"Learning and Inference for Hierarchically Split PCFGs"

2004

Abstract This paper describes a method for conducting evaluations of Treebank and non-Treebank parsers alike against the English language U. Penn Treebank (Marcus et al., 1993) using a metric that focuses on the accuracy of relatively non-controversial aspects of parse structure. Our conjecture is that if we focus on maximal projections of heads (MPH), we are likely to find much broader agreement than if we try to evaluate based on order of attachment.

2008

Recent parsing research has started addressing the questions a) how parsers trained on different syntactic resources differ in their performance and b) how to conduct a meaningful evaluation of the parsing results across such a range of syntactic representations. Two German treebanks, Negra and TüBa-D/Z, constitute an interesting testing ground for such research given that the two treebanks make very different representational choices for this language, which also is of general interest given that German is situated between the extremes of fixed and free word order. We show that previous work comparing PCFG parsing with these two treebanks employed PARSEVAL and grammatical function comparisons which were skewed by differences between the two corpus annotation schemes. Focusing on the grammatical dependency triples as an essential dimension of comparison, we show that the two very distinct corpora result in comparable parsing performance.

Computational Linguistics, 2007

This article presents an algorithm for translating the Penn Treebank into a corpus of Combinatory Categorial Grammar (CCG) derivations augmented with local and long-range word-word dependencies. The resulting corpus, CCGbank, includes 99.4% of the sentences in the Penn Treebank. It is available from the Linguistic Data Consortium, and has been used to train wide-coverage statistical parsers that obtain state-of-the-art rates of dependency recovery. In order to obtain linguistically adequate CCG analyses, and to eliminate noise and inconsistencies in the original annotation, an extensive analysis of the constructions and annotations in the Penn Treebank was called for, and a substantial number of changes to the Treebank were necessary. We discuss the implications of our findings for the extraction of other linguistically expressive grammars from the Treebank, and for the design of future treebanks.

Proceedings of the Fifth Workshop on …, 2006

Proceedings of the 23rd …, 2010

In this paper, we introduce our recent work on Chinese HPSG grammar development through treebank conversion. By manually defining grammatical constraints and annotation rules, we convert the bracketing trees in the Penn Chinese Treebank (CTB) to be an HPSG treebank. Then, a large-scale lexicon is automatically extracted from the HPSG treebank. Experimental results on the CTB 6.0 show that a HPSG lexicon was successfully extracted with 97.24% accuracy; furthermore, the obtained lexicon achieved 98.51% lexical coverage and 76.51% sentential coverage for unseen text, which are comparable to the state-of-the-art works for English.

Proceedings of the ACL-02 conference on Empirical methods in natural language processing - EMNLP '02, 2002

We describe an LR parser of parts-ofspeech (and punctuation labels) for Tree Adjoining Grammars (TAGs), that solves table conflicts in a greedy way, with limited amount of backtracking. We evaluate the parser using the Penn Treebank showing that the method yield very fast parsers with at least reasonable accuracy, confirming the intuition that LR parsing benefits from the use of rich grammars. 1 Unlike (Wright and Wrigley, 1991)'s approach who tries to transpose PCFG probabilities to LR tables, facing difficulties which, to the best of our knowledge, have not been yet solved to content (cf. also (Ng and Tomita, 1991; Wright et al., 1991; Abney et al., 1999)).

Proceedings of the 36th annual meeting on Association for Computational Linguistics -, 1998

Treebanks, such as the Penn Treebank (PTB), a ord a simple approach to obtaining a broad coverage grammar: simply read the grammar o the parse trees in the treebank. While such a grammar is easy to obtain, an observed linear rate of growth of rule set versus corpus size suggests, at least in the case of the PTB, that the derived grammar is far from complete and that much more treebanked text could be required to obtain one. However, we o er an alternative explanation in terms of underspeci cation of structures within the treebank. This hypothesis is explored by applying an algorithm to compact the derived grammar by eliminating redundant rules { rules whose right hand sides can be parsed by other rules. The size of the resulting compacted grammar, which is signi cantly less than that of the full treebank grammar, is shown to approach a limit.

We modify the PCFG language model by introducing a mixture over grammars. We then investigate the performance of such a mixture and analyze what information is gained by its introduction. We examine two different types of parsers and find optimal parameters (number of grammars, training iterations, etc.) for our language model.

We introduce LTAG-spinal, a novel variant of traditional Lexicalized Tree Adjoining Grammar (LTAG) with desirable linguistic, computational and statistical properties. Unlike in traditional LTAG, subcategorization frames and the argument-adjunct distinction are left underspecified in LTAG-spinal. LTAG-spinal with adjunction constraints is weakly equivalent to LTAG. The LTAG-spinal formalism is used to extract an LTAG-spinal Treebank from the Penn Treebank with Propbank annotation. Based on Propbank annotation, predicate coordination and LTAG adjunction structures are successfully extracted. The LTAG-spinal Treebank makes explicit semantic relations that are implicit or absent from the original PTB. LTAG-spinal provides a very desirable resource for statistical LTAG parsing, incremental parsing, dependency parsing, and semantic parsing. This treebank has been successfully used to train an incremental LTAG-spinal parser and a bidirectional LTAG dependency parser.

We describe experiments on learning latent variable grammars for various German treebanks, using a language-agnostic statistical approach. In our method, a minimal initial grammar is hierarchically refined using an adaptive split-and-merge EM procedure, giving compact, accurate grammars. The learning procedure directly maximizes the likelihood of the training treebank, without the use of any language specific or linguistically constrained features. Nonetheless, the resulting grammars encode many linguistically interpretable patterns and give the best published parsing accuracies on three German treebanks.