Data Sets

Noun Verb

This dataset contains naturally-occurring English sentences that feature non-trivial noun-verb ambiguity.

Title Noun Verb
Overview Noun Verb Dataset This dataset contains naturally-occurring English sentences that feature non-trivial noun-verb ambiguity. Motivation English part-of-speech taggers regularly make egregious errors related to noun-verb ambiguity, despite having achieved 97%+ accuracy on the WSJ Penn Treebank since 2002. These mistakes have been difficult to quantify and make taggers less useful to downstream tasks such as translation and text-to-speech synthesis. Below are some examples from the dataset: Certain insects can damage plumerias, such as mites, flies, or aphids. NOUN Mark which area you want to distress. VERB All tested existing part-of-speech taggers mistag both of these examples, tagging flies as a verb and Mark as a noun. Description The dataset contains sentences in CoNLL format. Each sentence has a single token that has been manually annotated as either VERB or NON-VERB. The sentences come from multiple domains. Where applicable, the url of the source page for the sentence is included in a comment line before the sentence. The dataset is split into Train/Dev/Test sections. For Dev and Test sections the annotations included either VERB or NON-VERB in XPOS, UPOS and FEATS columns. For the Train section, XPOS and UPOS columns are replaced with a (predicted) fine POS tag obtained by running automatic tagger and selecting the top tag that matched the gold coarse-grained VERB/NON-VERB label. The number of examples are shown below: Data Split Train Dev Test Total 23458 2367 5907 These numbers are slightly lower than those reported in Table 5 of the paper by 796, 33, and 93 examples, respectively.
Tags
  • DataChallenges
  • Education
  • File Size
    Download

    © copyright 2017 www.aimlmarketplace.com. All Rights Reserved.

    A Product of HunterTech Ventures