B O O K RE VIE W APPROCHES FORMELLES DE LA S E M A N T I Q U E NATURELLE (Formal Approaches to Natural Semantics) Editors: A. Borillo, M. Borillo, L. Farinas del Cerro & J. Virbel Laboratoire de Langages et Systemes Informatiques Toulouse, France 1982 Reviewed by: James Reggia and Vicki Roy Department of Computer Science University of Maryland College Park, M D 20742 This paperback book is a collection of presentations made at the Linguistics S u m m e r School for C o m p u t e r Scientists at Toulouse, France in September, 1981. It a t t e m p t s to fill a perceived educational gap in the interdisciplinary area between c o m p u t e r science (especially AI) and human communication (read "traditional linguistics"). All but t w o of the papers are written in French, with topics ranging from the t h e o r e t i c a l / philosophical to descriptions of i m p l e m e n t e d programs. We summarize t w o representative papers in this book, and then make some general comments. A s c o m b r e ("Argumentative Structures and Linguistics Expression of Reasoning") begins by contrasting the scientific method with the linguistic method. The f o r m e r is characterized as the simulation of natural mechanisms which can explain phenomena observed in nature. In contrast, the latter involves the application of artificial mechanisms to constructively reproduce natural phenomena (typically through formula generation). The subjective nature of linguistic research is emphasized. The relationships between syntax, semantics and pragmatics are explored, and an a r g u m e n t is made that pragrnatics can determine both syntax and semantics. Pierrel ("On the Use of Linguistic Information in Automatic Speech Recognition and Understanding") describes the use of linguistic knowledge in the MYRTILLE II system for automatic, continuous speech understanding. The goal of MYRTILLE II is the recognition and understanding of p s e u d o - n a t u r a l language, defined as language w i t h the same syntactic c o m p l e x i t y as spoken French and a vocabulary of more than 350 w o r d s in a relatively restricted domain. MYRTILLE II uses multiple k n o w l e d g e sources, from phonetics through pragmatics, and integrates syntactic and semantic aspects of parsing in a h y p o t h e s i z e - a n d - t e s t control structure. A "procedural nodes hierarchy" (similar to an ATN) and a lexicon organized into multiple hierarchies are involved. The t w o papers described above illustrate the scope of this collection, from the abstract linguistic discussion to the description of concrete implementations. Other papers address such issues as mental imagery, ambiguity, difficulties involved in formalizing linguistic knowledge, representation of t e m p o r a l events, the role of logic in discourse, Montague's universal grammar, and the formal representation of causality. The general emphasis is on linguistic concepts rather than applications. One is left with the sense that a group of linguists is making an effort to establish a dialogue w i t h their linguistically-naive colleagues in A I - o r i e n t e d natural language processing. As in m o s t collection of papers, those in this report vary in quality, but w e suspect that many AI researchers in natural language (presuming fluency in French) m i g h t find s o m e t h i n g of interest and certainly a different perspective in this book. ARTICLE MASTERING MASTER M I N D LOGICALLY Peter Koppstein Department of Political Science Washington University St. Louis March 1, 1984 INTRODUCTION Some readers of the recent note by Ehud Shapiro on "Playing Master Mind Logically" (SIGART 85, July 1983) may reasonably have concluded that Prolog was a suitable language for a "generate and test" strategy, as the author stated, but that perhaps there was something intrinsic to Prolog that made anything much more subtle unwieldy. The present note at any rate shows h o w a more "logical" strategy incorporating that previously described may readily be i m p l e m e n t e d w i t h o u t c o m p r o m i s i n g the elegance and simplicity of Ehud Shapiro's original problem. A related approach, using templates as generators, is also described. MASTER MIND The version of Master Mind described involves the

/lp/association-for-computing-machinery/review-of-approches-formelles-de-la-semantique-naturelle-formal-qfXcxzQVV9