A. HUMAN-INFORMATION IN GENERAL, PHILOSOPHICAL TERMS.

The irony in the "information age" is that "there is still no accepted integrating theory for dealing with data and information, let alone with knowledge, which is the focus of many so-called `knowledge-based systems'" (Fox 1989, 1).

Definition of human information (HI).
I perceive HI as a unity of three representations.
These representations were created in a long historical process together with the creation of humanity.
The first representation, is the "thought" in our minds, which represents the real world, a prerequisite for the existing of every animal system.
The second representation is the correspondence of sounds (words) in these thoughts.
The creation of this second representation was a revolutionary stage for mankind.
It was the main thing that differentiated humans from animals, and a prerequisite in the creation of human society.
Finally the third representation is the correspondence in the existing entity sounds-thoughts, and visual symbols (letters).
This third representation appeared many years after the second representation and was a second revolution in the evolution of humanity. It was the cause for the appearance of sciences as shown in Greek history.

HI has many forms.
The unity of sounds and symbols we call language.
In the world many languages exist that represent human thoughts.
So, HI exists in many forms.
But all these different forms have something in common.
All are representations of the same thing, the real world that exists independently of humans but which includes humans.

Structure of HI.
Furthermore, HI can be divided into smaller parts.
The indivisible elements which preserve its characteristics, I call concept.
The structure of every concept has a language element (which consists of a symbol and a sound component), and a thought, which we usually call meaning.
In this perspective, data, information, and knowledge are all conceptual systems (structures of concepts).
Its difference depends only in the degree of the complexity of their structure.
Of course, there is no clear criteria in this distinction.

Subjectivity of HI.
I must emphasize that the nature of HI depicts its main characteristic, its subjectivity.
It is a complex representation of the real world, but not reality itself.
So its validity (if it is true or false) depends on its comparison with reality, not just with other information.

B. THE COMPUTER SYSTEM.

The advent of CSs, machines to manipulate HI, came only when the evolution of technology made feasible to man to do another representation on the human information.
This new digital representation is at final analysis a combination of electric circuits, a representation of the symbols of HI.
In the early years only numbers and instructions were represented.
This is why the first functions of CSs were computations. This fact also gave its name to systems, computers.
But, through the years we are witness in the evolution of the complexity in this representation.
The next step was the representation of words which gave rise to the creation of data bases.
Furthermore, in the 1980's representations of relations between words gave rise to knowledge bases and the advent of the so called "expert systems" and other artificial intelligence products.

The support role of CSs.
But at least now and the next decade, subject to my paper, the CS is unable to deal with the main characteristic of HI, its subjectivity.
Only the human at the final analysis is the one who decides about its validity, because only he can compare it with the reality.
The system only in very limited ways can decide about the truth of HI, and this can be done only in an indirect way -- through comparisons with other existing conceptual systems.
In this perspective, the CSs have a support role, and this support role will continue for many years.

1. Future structure.

The evolution of CSs shows that the main characteristic of the structure will be in one word, network. Powerful workstations will have access to databases, libraries of software, expert systems, and powerful mainframes.

Components.
All the sciences do not have the same CSs needs (Info..., 12).
But existing problems demand more powerful hardware (Markoff 1990a, D1).
The history of CSs shows that the capabilities of CSs in terms of speed and memory capacity will continue, and will be another reason for more computer use in the sciences.
In relation to big systems, Parallel processing seems to be a solution in terms of speed for the short term future.
For example the Delta system will be installed next spring at the California Institute of Technology and is claimed to be the world's fastest computer (ibid, D1).
Also a report prepared for the Department of Energy and the President's office of Science and Technology Policy, estimates that the sales of parallel computers will exceed conventional ones at 1996 (Markoff 1990b, 34).
A significant role for more scientific uses of CSs will be the small systems.
In the November 1990 Comdex exposition the first generation of desktop 486 systems were displayed. "Scientists who need every bit of performance they can get are expected to be among the early users of desktop 486 systems" (Lewis 1990a, C5).
In terms of memory capacity, optical hard disks and diskettes with many times capacity than the present magnetic disks and diskettes will be widely used in ten years and will be another reason for further use of CSs in sciences which demand more storage capacity.
Improvements are also expected in networking hardware.
Fiber optics and wireless methods will play a role in this area, which is a major factor to improve the communication function of the systems.
Finally in ten years the research in optical computers is expected to lead to the first practical systems that will be a revolution in CSs hardware.

2. Future functions.

The purpose hear is to describe the current impediments in the functions, to show concrete trends to surpass them, and how the enhanced functions will give the possibility for further uses of CSs in sciences.

Communication.
Impediments:
Incompatibility is the word that expresses the substance of today's impediments (Info..., 21).
It is related with a lack of standards in network protocols and standards in the other functions.
These standards are necessary to systems in order to perform functions that other systems perform.
Another obstacle is the limited spread of existing networks.
Trends:
In relation to existing networks, a significant expansion is expected.
For example, the National Science Foundation has announced its intention to serve as a lead agency in the development of a "truly national research network" (Info..., 21).
In relation to standards, text formats for example which are a necessity to communicate text between scientists, it is expected to be resolved the next few years (Info..., 21).
Another example is the Britelite machine which demonstrated the previous month at the Comdex exposition that represents "the first major step toward [small] computers that can run software from a variety of operation systems" (Lewis 1990b, F1).
The enhanced communication function will be the main reason for wide spread use of CSs in sciences.
The report on the uses of information technology in research had emphasized this function and D.N. Langenberg the chair of the panel of the report "strongly believes" that the power of science, derives from communication (Info..., ix).

Computations.
Impediments:
The only obstacles in this function are hardware capabilities of the systems.
Trends:
The evolution in complexity of computations, numerical and symbolic (computations of functions rather than numbers), will continue.
An integration with the graphics function is also expected and this will enhance the interpretation of calculations.

Storage-retrieval.
Impediments:
Except the storage capacity the main obstacle is searches.
"Most information searches at present are incomplete, cumbersome, inefficient, expensive, and executable only by specialists" (Info..., 2).
Trends:
In ten years we will have a significant improvement in this function.
It is expected integration with expert systems and natural language processing techniques.
For example the Q&A database system of Symantec Corp. that was marketing in 1989 "can do searches in natural language" and works in desktop systems (Arora 1990, 74).
The enhanced capabilities of document and reference databases will be a very important reason for further use of CSs in sciences and an important step to improve the acquisition of knowledge.

Word processing.
Impediments:
This function in relation to other functions, is of the most powerless. "Spelling checkers typically look at words in isolation" (Smith 1990, 199).
All automatic abstracting just extract sentences and not create new sentences as a human (ibid, 248).
Trends:
Integration with natural languages processing techniques will improve this function.
For example systems like the CRITIQUE, a text processing system that detects syntax errors like number disagreement, wrong pronoun case, wrong verb form, and punctuation (Ravin 1990, 108) will be in wide spread by 2000.
Also input hand-drawn or camera images with scanners technology, so important in scientific texts, will improve this function.

Graphics.
Impediments:
Creation of standards and demand for powerful workstations are some of the existing obstacles.
Trends:
Improvements in this function are expected.
For example K.W. Brodlie predicts another trend in graphics. Visualization of the problem solving during the process of solving thus allowing the scientist both insight and control over the problem (Brodlie 1990, 200).
Also in relation to standards, the ISO began an effort since 1985 which continues to be improved on and to gain acceptability (Brodlie 1990, 189).
Another trend is the integration of graphics, sound, text, full-motion video (so called multimedia). In November 28, 1990, ten computer makers announced to offer multimedia PC's.
"The technology could be used to create educational or presentation material" (Fisher 1990, D6).
This will be another step to improve acquisition of knowledge and another technological cause for more computer use in the sciences.
Also, because computer graphics allows new possibilities in research, "computer `see' the unseen" (Kolata 1990, C1), this function is an example how CSs will be, from another point, a necessity for the sciences.

Advisory function.
Impediments:
The "very limited domain" that can be represented in CSs (Brooks 1987, 367) is the most significant obstacle.
Trends:
Improvements in knowledge representation techniques are surely expected.
According to an optimistic opinion "during the next ten years, expert systems can be expected to supplement and transform the functions of professionals in every knowledge industry" (Davidson et al 1990, 80).
Eventhough I can not predict the exact strength of this function, it will be another important reason for more computer uses in sciences.

Natural language processing.
Impediments:
Until now by this function scientists have only managed to represent the symbol portion of HI and the rules used to create symbol (word) structures, or the syntax of HI.
However, this is the limit of the function. In reality ambiguities and conflicts in the symbol subsystem are resolved by the sound subsystem (eg with different accent for the same word), or by meaning (same word, different meanings).
Not mentioning the most important of all, conflicts in HI, which only the reality resolves, either the comparison with it or the practice ("really works") as we say.
Trends:
An example of future trends is the START system published in the last publication of the AI laboratory at the Massachusetts Institute of Technology.
The "objective is to buillp a practical system that reads sentences, indexes the facts they convey, and retrieves those facts when asked" (Winston et al 1990, 134).
The validity of the system was tested many times and one example was from members of the press in the Jet Propulsion Laboratory who asked questions about the Voyager's Neptune encounter in 1989.
"The experiment was a success" (Katz 1990, 158).
Another trend that will improve this function is the Machine translation techniques (Smith 1990, 273).
An example here is the EUROTRA project of the European Communities which began in 1982.
In February 1987 the first small-scale translation system was completed, and worked for translation between German, English, and Danish (Maegaard 1989, 47).
In general, improvements in this function will obviously be a cause for further use of CSs in sciences.

Perception.
Impediments:
A major obstacle for practical applications of these systems is their speed in both processing and accessing data.
Perceptual systems involve manipulating fantastically large quantities of data and most potential areas of application of these systems demand that their tasks must be done in a timely fashion.
Trends:
Based on current levels of activity, the situation is expected to improve dramatically over the next ten years.
Raw computer processing power has been increasing and will continue to do so at unprecedented rates.
This has been accompanied by new types of data and knowledge representation structures together with corresponding processing algorithms that are designed to exploit the new hardware architectures.
For example, this field has been one of the major consumers of new types of parallel processing hardware and algorithmic techniques.
Particularly important applications for sciences will be in the area of visual based man-machine communication.
Visual based communication requires that a machine must be able to understand images as well as be able to generate them.
The problem of how to generate images is being solved by computer graphics researchers.
The problem of how to interpret input images is being be solved by the application of machine vision techniques.
So, the enhanced perception function will be another reason for more computer use in sciences.

3. Future user-system interface

In general, it is certain that in ten years a significant improvement in user-system interface will be done.
This will result in an improvement on all CSs functions and further computer uses in sciences.

Methods of operation.
Impediments:
"User-unfredliness" is the word we use to express current obstacles.
The cause for this is the quantity and complexity of conventions the user needs to know to operate the system.
"Powerful query languages are rather difficult to learn for nonprogrammers" (Thorpe et al 1989, 152).
Trends:
The future in this area are natural languages interfaces that "are becoming reasonable option and not a source of endless frustration" (Katz 1990, 158).
Especially in databases the assistant role of expert systems will play a major role to make the systems more easy to be operated.

Ways to interface.
Impediments:
The major obstacle is the "keyboard bottleneck" (Sanders 1988, 216). Also voice input is in its infancy (Smith 1990, 15).
Trends:
The new optical character recognition and scanners technology will reach a mature stage in ten years.
This will be a very important cause for further scientific uses of CSs because scientists mainly deal with texts.
The use of optical character recognition is for example forty times faster than a typist (Smith 1990, 7).
Understanding handwriting is another technology that will play a role.
This year, eg Tandy corporation introduced its 4.5 pound pen-based system that accept handwriting commands (Shapiro 1990, D5).
E.G. Glass, a consultant, estimates that sales of pen-based systems could reach $3 billion by the year 2000 (Shapiro 1990, D5).
Finally improvements in voice recognition are expected to result in a more easy and quick way to interface with the system.