Attestation committee
Accreditation committee
Expert committee
Dispositions, instructions
Normative acts
Nomenclature
Institutions
Scientific councils
Seminars
Theses
Scientific advisers
Scientists
Doctoral students
Postdoctoral students
CNAA logo

 română | русский | english

CNAA / Theses / 2008 / May /

Computational analysis of business correspondence


Author: Victoria Maxim
Degree:doctor of physics and mathematics
Speciality: 01.05.04 - Mathematical modelling, mathematical methods, software
Year:2008
Scientific advisers: Anatol Popescu
doctor habilitat, professor, Technical University of Moldova
Anatol Ciobanu
doctor habilitat, professor, Moldova State University
Institution: Institute of Mathematics and Computer Science, Academy of Sciences of Moldova
Scientific council: DH 01-01.05.04
Institute of Mathematics and Computer Science, Academy of Sciences of Moldova

Status

The thesis was presented on the 23 May, 2008
Approved by NCAA on the 19 June, 2008

Abstract

Adobe PDF document0.79 Mb / in romanian

Thesis

CZU 004.912

Adobe PDF document 2.12 Mb / in romanian
220 pages

Keywords

Term acquisition, Automatic analysis, Bilingual business term dictionary, Concept ontological aspect, Morphological codification, Business concepts, Business correspondence , Parallel bilingual corpora, Part of speech tagging, Term extraction, Machine learning, Isomorphism,,Commercial language, Programming language Perl, Programming language Delphi, Translation Memory, Linguistic model, Parsing, Natural language processing, Transformation rules, Phrase formal representation, Semantic representations, Semantic network, Synergetics, Noun group, Computational Linguistic Re

Summary

This work represents a study accomplished for the automatic analysis of the business letters with the aim to elaborate the technology and necessary instruments in order to create software in Assisted Machine Translation. It is based on the bilingual English- Romanian parallel corpora of business letters (42 300 words), and specialized dictionaries.

The thesis is based on synergy between Linguistics and Informatics, that’s why the issues approach was through analogies, synthesis and correlation of the results.

We have proposed the analysis of Commercial Language at different levels, including morphological, syntactic, semantic, stylistical and vocabulary. We bring arguments of its existence at each level.

This work also includes: a general presentation of the Natural Language Processing where we describe the Supervised Machine Learning method, the Synergy between Linguistics and Informatics, the Romanian studies and realizations in NLP, the architecture and levels of an NLP system, some of syntax (parsing), semantics (λ - calculus), pragmatics, discourse and Machine Translation issues.

We’ve created a generic ontology of business concepts, which can be enlarged by Ontology Editor Protégé.

We examine the parallel English-Romanian corpora of business letters and present the results of their computational analysis. There are suggestions on terminological and ontological analysis of business language; the basic elements and methodology of Machine Assisted Translation of business letters. We have extracted 1831 pairs of terminological units, for the complex terms were being created 43 models of transformational rules of English Noun Part (NP) into Romanian NP.

Using the elaborated computational resources, we’ve created software BusLet Translator. It is based on Object Oriented Programming, Translation Memory and the bilingual Dictionary of business terms.

This research points out the value aspect of the conceptual frame, from which have been obtained new scientific results. The theoretical research it is being used in teaching course of Computational Linguistics. We specify the necessity of subsequent scientific and didactic contributions.