mmteb-miracl-reranking.py

# coding=utf-8
# Copyright 2020 The HuggingFace Datasets Authors and the current dataset script contributor.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

# Lint as: python3

import json
import gzip
import datasets
from collections import defaultdict
from dataclasses import dataclass

_CITATION = """
"""

surprise_languages = ["de", "yo"]
new_languages = ["es", "fa", "fr", "hi", "zh"] + surprise_languages
languages = ["ar", "bn", "en", "es", "fa", "fi", "fr", "hi", "id", "ja", "ko", "ru", "sw", "te", "th", "zh"] + surprise_languages

_DESCRIPTION = "dataset load script for MIRACL"


def get_first_stage_runfile(lang):
    first_stages = [
        "bm25", "mdpr", "hybrid",
    ]
    return {
        first_stage: f"https://huggingface.co/datasets/miracl/miracl-reranking/resolve/main/data/{first_stage}/{lang}.gz" for first_stage in first_stages
    }


_DATASET_URLS = {
    lang: {
        "dev": {
            "topics": f"https://huggingface.co/datasets/miracl/miracl/resolve/main/miracl-v1.0-{lang}/topics/topics.miracl-v1.0-{lang}-dev.tsv",
            "qrels": f"https://huggingface.co/datasets/miracl/miracl/resolve/main/miracl-v1.0-{lang}/qrels/qrels.miracl-v1.0-{lang}-dev.tsv",
            **get_first_stage_runfile(lang),
        },
    } for lang in languages
}


def load_topic(fn):
    qid2topic = {}
    with open(fn, encoding="utf-8") as f:
        for line in f:
            qid, topic = line.strip().split("\t")
            qid2topic[qid] = topic
    return qid2topic


def load_qrels(fn):
    if fn is None:
        return None

    qrels = defaultdict(dict)
    with open(fn, encoding="utf-8") as f:
        for line in f:
            qid, _, docid, rel = line.strip().split("\t")
            qrels[qid][docid] = int(rel)
    return qrels


def load_runfile(fn, topk=100):
    file_handle = gzip.open(fn, "rb") if fn.endswith(".gz") else open(fn, "r")
    runs = defaultdict(dict)
    for line in file_handle:
        if not isinstance(line, str):
            line = line.decode()
        qid, _, docid, _, score, _ = line.strip().split()
        runs[qid][docid] = float(score)

    if topk > 0:
        for qid in runs:
            runs[qid] = dict(sorted(
                runs[qid].items(),
                key=lambda doc_score: doc_score[1],
                reverse=True,
            )[:topk])
    return runs


class MIRACLReranking(datasets.GeneratorBasedBuilder):
    BUILDER_CONFIGS = [datasets.BuilderConfig(
            version=datasets.Version("1.0.0"),
            name=lang, description=f"MIRACL Reranking in language {lang}."
        ) for lang in languages
    ]

    def _info(self):
        features = datasets.Features(
            query=datasets.Value("string"),
            positive=datasets.Sequence(datasets.Value("string")),
            negative=datasets.Sequence(datasets.Value("string")),
            candidates=datasets.Sequence(datasets.Value("string")),
        )

        return datasets.DatasetInfo(
            # This is the description that will appear on the datasets page.
            description=_DESCRIPTION,
            # This defines the different columns of the dataset and their types
            features=features,  # Here we define them above because they are different between the two configurations
            supervised_keys=None,
            # Homepage of the dataset for documentation
            homepage="https://project-miracl.github.io",
            # License for the dataset if available
            license="",
            # Citation for the dataset
            citation=_CITATION,
        )

    def _split_generators(self, dl_manager):
        lang = self.config.name
        downloaded_files = dl_manager.download_and_extract(_DATASET_URLS[lang]["dev"])

        splits = [
            datasets.SplitGenerator(
                name="dev",
                gen_kwargs={
                    "filepaths": downloaded_files,
                },
            ),
        ]

        return splits

    def _generate_examples(self, filepaths):
        def formulate_doc(title, text):
            return f"{title} {text}"

        lang = self.config.name
        miracl_corpus = datasets.load_dataset("miracl/miracl-corpus", lang)["train"]
        docid2doc = {doc["docid"]: formulate_doc(doc["title"], doc["text"]) for doc in miracl_corpus}

        topic_fn = filepaths["topics"]
        qrel_fn = filepaths["qrels"]
        runfile = filepaths["bm25"]

        qid2topic = load_topic(topic_fn)
        qrels = load_qrels(qrel_fn)
        runs = load_runfile(runfile, topk=100)
        for qid in qid2topic:
            data = {}

            pos_docids = [docid for docid, rel in qrels[qid].items() if rel == 1] if qrels is not None else []
            neg_docids = [docid for docid, rel in qrels[qid].items() if rel == 0] if qrels is not None else []

            data["query"] = qid2topic[qid]
            data["positive"] = [docid2doc[docid] for docid in pos_docids]
            data["negative"] = [docid2doc[docid] for docid in neg_docids]
            data["candidates"] = [docid2doc[docid] for docid in runs[qid]]

            yield qid, data