Previous work has shown that Chinese word segmentation is useful for machine translation to English, yet the way different segmentation strategies affect MT is still poorly understood. In this paper, we demonstrate that optimizing segmentation for an existing segmentation standard does not always yield better MT performance. We find that other factors such as segmentation consistency and granularity of Chinese “words ” can be more important for machine translation. Based on these findings, we implement methods inside a conditional random field segmenter that directly optimize segmentation granularity with respect to the MT task, providing an improvement of 0.73 BLEU. We also show that improving segmentation consistency using external lexicon and proper noun features yields a 0.32 BLEU increase. 1

For Chinese POS tagging, word segmentation is a preliminary step. To avoid error propagation and improve segmentation by utilizing POS information, segmentation and tagging can be performed simultaneously. A challenge for this joint approach is the large combined search space, which makes efficient decoding very hard. Recent research has explored the integration of segmentation and POS tagging, by decoding under restricted versions of the full combined search space. In this paper, we propose a joint segmentation and POS tagging model that does not impose any hard constraints on the interaction between word and POS information. Fast decoding is achieved by using a novel multiple-beam search algorithm. The system uses a discriminative statistical model, trained using the generalized perceptron algorithm. The joint model gives an error reduction in segmentation accuracy of 14.6 % and an error reduction in tagging accuracy of 12.2%, compared to the traditional pipeline approach. 1

Standard approaches to Chinese word segmentation treat the problem as a tagging task, assigning labels to the characters in the sequence indicating whether the character marks a word boundary. Discriminatively trained models based on local character features are used to make the tagging decisions, with Viterbi decoding finding the highest scoring segmentation. In this paper we propose an alternative, word-based segmentor, which uses features based on complete words and word sequences. The generalized perceptron algorithm is used for discriminative training, and we use a beamsearch decoder. Closed tests on the first and second SIGHAN bakeoffs show that our system is competitive with the best in the literature, achieving the highest reported F-scores for a number of corpora. 1

We show that the standard beam-search algorithm can be used as an efficient decoder for the global linear model of Zhang and Clark (2008) for joint word segmentation and POS-tagging, achieving a significant speed improvement. Such decoding is enabled by: (1) separating full word features from partial word features so that feature templates can be instantiated incrementally, according to whether the current character is separated or appended; (2) deciding the POS-tag of a potential word when its first character is processed. Early-update is used with perceptron training so that the linear model gives a high score to a correct partial candidate as well as a full output. Effective scoring of partial structures allows the decoder to give high accuracy with a small beam-size of 16. In our 10-fold crossvalidation experiments with the Chinese Treebank, our system performed over 10 times as fast as Zhang and Clark (2008) with little accuracy loss. The accuracy of our system on the standard CTB 5 test was competitive with the best in the literature. 1

We study a range of syntactic processing tasks using a general statistical framework that consists of a global linear model, trained by the generalized perceptron together with a generic beamsearch decoder. We apply the framework to word segmentation, joint segmentation and POStagging, dependency parsing, and phrase-structure parsing. Both components of the framework are conceptually and computationally very simple. The beam-search decoder only requires the syntactic processing task to be broken into a sequence of decisions, such that, at each stage in the process, the decoder is able to consider the top-n candidates and generate all possibilities for the next stage. Once the decoder has been defined, it is applied to the training data, using trivial updates according to the generalized perceptron to induce a model. This simple framework performs surprisingly well, giving accuracy results competitive with the state-of-the-art on all the tasks we consider. The computational simplicity of the decoder and training algorithm leads to significantly higher test speeds and lower training times than their main alternatives, including log-linear and large-margin training algorithms and dynamic-programming for decoding. Moreover, the framework offers the freedom to define arbitrary features which can make alternative training and decoding algorithms prohibitively slow. We discuss how the general framework is applied to each of the problems studied in this article, making comparisons with alternative learning and decoding algorithms. We also show how the comparability of candidates considered by the beam is an important factor in the performance. We argue that the conceptual and computational simplicity of the framework, together with its language-independent nature, make it a competitive choice for a range of syntactic processing tasks and one that should be considered for comparison by developers of alternative approaches. 1.

Many sequence labeling tasks in NLP require solving a cascade of segmentation and tagging subtasks, such as Chinese POS tagging, named entity recognition, and so on. Traditional pipeline approaches usually suffer from error propagation. Joint training/decoding in the cross-product state space could cause too many parameters and high inference complexity. In this paper, we present a novel method which integrates graph structures of two subtasks into one using virtual nodes, and performs joint training and decoding in the factorized state space. Experimental evaluations on CoNLL 2000 shallow parsing data set and Fourth SIGHAN Bakeoff CTB POS tagging data set demonstrate the superiority of our method over cross-product, pipeline and candidate reranking approaches. 1

In this paper, we present a discriminative word-character hybrid model for joint Chinese word segmentation and POS tagging. Our word-character hybrid model offers high performance since it can handle both known and unknown words. We describe our strategies that yield good balance for learning the characteristics of known and unknown words and propose an errordriven policy that delivers such balance by acquiring examples of unknown words from particular errors in a training corpus. We describe an efficient framework for training our model based on the Margin Infused Relaxed Algorithm (MIRA), evaluate our approach on the Penn Chinese Treebank, and show that it achieves superior performance compared to the state-ofthe-art approaches reported in the literature. 1

State-of-the-art Chinese word segmentation systems have achieved high performance when training data and testing data are from the same domain. However, they suffer from the generalizability problem when applied on test data from different domains. We introduce a multi-layer Chinese word segmentation system which can integrate the outputs from multiple heterogeneous segmentation systems. By training a second layer of large margin classifier on top of the outputs from several Conditional Random Fields classifiers, it can utilize a small amount of in-domain training data to improve the performance. Experimental results show consistent improvement on F1 scores and OOV recall rates by applying the approach. 1

This degree is submitted to the University of Cambridge

Machine Translation (MT) is a task with multiple components, each of which can be very challenging. This thesis focuses on a difficult language pair – Chinese to English – and works on several language-specific aspects that make translation more difficult. The first challenge this thesis focuses on is the differences in the writing systems. In Chinese there are no explicit boundaries between words, and even the definition of a “word” is unclear. We build a general purpose Chinese word segmenter with linguistically inspired features that performs very well on the SIGHAN 2005 bakeoff data. Then we study how Chinese word segmenter performance is related to MT performance, and provide a way to tune the “word ” unit in Chinese so that it can better match up with the English word granularity, and therefore improve MT performance. The second challenge we address is different word order between Chinese and English. We first perform error analysis on three state-of-the-art MT systems to see what the most prominent problems are, especially how different word orders cause translation errors. According to our findings, we propose two solutions to improve Chinese-to-English