Dataset Card for QA-SRL

Dataset Summary

we model predicate-argument structure of a sentence with a set of question-answer pairs. our method allows practical large-scale annotation of training data. We focus on semantic rather than syntactic annotation, and introduce a scalable method for gathering data that allows both training and evaluation.

Supported Tasks and Leaderboards

[More Information Needed]

Languages

This dataset is in english language.

Dataset Structure

Data Instances

We use question-answer pairs to model verbal predicate-argument structure. The questions start with wh-words (Who, What, Where, What, etc.) and contains a verb predicate in the sentence; the answers are phrases in the sentence. For example:

UCD finished the 2006 championship as Dublin champions , by beating St Vincents in the final .

Data Fields

Annotations provided are as follows:

• sentence: contains tokenized sentence
• sent_id: is the sentence identifier
• predicate_idx:the index of the predicate (its position in the sentence)
• predicate: the predicate token
• question: contains the question which is a list of tokens. The question always consists of seven slots, as defined in the paper. The empty slots are represented with a marker “_”. The question ends with question mark.
• answer: list of answers to the question

Data Splits

Please note This dataset only has wikipedia data. Newswire dataset needs CoNLL-2009 English training data to get the complete data. This training data is under license. Thus, newswire dataset is not included in this data.

Dataset Creation

Curation Rationale

[More Information Needed]

Source Data

Initial Data Collection and Normalization

We annotated over 3000 sentences (nearly 8,000 verbs) in total across two domains: newswire (PropBank) and Wikipedia.

Who are the source language producers?

[More Information Needed]

Annotations

Annotation process

non-expert annotators were given a short tutorial and a small set of sample annotations (about 10 sentences). Annotators were hired if they showed good understanding of English and the task. The entire screening process usually took less than 2 hours.

Who are the annotators?

10 part-time, non-exper annotators from Upwork (Previously oDesk)

Personal and Sensitive Information

[More Information Needed]

Considerations for Using the Data

Social Impact of Dataset

[More Information Needed]

Discussion of Biases

[More Information Needed]

Other Known Limitations

[More Information Needed]

Additional Information

Dataset Curators

Luheng He

Licensing Information

[More Information Needed]

Citation Information

@InProceedings{huggingface:dataset,
title = {QA-SRL: Question-Answer Driven Semantic Role Labeling},
authors={Luheng He, Mike Lewis, Luke Zettlemoyer},
year={2015}
publisher = {cs.washington.edu},
howpublished={\\url{https://dada.cs.washington.edu/qasrl/#page-top}},
}

Contributions

Thanks to @bpatidar for adding this dataset.