Rajakrishnan Rajkumar

Perceptron Reranking for CCG Realization

This paper shows that discriminative reranking with an averaged perceptron model yields substantial improvements in realization quality with CCG. The paper confirms the utility of including language model log probabilities as features in the model, which prior work on discriminative training with log linear models for HPSG realization had called into question. The perceptron model allows the combination of multiple n-gram models to be optimized and then augmented with both syntactic features and discriminative n-gram features. The full model yields a state-of-the-art BLEU score of 0.8506 on Section 23 of the CCGbank, to our knowledge the best score reported to date using a reversible, corpus-engineered grammar.

A More Precise Analysis of Punctuation for Broad-Coverage Surface Realization with CCG

This paper describes a more precise analysis of punctuation for a bi-directional, broad coverage English grammar extracted from the CCGbank (Hockenmaier and Steedman, 2007). We discuss various approaches which have been proposed in the literature to constrain overgeneration with punctuation, and illustrate how aspects of Briscoes (1994) influential approach, which relies on syntactic features to constrain the appearance of balanced and unbalanced commas and dashes to appropriate sentential contexts, is unattractive for CCG. As an interim solution to constrain overgeneration, we propose a rule-based filter which bars illicit sequences of punctuation and cases of improperly unbalanced apposition. Using the OpenCCG toolkit, we demonstrate that our punctuation-augmented grammar yields substantial increases in surface realization coverage and quality, helping to achieve state-of-the-art BLEU scores.

Exploiting Named Entity Classes in CCG Surface Realization

This paper describes how named entity (NE) classes can be used to improve broad coverage surface realization with the OpenCCG realizer. Our experiments indicate that collapsing certain multi-word NEs and interpolating a language model where NEs are replaced by their class labels yields the largest quality increase, with 4-grams adding a small additional boost. Substantial further benefit is obtained by including class information in the hyper-tagging (supertagging for realization) component of the system, yielding a state-of-the-art BLEU score of 0.8173 on Section 23 of the CCGbank. A targeted manual evaluation confirms that the BLEU score increase corresponds to a significant rise in fluency.

Grammar Engineering for CCG using Ant and XSLT

Corpus conversion and grammar extraction have traditionally been portrayed as tasks that are performed once and never again revisited (Burke et al., 2004). We report the successful implementation of an approach to these tasks that facilitates the improvement of grammar engineering as an evolving process. Taking the standard version of the CCGbank (Hockenmaier and Steedman, 2007) as input, our system then introduces greater depth of linguistic insight by augmenting it with attributes the original corpus lacks: Propbank roles and head lexicalization for case-marking prepositions (Boxwell and White, 2008), derivational re-structuring for punctuation analysis (White and Rajkumar, 2008), named entity annotation and lemmatization. Our implementation applies successive XSLT transforms controlled by Apache Ant (http://ant.apache.org/) to an XML translation of this corpus, finally producing an OpenCCG grammar (http://openccg.sourceforge.net/). This design is beneficial to grammar engineering both because of XSLTs unique suitability to performing arbitrary transformations of XML trees and the fine-grained control that Ant provides. The resulting system enables state-of-the-art BLEU scores for surface realization on section 23 of the CCGbank.