Monica Gavrila

Comparing CBMT Approaches for German-Romanian

There is no doubt that in the last couple of years corpus-based machine translation (CBMT) approaches have been in focus. Each of the CBMT approaches (statistical MT and example-based MT) has its advantages and disadvantages. Therefore, hybrid approaches have been developed. This paper presents a comparative study of CBMT approaches, using three types of systems: a statistical MT (SMT) system, an example-based MT (EBMT) system and a hybrid (EBMT-SMT) system. For our experiments we considered German-Romanian as language-pair, in this direction of translation, and a domain-restricted corpus, manually created and aligned (RoGER). The translation results have been both automatically and manually evaluated. For both evaluation types, the SMT system is ranked best in our experiments.

Text Genre – An Unexplored Parameter in Statistical Machine Translation

It is generally accepted that the performance of a statistical machine translation (SMT) system depends significantly on the concordance between the domain of training and test data. During the last years several methods have been proposed in order to deal with out- of-domain words. Less to no attention has been paid however to text genre within the same domain. In this paper we demonstrate that the style of the training corpus may influence the quality of the translation output even when the domain of the training and test data remains al- most unchanged, but the text genre changes. We use as training data the JRC-Acquis and as test data the Europarl corpus. We include also experiments with an out-of-domain test data, as comparison for the variation of performance of the SMT system.

ProLiV - a Tool for Teaching by Viewing Computational Linguistics

ProLiV - Animated Process-modeler of Complex (Computational) Linguistic Methods and Theories - is a fully modular, flexible, XML-based stand-alone Java application, used for computer-assisted learning in Natural Language Processing (NLP) or Computational Linguistics (CL). Having a flexible and extendible architecture, the system presents the students, by means of text, of visual elements (such as pictures and animations) and of interactive parameter set-up, the following topics: Latent Semantics Analysis (LSA), (computational) lexicons, question modeling, Hidden-Markov-Models (HMM), and Topic-Focus. These topics are addressed to first-year students in computer science and/or linguistics.