|
The word "informatics" has a number of inter-connected meanings:
At Edinburgh, we formed the School of Informatics from three previous departments: Computer Science, Artificial Intelligence and Cognitive Science. We chose the name "Informatics", interpreted in the broad sense (no. 4 in the above list) to signal the breadth of our interests. These interests include those of the previous constituent subject areas:
but also includes new interests, to be found at the interdisciplinary boundaries both between these previous areas and between them and other disciplines, such as biology, psychology, philosophy, sociology, physics, electronics, etc.
The time is ripe for this broader interpretation of informatics because old boundaries are breaking down. There is a need for a unifying science.
Most sciences are associated with Big Questions, that define their mission. In physics, for instance, the Big Questions include: "what is the nature of matter and energy?" and "what is the structure of the cosmos?". In biology, the Big Question is "what is the nature of life?". In informatics, we can identify at least three Big Questions:
(1) refers to the scientific purpose of informatics, (2) to its role in modelling cognition and (3) to its engineering goals. An advance in informatics might contribute to more than one of these questions. For instance, a new form of computation might suggest both new kinds of cognitive models and new kinds of smart products.
Informatics can also be seen as encompassing a space of computational techniques, which it is the job of informatics to explore. The word "techniques" is used here in a very broad sense to include the whole range of artifacts that informatics develops, combines and uses. These include, for instance, the following:
Exploring the space of techniques includes both the discovery and invention of new techniques, and the discovery of their properties and interrelationships. For instance, we can discover: what kind of tasks each technique is good for; under what circumstances one technique is better than another; which techniques complement each other and can be combined into a system to tackle a task. In the lecture on "The Need for Hypotheses in Informatics Research" we will explore the many dimensions along which techniques can be compared. We will, thereby, see that the space of techniques is also multi-dimensional, so the relationships between techniques can be complex. Technique A might be superior to technique B along one dimension, but inferior along another, which creates a complex of tradeoffs between the techniques.
Informatics research is driven by a collection of generic questions, which link research areas to applications. Each application requires answers to these questions and each research area provides some answers. It is in the nature of informatics that these answers are only ever partial; they improve our understanding and our capabilities, but leave the generic question open for attack by future researchers, perhaps from other research areas. Some examples of such generic questions are:
Each application will require some answer to many of these generic questions. We typically require systems that are easy to use and maintain, dependable in use, scale to 'industrial strength' problems, cheap to build and run, etc.
Computing researchers frequently argue as to whether informatics (or computer science) is science of engineering. The short answer is both --- and much more. A more detailed discussion of this issue will be given in my lecture on Informatics: The Methodological Magpie.
Meanwhile, we can remark that informatics is a science in at least two respects. Firstly, in its use of the scientific method of developing theories and subjecting them to experimental evaluation. Secondly, in the use of computational models of natural phenomena, in which computational techniques are used to construct theories and hypotheses for natural sciences, such as physics, astronomy, biology, psychology, etc. One of the most established exemplars of this second kind of science, is cognitive science, which builds computational models of cognitive processes. A newer example is systems biology, in which computational models are built, for instance, of protein construction and interaction in the cell.
Informatics can also be seen as engineering, in that it builds complex artifacts, ICT systems, out of computer programs coupled with both generic computers and more specialised hardware, such as robots or embedded devices in electronic and mechanical products. Here the techniques of software engineering show us how to build usable, dependable, scalable, maintainable and affordable ICT systems in a timely manner. Software engineering draws on multiple methodologies, including mathematical methodologies of formal methods and sociological methodologies for organising effective and efficient programming teams.
Of course, these are merely different views on what is often a complex and interrelated process of informatics research. Theory and experiment give us new techniques and a better understanding of their properties and relationships. This suggests new applications both to ICT systems and to computational modelling, and improves our ability to build better such applications. Experience with both building and using these applications feeds back into the science by raising new questions, suggesting new approaches and a deepening our understanding of the limitations of and tradeoffs between the various techniques.
One of the ways that informatics advances is by using the results of earlier research as a platform on which to build later research. A research platform could be a programming language, whose basic operations encapsulate common patterns of programming in earlier, less sophisticated languages. More commonly, it will be a generic application, i.e., one that can be used as a component in a diverse array of more specialised applications. Examples include: reasoning engines, which combine facts and rules to infer consequences; web browsers, that can be used as a common interface to different web services; machine learning systems, that can be used as the core of a data-mining system. Java is particularly rich in the many packages and components that are available.
A research platform is an example of a virtual machine that provides a new layer of functionality on which new systems may be built. This improves productivity because the user does not have to reimplement this functionality within their application. Users may use the time saved to be more ambitious in the scope of their application. It may also improve usability, dependability and efficiency, since it is worth investing a lot of effort in optimising these virtual machines so that may be recycled in many applications. By limiting the functionality provided to users, they may be protected from introducing certain kinds of bug into their applications. Maintainability may be improved because systems are composed of independent layers of virtual machines, whose implementation can be revised without, in principle, affecting the other layers. Of course, upper layers will be affected if functionality they relied on is removed or altered.
The impact of computers on intellectual life goes beyond email, web surfing and word processing. Computational ideas are beginning to influence research thinking in many branches of both the sciences and the humanities.
Many scientific questions can only be addressed by collecting and analysing vast amounts of data. The resulting data collections are often so large that they are prohibitively expensive to send to the processor; the processor must come to them. They are often also distributed across multiple sites, meaning that Grid-like organisation of data sources and high-performance processors are required to combine and analyse the data, extract answers and present them in a digestible form. Many sciences now present research questions in this form. They include particle physics, astronomy, genetics and protoeomics, and the earth sciences.
Computational metaphors are also infiltrating the hypotheses and theories of other sciences. Systems biology is probably one of the best known examples of such computational infiltration, particularly the use of computational procedures to represent the interaction of biological processes. Perhaps the longest standing computational metaphor is that of the brain as a computer and the mind as the program running on it.
We have presented informatics as the study of the structure, behaviour, and interactions of natural and engineered computational systems. Informatics is both science and engineering, whose results and experiences feed in both directions in a virtuous cycle. We explore a diverse, multi-dimensional space of techniques, their properties, interrelationships and applications. This exploration is driven by a set of enduring generic questions, to which the answers are incrementally and partially revealed. One indicator of progress is the construction of research platforms, that provide encapsulated functionality, make efficient, dependable, maintainable and usable implementations widely available, and increase productivity and the ambition of the field. The techniques of informatics are infiltrating many other subject areas, influencing the kinds of questions they ask and the answers they accept. Computational metaphors are being incorporated into the hypotheses and theories of many other fields.
I always welcome feedback on these notes, especially if you have detected missing material.
Alan Bundy|
Informatics Forum, 10 Crichton Street, Edinburgh, EH8 9AB, Scotland, UK
Tel: +44 131 651 5661, Fax: +44 131 651 1426, E-mail: school-office@inf.ed.ac.uk Please contact our webadmin with any comments or corrections. Logging and Cookies Unless explicitly stated otherwise, all material is copyright © The University of Edinburgh |
