morehopqa.py

import datasets
from datasets import load_dataset
import json
import os

class MoreHopQAConfig(datasets.BuilderConfig):
    """BuilderConfig for MoreHopQA."""
    def __init__(self, data_path, **kwargs):
        """BuilderConfig for MoreHopQA.
        Args:
            data_path: string, path to the data files containing the dataset.
            **kwargs: keyword arguments forwarded to super.
        """
        super(MoreHopQAConfig, self).__init__(**kwargs)
        self.data_path = data_path

class MoreHopQA(datasets.GeneratorBasedBuilder):
    """MoreHopQA: A dataset for multi-hop question answering."""
    BUILDER_CONFIG_CLASS = MoreHopQAConfig

    BUILDER_CONFIGS = [
        MoreHopQAConfig(
            name="verified",
            version=datasets.Version("1.0.0", ""),
            description="MoreHopQA: A dataset for multi-hop question answering.",
            data_path="data/with_human_verification.json",
        ),
        MoreHopQAConfig(
            name="unverified",
            version=datasets.Version("1.0.0", ""),
            description="MoreHopQA: A dataset for multi-hop question answering.",
            data_path="data/without_human_verification.json",
        ),
    ]

    DEFAULT_CONFIG_NAME = "verified"

    def _info(self):
        return datasets.DatasetInfo(
            features=datasets.Features({
                "question": datasets.Value("string"),
                "context": datasets.Sequence({
                    "title": datasets.Value("string"),
                    "paragraphs": datasets.Sequence(datasets.Value("string"))
                }),
                "answer": datasets.Value("string"),
                "previous_question": datasets.Value("string"),
                "previous_answer": datasets.Value("string"),
                "question_decomposition": datasets.Sequence({
                    "sub_id": datasets.Value("string"),
                    "question": datasets.Value("string"),
                    "answer": datasets.Value("string"),
                    "paragraph_support_title": datasets.Value("string")
                }),
                "question_on_last_hop": datasets.Value("string"),
                "answer_type": datasets.Value("string"),
                "previous_answer_type": datasets.Value("string"),
                "no_of_hops": datasets.Value("int32"),
                "reasoning_type": datasets.Value("string"),
            }),
        )

    def _split_generators(self, dl_manager):
        """Returns SplitGenerators."""
        return [
            datasets.SplitGenerator(
                name=datasets.Split.TEST,
                gen_kwargs={"filepath": os.path.join(self.config.data_path)}
            )
        ]

    def _generate_examples(self, filepath):
        """Yields examples."""
        with open(filepath, encoding="utf-8") as f:
            data = json.load(f)
            for item in data:
                yield item['_id'], {
                    "question": item["question"],
                "context": [{
                        "title": subitem[0],  # title is the first item in the sublist
                        "content": subitem[1]  # paragraphs are the second item
                    } for subitem in item.get("context", [])],
                "answer": item["answer"],
                "previous_question": item["previous_question"],
                "previous_answer": item["previous_answer"],
                "question_decomposition": [{
                    "sub_id": subitem["sub_id"],
                    "question": subitem["question"],
                    "answer": subitem["answer"],
                    "paragraph_support_title": subitem["paragraph_support_title"]
                } for subitem in item["question_decomposition"]],
                "question_on_last_hop": item["ques_on_last_hop"],
                "answer_type": item["answer_type"],
                "previous_answer_type":item["previous_answer_type"],
                "no_of_hops": item["no_of_hops"],
                "reasoning_type": item["reasoning_type"],
                }