A Hybrid Approach for Automatic Extraction of Bilingual Multiword Expressions from Parallel Corpora

A Hybrid Approach for Automatic Extraction of Bilingual Multiword Expressions from Parallel Corpora Nasredine Semmar CEA, LIST, Vision and Content Engineering Laboratory F-91191, Gif-sur-Yvette, France nasredine.semmar@cea.fr Abstract Specific-domain bilingual lexicons play an important role for domain adaptation in machine translation. The entries of these types of lexicons are mostly composed of MultiWord Expressions (MWEs). The manual construction of MWEs bilingual lexicons is costly and time-consuming. We often use word alignment approaches to automatically construct bilingual lexicons of MWEs from parallel corpora. We present in this paper a hybrid approach to extract and align MWEs from parallel corpora in a one-step process. We formalize the alignment process as an integer linear programming problem in order to find an approximated optimal solution. This process generates lists of MWEs with their translations, which are then filtered using linguistic patterns for the construction of the bilingual lexicons of MWEs. We evaluate the bilingual lexicons of MWEs produced by this approach using two methods: a manual evaluation of the alignment quality and an evaluation of the impact of this alignment on the translation quality of the phrase-based statistical machine translation system Moses. We experimentally show that the integration of the bilingual MWEs and their linguistic information into the translation model improves the performance of Moses. Keywords: Bilingual lexicon, Multiword expression, Terminology extraction, Domain adaptation, Statistical machine translation 1. Introduction A MultiWord Expression (MWE) is a combination of words for which syntactic or semantic properties of the whole expression cannot be obtained from its components (Sag et al., 2002). Such units could be collocations, compound words, named entities, idioms, etc. They constitute an important part of the lexicon of any natural language (Jackendoff, 1997). Bilingual lexicons of MWEs play a vital role in Machine Translation (MT) and Cross- Language Information Retrieval (CLIR) because for a specific domain the specialized vocabulary is largely dominated by MWEs. The manual construction of these lexicons is costly and time-consuming. Word alignment approaches are often used to automatically construct bilingual lexicons from parallel corpora. Several word alignment approaches have been explored (Daille et al., 1994; Barbu, 2004) and many automatic word alignment tools are available, such as Giza++ (Och and Ney, 2000). However, most of these tools are efficient only to align single words (Fraser and Marcu, 2007). In this paper, we describe and evaluate a hybrid approach to automatically extract and align MWEs from an English-French parallel corpus. In contrast to traditional approaches for MWEs alignment which consist in firstly identifying monolingual MWEs candidates and secondly applying alignment to find bilingual correspondences, our approach extracts and aligns MWEs in a one-step process. The remainder of the paper is organized as follows. We define in Section 2 the notion of Multiword Expression and describe different types of MWEs with examples. In Section 3, we survey previous works addressing the tasks of extracting and aligning MWEs from parallel corpora. Section 4 introduces our hybrid approach to build bilingual lexicons of MWEs from sentence aligned parallel corpora. The experimental results are reported and …

Content truncated to 500 words