Abstract

The current paper is devoted to a formal analysis of the space category and, especially, to questions bound with the presentation of space relations in a formal NLP model. The aim is to demonstrate how linguistic and cognitive problems relating to spatial categorization, definition of spatial entities, and the expression of different locative senses in natural languages can be solved in an artificial intelligence system. We offer a description of the locative groups in the ABBYY Compreno formalism – an integral NLP framework applied for machine translation, semantic search, fact extraction, and other tasks based on the semantic analysis of texts. The model is based on a universal semantic hierarchy of the thesaurus type and includes a description of all possible semantic and syntactic links every word can attach. In this work we define the set of semantic locative relations between words, suggest different tools for their syntactic presentation, give formal restrictions for the word classes that can denote spaces, and show different strategies of dealing with locative prepositions, especially as far as the problem of their machine translation is concerned.

Keywords: ABBYY Compreno, space relations, NLP model, semantics, spatial categorization, semantic hierarchy

How to Cite:

Leontyev, A. & Petrova, M., (2015) “A Description Of Space Relations In An NLP Model: The ABBYY Compreno Approach”, Baltic International Yearbook of Cognition, Logic and Communication 1(2015). doi: https://doi.org/10.4148/1944-3676.1096