly
available. A dozen or more specialists in a
particular field {the "invisible college" for that
topic) can contribute simultaneously to work
on ideas being written on one "mental note
pad " via electronic dialogue support systems
which help them to respond to each other's
ideas {even if they are a continent apart) with
a rapidity that allows each of them to maintain
thinking momentum.
|
|
Thinking momentum
is constantly interrupted when
access to new documents is required. (Long
delays, 2 - 3 months, are normal; 50 months or
more from initiation of research to appearance in
abstracts.)
|
Thinking momentum
is maintained since the
essence of any new domain of knowledge is
always accessible - all the links and entities are
there {delays are measured in seconds for data
links).
{This mode of
operation should be compared
with some discussions between academics
interested in the same topic in which progress
is frustrated because if someone thinks of a
good idea he wants to "publish " it (to gain
credit) before contributing to the thinking
momentum of his colleagues - this may mean a
delay of months)
|
|
Author has "published"
when document is in
circulation and "available"; index entries are of little
significance to the author. Texts must be at least
several pages in length before they are considered
"documents" worthy of registration in an
information system. The documentation system is
embarrassed when faced with obtaining
"ephemeral" or "phantom" material which has not
been made commercially available through the
normal publishing channels.
|
Author has "published"
when the appropriate
knowledge structure in the "index " has been
modified; incorporation in "index " (through a
terminal) is of highest priority for the author.
Acceptable amendments to the knowledge
structure can be as little as a single line of text
in length, or simply the indication of a
relationship between existing, but hitherto
unrelated, items of knowledge. Even in the
course of rapid change to the knowledge
structure the paternity of each emerging
formulation is identified and registered (if the
author so desires).
|
|
Author's status,
credibility, pride and interest are
primarily associated with visible documents on
library shelves and only secondarily with the
research community's collective judgment on their
value. The documentation problem is aggravated
by the "publish or perish" code which governs
much of academic life. Unless an academic
produces a document he is "invisible" and loses
status.
|
Author's status,
credibility, pride and interest
are associated with the visible entities and
links in the graphic representation accessible
to all. By switching emphasis to the specific
entities and relationships which the academic
has formulated, successfully confirmed or
criticized - his status is determined by the
bonds and entities with which he is associated.
Each of his contributions is "visible" until it is
superseded. They are not subject to the
vagaries of document distribution patterns and
the journal referee system.
|
|
The key figures
in a discipline and the relationships
between their spheres of influence are unclear.
|
The "luminaries"
in a particular discipline are
all visible together with the relationships
between their spheres of influence.
|
|
The direction of
research is governed in part by
shifting fashions of credibility, status and politically
determined funding (e.g. "environment",
"resources", "population") which obscure the basic
knowledge structure. This is only partly evident in
print but is controlled by an ongoing informal
dialogue centred upon the elders of the discipline
who legitimate consideration of particular entities
and relationships.
|
It is quite
evident which issues are currently
under debate and the manner in which the
demise of a set of entities and relationships will
weaken the status of a whole set of dependent
elements. Current fashions would not obscure
the basic knowledge structure. Ideally the
system would also act as a continually updated
voting board for each element, providing an
opportunity for members of the profession to
indicate their approval, whilst at the same time
providing an appropriate focus for counter-
arguments and alternatives.
|
|
The world's publishing
and purchasing capacity,
and the consequent necessity for the journal
referee system and increasing costs, limits
arbitrarily (and in many cases inequitably) the
number and variety of viewpoints which can be
expressed on any subject. The nature of the
referee system leads to an inhibition of innovation.
|
The reduction
of the volume of text required to
"carry" any conceptual innovation, and the
integration of the referee and editorial system
at the computer level permits a greater number
and variety of viewpoints and more subtle and
equitable mechanisms for the expression of
peer-group support or criticism.
|
|
Considerable intellectual,
administrative and
technical investments are made in achieving a
unified standard of classification and description
which determine the structural specifications of
information systems. Relationships between
different standardized schemes of this type are not
facilitated nor are experiments with amendments
or alternatives to any particular scheme.
|
The information
is handled in a very flexible
format. A choice may be made at any time
between a variety of classification schemes.
Some of these may be universal schemes,
others may be specialized, and others may be
experimentally employed by the user.
Considerable use is made of computer power
to switch between classification schemes and
to restructure them {tentatively) in the light of
new insights and relationship coding schemes.
|
1. Constraints
