Using Transliteration of Proper Names from Arabic to Latin Script to Improve English-Arabic Word Alignment

International Joint Conference on Natural Language Processing, pages 1022–1026, Nagoya, Japan, 14-18 October 2013. Using Transliteration of Proper Names from Arabic to Latin Script to Improve English-Arabic Word Alignment Nasredine Semmar Institut CEA LIST, Laboratoire Vision et Ingénierie des Contenus CEA Saclay – Nano-INNOV, F-91191 Gif-sur-Yvette Cedex, France nasredine.semmar@cea.fr Houda Saadane LIDILEM, Université Grenoble III Domaine Universitaire 1180, avenue centrale, F-38400 Saint Martin d'Hères, France houda.saadane@e.u- grenoble3.fr Abstract Bilingual lexicons of proper names play a vital role in machine translation and cross-language information retrieval. Word alignment ap- proaches are generally used to construct bilin- gual lexicons automatically from parallel cor- pora. Aligning proper names is a task particu- larly difficult when the source and target lan- guages of the parallel corpus do not share a same written script. We present in this paper a system to transliterate automatically proper names from Arabic to Latin script, and a tool to align single and compound words from English-Arabic parallel texts. We particularly focus on the impact of using transliteration to improve the performance of the word align- ment tool. We have evaluated the word align- ment tool integrating transliteration of proper names from Arabic to Latin script using two methods: A manual evaluation of the align- ment quality and an evaluation of the impact of this alignment on the translation quality by using the open source statistical machine translation system Moses. Experiments show that integrating transliteration of proper names into the alignment process improves the F- measure of word alignment from 72% to 81% and the translation BLEU score from 20.15% to 20.63%. 1 Introduction Bilingual lexicons of proper names play a vital role in Machine Translation (MT) and Cross- Language Information Retrieval (CLIR). Word alignment approaches are generally used to con- struct bilingual lexicons automatically from par- allel corpora. Aligning proper names requires both recognition of the proper names present in the parallel corpus and their alignment (Abuleil and Evens, 2004). This task is particularly diffi- cult when the source and target languages of the parallel corpus do not share a same written script. A solution to this issue consists in writing the proper names present in the parallel corpus in the same written script. This operation is named transliteration and consists in replacing each grapheme of a writing system by another graph- eme or a group of graphemes of another writing system, regardless of pronunciation. In order to study the impact of using translit- eration to improve the performance of a word alignment tool, we present in this paper a system to transliterate automatically proper names from Arabic to Latin script, and a tool to align single and compound words from English-Arabic paral- lel texts. The remainder of the paper is organized as fol- lows: Section 2 recalls in some previous work addressing tasks of transliteration and bilingual lexicon extraction from parallel corpora. In sec- tion 3, we present briefly the system for automat- ic transliteration of proper names from Arabic to Latin script. Section 4 describes the process of …

Content truncated to 500 words