Librería Samer Atenea
Librería Aciertas (Toledo)
Kálamo Books
Librería Perelló (Valencia)
Librería Elías (Asturias)
Donde los libros
Librería Kolima (Madrid)
Librería Proteo (Málaga)
Language is a set of terms (word forms, word groups, phrases) and rules that denote or encode objects, concepts and their transformations in time and space. Using these terms and grammatical rules the humans produce new codes - the natural (oral, written, or sign) speech to describe the surrounding reality to other humans. The speech has two: the expression (roughly S-structure) and the content (roughly D-structure) planes. Content encodes the meaning. The speech content is described as a tree of word phrases - the Context Tree (syntactic or phrase structure tree). This structure, which is universal for natural languages, encodes the meaning of speech. The universality is achieved not by using common abstract types in the Context Tree, but rather by including specific types from all languages. The Context Tree is not an abstract-universal, but rather a specific-cumulative structure. Such approach minimizes losses at representing thre Content Tree build for a sentence in one language in another (translation) to a theoretical minimum.The main goal of the book is to design data structures and algorithms for building the Context Tree. It touches also the difficulties of translation: encoding the meaning by Context Tree in one language and decoding it in another.The design of the Context Tree that can represent natural speech in any language is examined using Eastern Armenian language. Morphological data and syntactic rules of the Eastern Armenian language described in tabular form and by algorithms: the wordforms generation, stemming, tagging, and lemmatization. These algorithms along with the dictionaries of morphemes allow delinearizing/linearizing sentences into/from Context Trees. They are the core for building various natural speech processing applications such as spell checking, translation, corpus texts tagging and indexing, etc.Recently significant progress has been made in statistical approaches to natural speech processing. There is a section with quick overview of history and theory of statistical approaches to natural speech processing. A more detailed analysis of deep learning (artificial intelligence) with Recurrent Neural Networks is presented for the one of the most successful implementations - the sequence to sequence learning model.The book is written in Eastern Armenian.
