Dataset Card for SharcModified

Dataset Summary

ShARC, a conversational QA task, requires a system to answer user questions based on rules expressed in natural language text. However, it is found that in the ShARC dataset there are multiple spurious patterns that could be exploited by neural models. SharcModified is a new dataset which reduces the patterns identified in the original dataset. To reduce the sensitivity of neural models, for each occurence of an instance conforming to any of the patterns, we automatically construct alternatives where we choose to either replace the current instance with an alternative instance which does not exhibit the pattern; or retain the original instance. The modified ShARC has two versions sharc-mod and history-shuffled.

Supported Tasks and Leaderboards

[More Information Needed]

Languages

The dataset is in english (en).

Dataset Structure

Data Instances

Example of one instance:

{
    "annotation": {
        "answer": [
            {
                "paragraph_reference": {
                    "end": 64,
                    "start": 35,
                    "string": "syndactyly affecting the feet"
                },
                "sentence_reference": {
                    "bridge": false,
                    "end": 64,
                    "start": 35,
                    "string": "syndactyly affecting the feet"
                }
            }
        ],
        "explanation_type": "single_sentence",
        "referential_equalities": [
            {
                "question_reference": {
                    "end": 40,
                    "start": 29,
                    "string": "webbed toes"
                },
                "sentence_reference": {
                    "bridge": false,
                    "end": 11,
                    "start": 0,
                    "string": "Webbed toes"
                }
            }
        ],
        "selected_sentence": {
            "end": 67,
            "start": 0,
            "string": "Webbed toes is the common name for syndactyly affecting the feet . "
        }
    },
    "example_id": 9174646170831578919,
    "original_nq_answers": [
        {
            "end": 45,
            "start": 35,
            "string": "syndactyly"
        }
    ],
    "paragraph_text": "Webbed toes is the common name for syndactyly affecting the feet . It is characterised by the fusion of two or more digits of the feet . This is normal in many birds , such as ducks ; amphibians , such as frogs ; and mammals , such as kangaroos . In humans it is considered unusual , occurring in approximately one in 2,000 to 2,500 live births .",
    "question": "what is the medical term for webbed toes",
    "sentence_starts": [
        0,
        67,
        137,
        247
    ],
    "title_text": "Webbed toes",
    "url": "https: //en.wikipedia.org//w/index.php?title=Webbed_toes&oldid=801229780"
}

Data Fields

• example_id: a unique integer identifier that matches up with NQ
• title_text: the title of the wikipedia page containing the paragraph
• url: the url of the wikipedia page containing the paragraph
• question: a natural language question string from NQ
• paragraph_text: a paragraph string from a wikipedia page containing the answer to question
• sentence_starts: a list of integer character offsets indicating the start of sentences in the paragraph
• original_nq_answers: the original short answer spans from NQ
• annotation: the QED annotation, a dictionary with the following items and further elaborated upon below:
  • referential_equalities: a list of dictionaries, one for each referential equality link annotated
  • answer: a list of dictionaries, one for each short answer span
  • selected_sentence: a dictionary representing the annotated sentence in the passage
  • explanation_type: one of "single_sentence", "multi_sentence", or "none"

Data Splits

The dataset is split into training and validation splits.

Dataset Creation

Curation Rationale

[More Information Needed]

Source Data

[More Information Needed]

Initial Data Collection and Normalization

[More Information Needed]

Who are the source language producers?

[More Information Needed]

Annotations

[More Information Needed]

Annotation process

[More Information Needed]

Who are the annotators?

[More Information Needed]

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

[More Information Needed]

Licensing Information

Unknown.

Citation Information

@misc{lamm2020qed,
    title={QED: A Framework and Dataset for Explanations in Question Answering},
    author={Matthew Lamm and Jennimaria Palomaki and Chris Alberti and Daniel Andor and Eunsol Choi and Livio Baldini Soares and Michael Collins},
    year={2020},
    eprint={2009.06354},
    archivePrefix={arXiv},
    primaryClass={cs.CL}
}

Contributions

Thanks to @patil-suraj for adding this dataset.