Knowledge representation and the semantics of natural language. With CD-ROM.

From MaRDI portal





As the author points out (Preface, p.~v): ``Natural Language is not only the most important means of communication between human beings, it is also used over historical periods for the preservation of cultural achievements and their transmission from one generation to the other. It has been shown indeed in particular by \textit{Eric Havelock} [Preface to Plato] that this was the primary function of Homeric state of mind (doxa) in pre-literate Greece. By a strange twist of history we are seeing nowadays a return to oral rather than written communication for the popular and youth strata, in particular in the media -- although this is not exactly the pre-literate state of mind [\textit{Walter J. Ong}, Orality and literacy]. Anyway, it's obvious given the enormous increase of digitalised information that the importance of natural language-based communication can only grow. We need to understand it theoretically and practically, i.e., its automated treatment. One of the major components is the semantic representation of knowledge, since it connects all components of natural language processing systems. The book under review undertakes to give a method for the semantic representation of natural language. It purports to be a universal knowledge representation in human sciences (linguistics, cognitive psychology, philosophy of language) as well as in computational linguistics and artificial intelligence and thus unify to some extent these disciplines. The universally applicable formalism is by means of Multilayered Extended Semantic Networks (MultiNets for short) that this book proceeds to describe. It is used at the University of Hagen as the common core of all aspects of natural language processing, be they theoretical or practical. This semantic representation of natural language by MultiNets is largely independent of the considered language. However the syntactic constructs in different languages used to describe the same content are different. In this book the syntactic phenomena and their description via MultiNets apply to German. Some familiarity with German is therefore useful, although the examples are mostly translated into English showing the similarity of semantic interpretation for languages as close to each other as German and English. Although the present book is concerned mainly with with the description of a system for semantic knowledge representation and not with the logical formalism working with this language, there are many references to the axiomatic underpinnings and inferential power of the expressional means -- this is important for their operational semantics in a question-answering system for instance. A different set of axioms and inference rules could be chosen to get a different manifestation of a knowledge base, they are built on the same representational means. There are two parts in the book: Part I deals with the most important problems of semantic knowledge representation and the semantic interpretation of typical natural language phenomena. Part II provides the basic constructs of MultiNet and gives a systematic compilation of the representational means according to a uniform schema. From the practical point of view, it is important to know that the knowledge representation method of the book is supported by several software tools described in the book. MultiNet has also been used for different projects as an interface between clients and a data server or as a semantic interlingua or as a supporting means for disabled people or a communication tool between students and an automatic tutor. A detailed review of the contents of the book is beyond this brief review. It may help the reader to know that MultiNet is the latest in two lines of development: -- the first, including structured inheritance networks and the knowledge representation KL-ONE is strongly influenced by the framework concept; they are also associated with logic and model-theoretic foundations such as SB-ONE, Terminological Logics and Description Logics; -- the second is closer to the cognitive sciences and semantic concepts going back to Wittgenstein and including extensional as well as intentional aspects included in the knowledge representation itself. All these are of course interconnected and have drawn from logic-oriented systems such as Generalized Quantifier Theory, File Change Theory, Discourse Representation Theory.





Describes a project that uses

Uses Software






This page was built for publication: Knowledge representation and the semantics of natural language. With CD-ROM.

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5714320)