To an increasing extent, scientific progress is being driven by a quest for automation of decision-making and control. Soft computing (SC) and computing with words (CW) are important additions to the armamentarium of systems analysis, providing basic concepts and techniques for the conception, design, and utilization of systems that have a high degree of machine intelligence and require minimal human intervention in an environment of uncertainty, imprecision, and partial truth.
What is soft computing? Basically, soft computing may be viewed in two related perspectives. In one view, soft computing--in contrast to hard computing--is aimed at an accommodation with the pervasive imprecision of the real world, exploiting the tolerance for imprecision, uncertainty, and partial truth to achieve tractability, robustness, low solution cost, and better rapport with reality. In another view, soft computing is a coalition or consortium of methodologies that share this objective. At present, the principal members of the coalition are: fuzzy logic (FL); neurocomputing (NC); evolutionary computing (EC); probabilistic computing (PC); chaotic computing (CC); and machine learning (ML). What is important about these methodologies is that, for the most part, they are complementary rather than competitive. A central tenet of soft compuring is that, in general, better performance can be achieved by using the constituent methodologies in combination rather than in a stand-alone mode. A prominent example is that of neurofuzzy systems. Other combinations, e.g., neurogenetic, neurochaotic, and neuro-fuzzy-genetic systems are beginning to grow in visibility and importance.
What is computing with words? Basically, computing with words adds to conventional modes of computing the capability to compute with interpreted words and propositions drawn from a natural language. An example is: suppose that I know that Robert returns from work at about 6:00 p.m. What is the probability that Robert is home at 6:30 p.m.? What is the earliest time at which the probability that Robert is home is high? Another example is: assume that a box contains about 20 balls, most of which are large and a few are small. What is the probability that a ball drawn at random is neither large nor small?
A basic component of computing with words is what may be called Precisiated Natural Language (PNL). Basically, PNL is a subset of a natural language, NL, which consists of propositions that are precisiable through translation into what is called the Generalized Constraint Language (GCL). A generalized constraint is an expression of the form X isr R, where X is the constrained variable, R is the constraining relation and r is an indexing variable whose values define the ways in which R constrains X. Among the principal types of constraints are: possibilistic (r = blank); veristic (r = v); probabilistic (r = p); random set (r = rs); fuzzy graph (r = fg); Pawlak set (r = ps); and usuality r = u). Generalized constraints may be combined, modified, or qualified, e.g., (X is small) is very unlikely.
One of the important functions of PNL is that of serving as a definition language. In particular, PNL makes it possible to define such concepts as smoothness and usual value in a form that lends them to computation. In addition, PNL provides a basis for re-refinition of the basic concepts of optimality, stability, independence, etc. Conventional, crisp definitions of these concepts may lead to counterintuitive conclusions.
An important derivative of computing with words is the computational theory of perceptions (CTTP). The point of departure in this theory is the assumption that perceptions are described in a natural language. With this assumption, the techniques of computing with words may be used to compute with perceptions. This capability plays an important role in what may be called perception-based decision analysis (PDA). In PDA, decision-relevant information is assumed to be a mixture of measurements and perceptions. The role model for PDA is the human mind.
In combination, soft computing and computing with words add to systems analysis and decision analysis a collection of new and important tools. These tools open the door to a radical enlargement of the role of natural languages in systems analysis, decision, and control.
Lotfi A. Zadeh is Professor in the Graduate School and director, Berkeley
initiative in Soft Computing (BISC), Computer Science Division and the
Electronics Research Laboratory, Department of EECs, Univeristy of California,
Berkeley, CA 94720-1776; Telephone: 510-642-4959; Fax: 510-642-1712;E-Mail:
zadeh@cs.berkeley.edu. Research supported in part by ONR Contract N00014-99-C-0298,
NASAContract NCC2-1006, NASA Grant NAC2-117, ONR Grant N00014-96-1-0556,
ONR Grant FDN0014991035, ARO Grant DAAH 04-961-0341 and the BISC Program
of UC Berkeley