Average Reviews:
(More customer reviews)This book has a wealth of solid information about knowledge representation (KR). It is focused on Multinets, an advanced type of semantic network. The author brings years of experience to this subject. The book deals thoroughly with many aspects of KR such as relationships, categories, generalization, real vs hypothetical, cardinality, and variability, just to name a few. Although the book is focused on Multinets, the concepts should be relevant to other knowledge representation schemes.
Frequent natural language statements clarify the issues under discussion. For example, here are two of the statements on page 49 that illustrate different types of information that can be communicated by "is".
"'A bachelor is an adult unmarried man.' (Relation EQU), 'The cherry is red.' (Relation PROP), . . ."
In other words, the first "is" links equivalent ideas, the second "is" indicates "red" as a property.
The book also contains numerous, well designed, helpful diagrams.
Unfortunately, the jargon is very thick, and difficult to navigate. For example, from page 20: "As already mentioned, the arcs of the semantic network have to be considered as epistemologically and cognitively justified categories which function as fundamental deep semantic relations in the framework of meaning representation." After reading this sentence several times, I think part of what it means is "the arcs of semantic networks should match actual deep semantic relations". But I'm not sure. I frequently find myself reading sentences several times, then moving on, only partly satisfied.
If you can get past the jargon, the book contains a thorough, solidly supported treatment of the knowledge representation of natural language.
Perhaps someone with more prior experience with knowledge representation would find the book more approachable.
Based on the quantity of information, the strong relevance to knowledge representation and the clarifying examples, I give this a generous 4 star rating. However, I'm worried that some readers will find the jargon so impenetrable that the book would lose most of its value to them.
Click Here to see more reviews about: Knowledge Representation and the Semantics of Natural Language (Cognitive Technologies)
The book presents an interdisciplinary approach to knowledge representation and the treatment of semantic phenomena of natural language, which is positioned between artificial intelligence, computational linguistics, and cognitive psychology. The proposed method is based on Multilayered Extended Semantic Networks (MultiNets), which can be used for theoretical investigations into the semantics of natural language, for cognitive modeling, for describing lexical entries in a computational lexicon, and for natural language processing (NLP).Part I deals with fundamental problems of semantic knowledge representation and semantic interpretation of natural language phenomena. Part II provides a systematic description of the representational means of MultiNet, one of the most comprehensive and thoroughly specified collections of relations and functions used in real NLP applications.MultiNet is embedded into a system of software tools comprising a workbench for the knowledge engineer, a semantic interpreter translating natural language expressions into formal meaning structures, and a workbench for the computer lexicographer. The book has been used for courses in artificial intelligence at several universities and is one of the cornerstones for teaching computational linguistics in a virtual electronic laboratory.
