Non-Projective Parsing for Statistical Machine Translation

Proceedings of the 2009 Conference on Empirical Methods in Natural Language Processing, pages 200–209, Singapore, 6-7 August 2009. c⃝2009 ACL and AFNLP Non-Projective Parsing for Statistical Machine Translation Xavier Carreras Michael Collins MIT CSAIL, Cambridge, MA 02139, USA {carreras,mcollins}@csail.mit.edu Abstract We describe a novel approach for syntax- based statistical MT, which builds on a variant of tree adjoining grammar (TAG). Inspired by work in discriminative depen- dency parsing, the key idea in our ap- proach is to allow highly flexible reorder- ing operations during parsing, in combina- tion with a discriminative model that can condition on rich features of the source- language string. Experiments on trans- lation from German to English show im- provements over phrase-based systems, both in terms of BLEU scores and in hu- man evaluations. 1 Introduction Syntax-based models for statistical machine trans- lation (SMT) have recently shown impressive re- sults; many such approaches are based on ei- ther synchronous grammars (e.g., (Chiang, 2005)), or tree transducers (e.g., (Marcu et al., 2006)). This paper describes an alternative approach for syntax-based SMT, which directly leverages meth- ods from non-projective dependency parsing. The key idea in our approach is to allow highly flexible reordering operations, in combination with a dis- criminative model that can condition on rich fea- tures of the source-language input string. Our approach builds on a variant of tree adjoin- ing grammar (TAG; (Joshi and Schabes, 1997)) (specifically, the formalism of (Carreras et al., 2008)). The models we describe make use of phrasal entries augmented with subtrees that pro- vide syntactic information in the target language. As one example, when translating the sentence wir m¨ussen auch diese kritik ernst nehmen from German into English, the following sequence of syntactic phrasal entries might be used (we show each English syntactic fragment above its associ- ated German sub-string): S NP we VP must ADVP also NP these criticisms ADVP seriously VP take wir m¨ussen auch diese kritik ernst nehmen TAG parsing operations are then used to combine these fragments into a full parse tree, giving the final English translation we must also take these criticisms seriously. Some key aspects of our approach are as fol- lows: • We impose no constraints on entries in the phrasal lexicon. The method thereby retains the full set of lexical entries of phrase-based systems (e.g., (Koehn et al., 2003)).1 • The model allows a straightforward integra- tion of lexicalized syntactic language models—for example the models of (Charniak, 2001)—in addi- tion to a surface language model. • The operations used to combine tree frag- ments into a complete parse tree are signifi- cant generalizations of standard parsing operations found in TAG; specifically, they are modified to be highly flexible, potentially allowing any possible permutation (reordering) of the initial fragments. As one example of the type of parsing opera- tions that we will consider, we might allow the tree fragments shown above for these criticisms and take to be combined to form a new structure with the sub-string take these criticisms. This step in …

Content truncated to 500 words