Reranking and Self-Training for Parser Adaptation

@inproceedings{McClosky2006RerankingAS,
  title={Reranking and Self-Training for Parser Adaptation},
  author={David McClosky and Eugene Charniak and Mark Johnson},
  booktitle={ACL},
  year={2006}
}
Statistical parsers trained and tested on the Penn Wall Street Journal (WSJ) treebank have shown vast improvements over the last 10 years. Much of this improvement, however, is based upon an ever-increasing number of features to be trained on (typically) the WSJ treebank data. This has led to concern that such parsers may be too finely tuned to this corpus at the expense of portability to other genres. Such worries have merit. The standard "Charniak parser" checks in at a labeled precision… 

Figures and Tables from this paper

Parser Evaluation and the BNC: Evaluating 4 constituency parsers with 3 metrics
TLDR
This work evaluates discriminative parse reranking and parser self-training on a new English test set using four versions of the Charniak parser and a variety of parser evaluation metrics and finds that reranking leads to a performance improvement on the new test set (albeit a modest one).
Improve Chinese Parsing with MaxEnt Reranking Parser
TLDR
After the authors adapted the parser to Chinese with few modifications of language dependent features, the parser worked well overall on Chinese as described in Lian's work, however, there was still room for improvement.
Viterbi Training Improves Unsupervised Dependency Parsing
We show that Viterbi (or "hard") EM is well-suited to unsupervised grammar induction. It is more accurate than standard inside-outside re-estimation (classic EM), significantly faster, and simpler.
Adapting WSJ-Trained Parsers to the British National Corpus using In-Domain Self-Training
TLDR
It is shown that retraining this parser with a combination of one million BNC parse trees (produced by the same parser) and the original WSJ training data yields improvements of 0.4% on WSJ Section 23 and 1.7% on the new BNC gold standard set.
Extending a Parser to Distant Domains Using a Few Dozen Partially Annotated Examples
TLDR
It is shown that recent advances in word representations greatly diminish the need for domain adaptation when the target domain is syntactically similar to the source domain, and a simple way to adapt a parser using only dozens of partial annotations is provided.
Self-Training for Enhancement and Domain Adaptation of Statistical Parsers Trained on Small Datasets
TLDR
This paper uses selftraining in order to improve the quality of a parser and to adapt it to a different domain, using only small amounts of manually annotated seed data, for the first time that self-training with small labeled datasets is applied successfully to these tasks.
Minimally Supervised Domain-Adaptive Parse Reranking for Relation Extraction
TLDR
The paper demonstrates how the generic parser of a minimally supervised information extraction framework can be adapted to a given task and domain for relation extraction (RE) and acquired reranking model improves the performance of RE in both training and test phases with the new first parses.
The Berkeley Parser at the EVALITA 2009 Constituency Parsing Task
TLDR
The Berkeley Parser is used, obtaining the best F1, that is 78:73.85% with respect to the best result obtained at EVALITA 2007 by the Bikel's parser.
Self-Training Tree Substitution Grammars for Domain Adaptation
TLDR
A good parsing model should be generalizable; that is, it should be applicable to multiple languages, as well as multiple domains within each language, and should be able to parse other datasets as well.
Cross-Domain Effects on Parse Selection for Precision Grammars
TLDR
It is found it is possible to considerably improve parse selection accuracy through construction of even small-scale in- domain treebanks, and learning of parse selection models over in-domain and out-of-domain data, and more sophisticated strategies for combining data from these sources to train models are investigated.
...
...

References

SHOWING 1-10 OF 21 REFERENCES
Discriminative Reranking for Natural Language Parsing
TLDR
The boosting approach to ranking problems described in Freund et al. (1998) is applied to parsing the Wall Street Journal treebank, and it is argued that the method is an appealing alternative-in terms of both simplicity and efficiency-to work on feature selection methods within log-linear (maximum-entropy) models.
Evaluating and Integrating Treebank Parsers on a Biomedical Corpus
TLDR
Inital experiments with unsupervised parse combination techniques showed that integrating the output of several parsers can ameliorate some of the performance problems they encounter on unfamiliar text, providing accuracy and coverage improvements, and a novel measure of trustworthiness.
Corpus Variation and Parser Performance
TLDR
This work examines how other types of text might a ect parser performance, and how portable parsing models are across corpora by comparing results for the Brown and WSJ corpora, and considers which parts of the parser's probability model are particularly tuned to the corpus on which it was trained.
Learning to Parse Natural Language with Maximum Entropy Models
TLDR
A machine learning system for parsing natural language that learns from manually parsed example sentences, and parses unseen data at state-of-the-art accuracies, and it is demonstrated that the parser can train from other domains without modification to the modeling framework or the linguistic hints it uses to learn.
Parsing Biomedical Literature
TLDR
It is shown how existing domain-specific lexical resources may be leveraged to augment PTB-training: part-of-speech tags, dictionary collocations, and named-entities, without requiring in-domain treebank data.
Supervised Grammar Induction using Training Data with Limited Constituent Information
TLDR
It is shown that the most informative linguistic constituents are the higher nodes in the parse trees, typically denoting complex noun phrases and sentential clauses, and an adaptation strategy is proposed, which produces grammars that parse almost as well as Grammars induced from fully labeled corpora.
Coarse-to-Fine n-Best Parsing and MaxEnt Discriminative Reranking
TLDR
This paper describes a simple yet novel method for constructing sets of 50- best parses based on a coarse-to-fine generative parser that generates 50-best lists that are of substantially higher quality than previously obtainable.
Effective Self-Training for Parsing
We present a simple, but surprisingly effective, method of self-training a two-phase parser-reranker system using readily available unlabeled data. We show that this type of bootstrapping is possible
Bootstrapping statistical parsers from small datasets
TLDR
Experimental results show that unlabelled sentences can be used to improve the performance of statistical parsers and it is shown that boot-strapping continues to be useful, even though no manually produced parses from the target domain are used.
Building a Large Annotated Corpus of English: The Penn Treebank
TLDR
As a result of this grant, the researchers have now published on CDROM a corpus of over 4 million words of running text annotated with part-of- speech (POS) tags, which includes a fully hand-parsed version of the classic Brown corpus.
...
...