A Discriminative Matching Approach to Word Alignment

By B. Taskar, Simon Lacoste-Julien, Daniel Klein. 2005

190 citations recorded by the source corpus.

We present a discriminative, large-margin approach to feature-based matching for word alignment. In this framework, pairs of word tokens receive a matching score, which is based on features of that pair, including measures of association between the words, distortion between their positions, similarity of the orthographic form, and so on. Even with only 100 labeled training examples and simple features which incorporate counts from a large unlabeled corpus, we achieve AER performance close to IBM Model 4, in much less time. Including Model 4 predictions as features, we achieve a relative AER reduction of 22% in over intersected Model 4 alignments.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Natural Language Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computer Vision and Pattern Recognition

In Natural Language Processing

Citations in the corpus

SIGMa: simple greedy matching for aligning large knowledge bases (2012, background): “Another approach to alignment arises from the word alignment problem in natural language processing [23], which has been formulated as a maximum weighted bipartite matching problem [29] (thus exploiting the 1–1 matching structure).”
Using Semantic Roles to Improve Question Answering (2007, methodology): “Edge covers have been successfully applied in several natural language processing tasks, including machine translation (Taskar et al., 2005) and annotation projection (Pad´o and Lapata, 2006).”
An Introduction to Conditional Random Fields (2010, background): “Examples of the types of structured outputs that are considered include parse trees of natural language sentences [123, 31], alignments between sentences in different languages [124], and route plans in mobile robotics [97].”
Word Alignment via Quadratic Assignment (2006, methodology): “, and are documented in Taskar et al. (2005). We also augment those features with the predictions of IBM Model 2 run on the training and test sentences.”
Structured Prediction, Dual Extragradient and Bregman Projections (2006, methodology): “This approach, explored in Taskar et al. (2004a, 2005a), yields a joint convex optimization problem.”
Accelerated training of max-margin Markov networks with kernels (2011, methodology): “The max-oracle again relies on graphical models for dynamical programming [19], and many existing combinatorial optimizers can also be used, such as in the applications of matching [20] and context free grammar parsing [21].”
A Discourse Commitment-Based Framework for Recognizing Textual Entailment (2007, methodology): “Following Commitment Extraction, we used an word alignment technique first introduced in (Taskar et al., 2005b) in order to select the commitment”
Recurrent Neural Networks for Word Alignment Model (2014, methodology): “These models are roughly clustered into two groups: generative models, such as those proposed by Brown et al. (1993), Vogel et al. (1996), and Och and Ney (2003), and discriminative models, such as those proposed by Taskar et al. (2005), Moore (2005), and Blunsom and Cohn (2006).”
Structured Prediction via the Extragradient Method (2005, background): “The features on the word pair (ej , fk) include measures of association, orthography, relative position, predictions of generative models (see [22] for details).”
Discriminative Unsupervised Alignment of Natural Language Instructions with Corresponding Video Segments (2015, background): “Early attempts (Blunsom and Cohn, 2006; Taskar et al., 2005) towards discriminative word alignment relied on supervised hand-aligned parallel corpora.”
Database-Text Alignment via Structured Multilabel Classification (2007, background): “While most of the research focuses on unsupervised models [Brown et al., 1993], recently a number of supervised discriminative approaches have been proposed [Taskar et al., 2005; Lacoste-Julien et al., 2006].”