Dataset Card for Grail QA

Dataset Summary

What is GrailQA?

Strongly Generalizable Question Answering (GrailQA) is a new large-scale, high-quality dataset for question answering on knowledge bases (KBQA) on Freebase with 64,331 questions annotated with both answers and corresponding logical forms in different syntax (i.e., SPARQL, S-expression, etc.). It can be used to test three levels of generalization in KBQA: i.i.d., compositional, and zero-shot.

Why GrailQA?

GrailQA is by far the largest crowdsourced KBQA dataset with questions of high diversity (i.e., questions in GrailQA can have up to 4 relations and optionally have a function from counting, superlatives and comparatives). It also has the highest coverage over Freebase; it widely covers 3,720 relations and 86 domains from Freebase. Last but not least, our meticulous data split allows GrailQA to test not only i.i.d. generalization, but also compositional generalization and zero-shot generalization, which are critical for practical KBQA systems.

Supported Tasks and Leaderboards

[More Information Needed]

Languages

English and Graph query

Dataset Structure

Data Instances

[More Information Needed]

Data Fields

qid (str)
question (str)
answer (List): Defaults to [] in test split.
- answer_type (str)
- answer_argument (str)
- entity_name (str): Defauts to "" if answer_type is not Entity.
function (string): Defaults to "" in test split.
num_node (int): Defaults to -1 in test split.
num_edge (int): Defaults to -1 in test split.
graph_query (Dict)
- nodes (List): Defaults to [] in test split.
  - nid (int)
  - node_type (str)
  - id (str)
  - class (str)
  - friendly_name (str)
  - question_node (int)
  - function (str)
- edges (List): Defaults to [] in test split.
  - start (int)
  - end (int)
  - relation (str)
  - friendly_name (str)
sqarql_query (str): Defaults to "" in test split.
domains (List[str]): Defaults to [] in test split.
level (str): Only available in validation split. Defaults to "" in others.
s_expression (str): Defaults to "" in test split.

Notes: Only qid and question available in test split.

Data Splits

Dataset Split	Number of Instances in Split
Train	44,337
Validation	6,763
Test	13,231

Dataset Creation

Curation Rationale

[More Information Needed]

Source Data

Initial Data Collection and Normalization

[More Information Needed]

Who are the source language producers?

[More Information Needed]

Annotations

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

[More Information Needed]

Citation Information

[More Information Needed]

Contributions

Thanks to @mattbui for adding this dataset.