Reporting Preliminary Automatic Comparable Corpora Compilation Results
Ekaterina Stambolieva · Recent Advances in Natural Language Processing · 2013
Translation and translation studies rely heavily on distinctive text resources, such as comparable corpora. Comparable corpora gather greater diversity of language-dependent phrases in comparison to multilingual electronic dictionaries or parallel corpora; and present a robust language resource. Therefore, we see comparable corpora compilation as impending in this technological era and suggest an automatic approach to their gathering. The originality of the research lies within the newly-proposed methodology that is guiding the compilation process. We aim to contribute to translation and translation studies professionals’ work by suggesting an approach to obtaining comparable corpora without intermediate human evaluation. This contribution reduces time and presents such professionals with non-static text resources. In our experiment we compare the automatic compilation results to the labels, which two human evaluators have given to the