The semantical structure of natural language. Knowledge representation using MultiNet

Publication date 28 March 2001

knowledge representation multi-layer extended semantic networks MultiNet natural language NL software systems

Logic of natural languages (03B65) Research exposition (monographs, survey articles) pertaining to computer science (68-02) Knowledge representation (68T30) Theory of languages and software systems (knowledge-based systems, expert systems, etc.) for artificial intelligence (68T35) Natural language processing (68T50) Linguistics (91F20)

Review text

The main purpose of this really interesting book is to propose the MultiNet methodology of multi-layer extended semantic networks as a powerful tool for Natural Language (NL) representation and automated processing, and to support the MultiNet formalism by several performant NL software systems. The author successfully proves in his book that MultiNet, as a knowledge representation device used to handle the semantic problems of NL, may offer essential advantages over the classical tool of NL knowledge representation, i.e. the classical predicate logic. This type of multi-layer extended semantic network has been used, as a common formalism platform, to implement several NL processing systems that cover all the important aspects of NL automatic processing, from information systems of NL development to NL interfaces and lexicographic tools. The inferential expressive power of MultiNet is obtained in the framework of an operational semantics associated to Question Answering system contexts. The complete meaning of a Question-Answering context represented in MultiNet is obtained by appropriate choosing of various Axioms and Inference Rules associated to concrete relations, respectively functions, and corresponding to salient logical-type features of a NL knowledge base. The book consists of two parts: Part I is treating the basic problems of semantic representation of knowledge, viz. the semantic interpretation of NL phenomena. The MultiNet semantic networks, as a formal representation language is described in Part II. This one contains a systematic classification of the whole repertoire of MultiNet representation devices, always described according to a unitary scheme. The best approach to reading the book would be to survey first the Part II, and then to read carefully in Part I, using Part II as a reference guide to each function of the representation tools. The following appendices make easier the reading and understanding of the book contents: Appendix A encloses the list of all the abbreviations within the book. Appendix B contains the relations and functions used by MultiNet, in lexicographic ordering. Appendix C provides a classification of the semantic templates, formulated as predicate types, completing the meaning of relations for the interrogation templates, and explaining the chosen sequence of arguments. Appendix D specifies the default value of the (argument) ordering for the relation \(R\) that concerns the inherent situational knowledge of a node \(N\) depending on \(R\), either \(N\) is the first of the second argument. Finally, the Appendix E comprises all the typical axioms (and examples) that provide the inferential expressive power of MultiNet representation devices. The multi-layer extended semantic networks of MultiNet have been utilized both as a theoretical formalism and as a practical tool for the proposed knowledge representation methods, e.g. within a lexical knowledge engineer toolbox, or a syntactic-semantic analysis system that translates automatically a NL phrase into its meaning representation, etc. MultiNet was also used and evaluated within the following research projects: (a) NRW project of ``Virtual Knowledge Factory, as interface formal language between User-Client and image-data bank. (b) DFG project ``NLI for Z39.50 (NL Interface for Internet protocol). (c) High-school project ``Interactive AI-Laboratory, as central formalism for knowledge representation. All these projects and systems have been used evaluated also within AI (Artificial Intelligence) and Knowledge Processing universitary courses, at the Technical University Dresden (Germany) and the FernUniversität (TeleUniversity) Hagen (Germany). The CD accompanying the book provides, in useful formats (mainly .PDF), the complete contents of the book together with running examples of the NL processing systems based on MultiNet formalism.

Recommendations

Cited in

(11)

This page was built for publication: The semantical structure of natural language. Knowledge representation using MultiNet

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2706824)