Joint Alignment and Artificial Data Generation: An Empirical Study of Pivot-based Machine Transliteration

Proceedings of the 5th International Joint Conference on Natural Language Processing, pages 1207–1215, Chiang Mai, Thailand, November 8 – 13, 2011. c⃝2011 AFNLP Joint Alignment and Artificial Data Generation: An Empirical Study of Pivot-based Machine Transliteration Min Zhang1 Xiangyu Duan1 Ming Liu1 Yunqing Xia2 Haizhou Li1 1Institute for Infocomm Research, A-STAR, Singapore {mzhang,xduan,mliu,hli}@i2r.a-star.edu.sg 2Dept. of Comp. Sci. & Tech., Tsinghua University, Beijing yqxia@tsinghua.edu.cn Abstract In this paper, we first carry out an investi- gation on two existing pivot strategies for statistical machine transliteration, namely system-based and model-based strategies, to figure out the reason why the previous model-based strategy performs much worse than the system-based one. We then propose a joint alignment algorithm to op- timize transliteration alignments jointly across source, pivot and target languages to improve the performance of the model- based strategy. In addition, we further propose a novel synthetic data-based strategy, which artificially generates source-target data using pivot language. Experimental results on benchmarking da- ta show that the proposed joint alignment optimization algorithm significantly im- proves the accuracy of model-based strat- egy and the proposed synthetic data-based strategy is very effective for pivot-based machine transliteration. 1 Introduction Machine transliteration refers to the phonetic translation of names across languages by comput- er. With the rapid growth of the Internet data and the dramatic changes in the user demographics especially among the non-English speaking parts of the world, machine transliteration is important in many cross-lingual NLP, MT and CLIR appli- cations as their performances have been shown to positively correlate with the correct conversion of names between the languages in several studies (Demner-Fushman and Oard, 2002; Mandl and Womser-Hacker, 2005; Hermjakob et al., 2008; Udupa et al., 2009). However, the traditional source for name equivalence, the bilingual dictio- naries — whether handcrafted or statistical built — offer only limited support because new names always emerge. All of the above points to the critical need for high-performance machine transliteration tech- nology. Much research effort has been made to address this issue in the research community (Knight and Graehl, 1998; Meng et al., 2001; Al- Onaizan and Knight, 2002; Oh and Choi, 2002; Klementiev and Roth, 2006; Sproat, 2006; Zelen- ko and Aone, 2006; Li et al., 2004, 2009a, 2009b; Sherif and Kondrak, 2007; Bertoldi et al., 2008; Goldwasser and Roth, 2008). These previous work falls into three categories, i.e., grapheme- based, phoneme-based and hybrid methods (Li et al., 2009a, 2009b). The report of the first machine transliteration shared task NEWS 2009 (Li et al., 2009a, 2009b) provides common benchmarking data in diverse language pairs and systematically evaluate the state-of-the-art technologies using their provided data. Although promising results have been reported, one of major issues is that the current translitera- tion methods rely heavily on significant amount of source-target parallel data to learn translitera- tion model. However, such corpora are not always available and the amounts of the currently availa- ble corpora, even for language pairs with English involved, are far from enough for training, letting alone many low-density language pairs. Indeed, …

Content truncated to 500 words