James R. Curran
James R. Curran wrote 8 of the papers Mova reads, in 2004–2015, most of them on Dependency parsing, Treebanks, Part-of-speech tagging; often with Jonathan K. Kummerfeld, Matthew Honnibal, Stephen Clark.
Topics: Dependency parsing 1 · Treebanks 1 · Part-of-speech tagging 1
Ideas Mova took (2)
- kept — en/src/parse.rs constraint mode: force the gold head and label for one error class at a time during beam decoding, and report direct and cascaded gains per class, as the paper does; measure on EWT dev (test only for the final numbers). (Identifying Cascading Errors using Constraints in… · en parse-diag constrained beam: forcing punct / core args / function words / root fixes 11.7 / 10.6 / 10.3 / 10.3% of the other errors; coordination 3.4%, PP…)
- kept — en beam-curve coverage: report, for each beam width, the share of sentences where the constrained decoder finds no tree; measure it on EWT dev with the class constraints from I1. (Identifying Cascading Errors using Constraints in… · no-tree count per class at beam 8 on EWT dev: clauses 41, PP 23, coordination 14, punct 12, others <= 6 sentences of 2001; commit 3ba5388e8)
Papers
- Identifying Cascading Errors using Constraints in Dependency Parsing (2015)
2 ideas kept
Parser evaluation counts wrong arcs, but one wrong arc can force several more, so the count hides which error is the root cause. - Parser Showdown at the Wall Street Corral: An Empirical Investigation of Error Types in Parser Output (2012)
A single F-score on WSJ section 23 says nothing about which errors a constituency parser makes. The authors repair each wrong parse into the gold tree with a few tree transformations (subtree movement, node creation, node deletion), and name each repair by its linguistic type: PP attachment, clause attachment… - Parsing Noun Phrases in the Penn Treebank (2011)
The Penn Treebank does not mark the internal structure of noun phrases, so statistical parsers cannot learn it. - Morphological Analysis Can Improve a CCG Parser for English (2010)
The authors test whether English verb morphology, which is poor, still helps a lexicalised parser. In Combinatory Categorial Grammar the verb category often encodes both its argument structure and its inflection, so each verb form gets its own category. The paper splits the inflectional suffix (-s, -ed, -ing and so… - Evaluating a Statistical CCG Parser on Wikipedia (2009)
The authors ask how well a statistical CCG parser (C&C) works on Wikipedia text. They hand-label 200 random sentences, which took about 20 hours. Accuracy is 4.3% lower than on newswire, and the parser is 54% slower. Self-training on Simple English Wikipedia raises accuracy on Wikipedia by 0.8% and narrows the gap to… - Classification of Verb-Particle Constructions with the Google Web1T Corpus (2008)
The paper builds a type-level classifier for verb-particle constructions (VPCs), with no parser and no sentence context. - Wide-Coverage Efficient Statistical Parsing with CCG and Log-Linear Models (2007)
The article gives a full log-linear parsing model for a CCG grammar extracted automatically from CCGbank. Discriminative training needs the incorrect parses of every training sentence, which is feasible only because a Maximum Entropy supertagger keeps the charts small. Estimation needs up to 25 GB of memory and runs… - The Importance of Supertagging for Wide-Coverage CCG Parsing (2004)
The paper shows how a supertagger, which gives each word only a few CCG categories, speeds up a wide-coverage log-linear CCG parser.
Co-authors (8)
- Jonathan K. Kummerfeld ×3
- Matthew Honnibal ×2
- Stephen Clark ×2
- Dan Klein
- David Hall
- David Vadas
- Dominick Ng
- Joel Nothman
This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/authors/james-r-curran/