trec-cast_v1_2020.py

"""
""" # TODO
try:
    import ir_datasets
except ImportError as e:
    raise ImportError('ir-datasets package missing; `pip install ir-datasets`')
import datasets

IRDS_ID = 'trec-cast/v1/2020'
IRDS_ENTITY_TYPES = {'queries': {'query_id': 'string', 'raw_utterance': 'string', 'automatic_rewritten_utterance': 'string', 'manual_rewritten_utterance': 'string', 'manual_canonical_result_id': 'string', 'topic_number': 'int64', 'turn_number': 'int64'}, 'qrels': {'query_id': 'string', 'doc_id': 'string', 'relevance': 'int64', 'iteration': 'string'}}

_CITATION = '@inproceedings{Dalton2020Cast,\n  title={CAsT 2020: The Conversational Assistance Track Overview},\n  author={Jeffrey Dalton and Chenyan Xiong and Jamie Callan},\n  booktitle={TREC},\n  year={2020}\n}'

_DESCRIPTION = "" # TODO

class trec_cast_v1_2020(datasets.GeneratorBasedBuilder):
    BUILDER_CONFIGS = [datasets.BuilderConfig(name=e) for e in IRDS_ENTITY_TYPES]

    def _info(self):
        return datasets.DatasetInfo(
            description=_DESCRIPTION,
            features=datasets.Features({k: datasets.Value(v) for k, v in IRDS_ENTITY_TYPES[self.config.name].items()}),
            homepage=f"https://ir-datasets.com/trec-cast#trec-cast/v1/2020",
            citation=_CITATION,
        )

    def _split_generators(self, dl_manager):
        return [datasets.SplitGenerator(name=self.config.name)]

    def _generate_examples(self):
        dataset = ir_datasets.load(IRDS_ID)
        for i, item in enumerate(getattr(dataset, self.config.name)):
            key = i
            if self.config.name == 'docs':
                key = item.doc_id
            elif self.config.name == 'queries':
                key = item.query_id
            yield key, item._asdict()

    def as_dataset(self, split=None, *args, **kwargs):
        split = self.config.name # always return split corresponding with this config to avid returning a redundant DatasetDict layer
        return super().as_dataset(split, *args, **kwargs)