During the past few years there has been a rapid advance in the technology of time-sharing systems and software to permit quick access to large files of structured data. This has led to a growing interest in communicating with computer files directly in a natural language such as English. The natural language systems described in the literature are largely small-scale research vehicles dealing with small data bases of restricted subject scope. Giuliano (1965), among others, has questioned the generalization of these systems to wider universes of discourse. Developments in this area have been reviewed by Simmons (1966), and by Bobrow, Fraser and Quillan (1967).

Work performed under the auspices of the U. S. Atomic Energy Commission. To view the rest of this content please follow the download PDF link above.

Attempts to write'intelligent' computer programs have commonly involved the choice for attack of some particular aspect of intelligent behaviour, together with the choice of some relevant task, or range of tasks, which the program must perform. The emphasis is sometimes on the generality of the program's ability, sometimes on the importance of the particular task which it can perform. Well-known examples of such programs are Newell, Shaw, and Simon's General Problem Solver (1959; see also Ernst and Newell, 1967), which is applicable to a wide range of simple problems, Samuel's checker (draughts) playing program (1959, 1967), and the program written by Evans (1964), which solves geometric analogy problems. However, there is another approach to the goal of machine intelligence which stresses the relationship of an organism to its environment and which sets out from the start to understand what is involved in this relationship. Long ago Grey Walter (1953) experimented with mechanical'tortoises' which could range over the floor in a lifelike manner. Toda (1962), in a whimsical and illuminating paper, has discussed the problems facing an automaton in a simple artificial environment. Friedman (1967), a psychologist, has described a computer simulation of instinctive behaviour involving an automaton equipped with sensory and motor systems.

The generalized resolution principle is a single inference principle which provides, by itself, a complete formulation of the quantifier-free first-order predicate calculus with equality. It is a natural generalization of the various versions and extensions of the resolution principle, each of which it includes as special cases; but in addition it supplies all of the inferential machinery which is needed in order to be able to treat the intended interpretation of the equality symbol as'built in', and obviates the need to include special axioms of equality in the formulation of every theorem-proving problem which makes use of that notion. The completeness theory of the generalized resolution principle exploits the very intuitive and natural idea of attempting to construct counterexamples to the theorems for which proofs are wanted, and makes this the central concept. It is shown how a proof of a theorem is generated automatically by the breakdown of a sustained attempt to construct a counterexample for it. The kind of proof one gets depends entirely on the way in which the attempt to construct a counterexample is organized, and the theory places virtually no restrictions on how this shall be done. Consequently there is a very wide freedom in the form which proofs may take: the individual inferences in a proof may be very'small' or very'large' (in a scale of measurement which, roughly speaking, weighs the amount of computing necessary to check that the inference is correct). It is even correct to infer the truth of a true proposition in just one step, but, presumably, to offer such a proof to someone who wishes to be convinced of the proposition's truth would not be helpful epistemologically. His conviction would come, not from contemplating the proof itself, but rather from examining the computation which shows the correctness of its single inference step.

Nature, 218, 19-22.

Simmons has presented a survey of some fifteen experimental question-answering and related systems which have been constructed since 1959. These systems take input questions in natural English (subject to varying constraints) and attempt to answer the questions on the basis of a body of information, called the data base, which is stored inside the computer. This process can be conceptually divided into three phases---syntatic analysis, semantic analysis, and retrieval, as illustrated schematically in Figure 1. The first phase consists of parsing the input sentence into a structure which explicitly represents the grammatical relationships among the words of the sentence. The remaining phase consists of procedures for either retrieving the answer directly from the data base, or else deducing the answer from information contained in the data base.

Cambridge, Mass.: MIT Press

Department of Computer Science University of Edinburgh 1. INTRODUCTION In this paper we describe a program which will assign deep and surface structure analyses to an infinite number of English sentences.1 The design of this program differs in several respects from that of other automatic parsers presently in existence. All these differences are a consequence of the particular aim we have pursued in writing the program, which represents an attempt to construct a device that will not only assign a syntactic analysis to any English sentence-that is, a record of the syntactic structure that the native speaker Perceives in any English sentence-but which also, to some extent, simulates the way in which he perceives this structure. This is not to say that the analyzer differs from others because we have based its design upon the findings of psycholinguistic experiments. For one thing very few experiments on the perception of syntactic structure have been carried out and for the most part the results have been fairly inconclusive. But it is the case that we have, as far as possible, treated the task of constructing an automatic parser as being itself a psycholinguistic experiment. That is to say, any proposal regarding the possible operation of the program has been judged (mainly as the result of introspection) according to whether or not it seemed to be consistent with human behaviour. And this has led to our incorporating certain features which are absent from other automatic parsing systems.

A not-for-profit organization, IEEE is the world's largest technical professional organization dedicated to advancing technology for the benefit of humanity.

This paper is a survey of the current machine translation research in the US, Europe and Japan. A short history of machine translation is presented first, followed by an overview of the current research work. Representative examples of a wide range of different approaches adopted by machine translation researchers are presented. These are described in detail along with a discussion of the practicalities of scaling up these approaches for operational environments. In support of this discussion, issues in, and techniques for, evaluating machine translation systems are addressed.