update

.gitignore

@@ -0,0 +1 @@
+SubjQA

process.py

@@ -0,0 +1,114 @@
+""" Script to process raw SQuAD file for Question Generation format
+You need to run `python -m spacy download en_core_web_sm`.
+Split when uploading to dataset hub by
+```
+gsplit -l 600 -d --additional-suffix=.jsonl test.jsonl test
+gsplit -l 600 -d --additional-suffix=.jsonl train.jsonl train
+gsplit -l 600 -d --additional-suffix=.jsonl valid.jsonl valid
+```
+"""
+import json
+import os
+import re
+from tqdm import tqdm
+from itertools import chain
+
+import pandas as pd
+import spacy
+
+
+SPLITTER = spacy.load('en_core_web_sm')
+HIGHLIGHT_TOKEN = '<hl>'
+
+
+def get_sentence(document: str):
+    return [str(s) for s in SPLITTER(document).sents]
+
+
+def process_single_data(question, paragraph, answer):
+    """ Convert single raw json data into QG format """
+    example = {'question': question, 'paragraph': paragraph, 'answer': answer}
+    start = example['paragraph'].find(example['answer'])
+    end = start + len(answer)
+    assert paragraph[start:end] == answer
+    # get sentence
+    before_tmp = get_sentence(example['paragraph'][:start])
+    if len(before_tmp) == 0:
+        before = ''
+        before_sentence = ''
+    else:
+        if before_tmp[-1].endswith('.'):
+            before = ' '.join(before_tmp)
+            before_sentence = ''
+        else:
+            before = ' '.join(before_tmp[:-1])
+            before_sentence = before_tmp[-1]
+            before_sentence = before_sentence if before_sentence.endswith(' ') else '{} '.format(before_sentence)
+    after_tmp = get_sentence(example['paragraph'][start + len(example['answer']):])
+    if len(after_tmp) == 0:
+        after = ''
+        after_sentence = ''
+    else:
+        after = ' '.join(after_tmp[1:])
+        after_sentence = after_tmp[0]
+        after_sentence = after_sentence if after_sentence.startswith(' ') else ' {}'.format(after_sentence)
+    example['sentence'] = '{}{}{}'.format(before_sentence, example['answer'], after_sentence)
+
+    # get paragraph_sentence
+    before = '' if before == '' else '{} '.format(before)
+    after = '' if after == '' else ' {}'.format(after)
+    source_text = '{0}{1} {2} {1}{3}'.format(before, HIGHLIGHT_TOKEN, example['sentence'], after)
+    example['paragraph_sentence'] = re.sub(r'\s+', ' ', source_text)
+
+    # get paragraph_answer
+    source_text = '{0}{1} {2} {1}{3}'.format(
+        example['paragraph'][:start], HIGHLIGHT_TOKEN, example['answer'],
+        example['paragraph'][start + len(example['answer']):])
+    example['paragraph_answer'] = re.sub(r'\s+', ' ', source_text)
+
+    # get sentence_answer
+    if len(before_tmp) == 0 or before_tmp[-1].endswith('.'):
+        before = ''
+    else:
+        before = before_tmp[-1] if before_tmp[-1].endswith(' ') else '{} '.format(before_tmp[-1])
+    if len(after_tmp) == 0:
+        after = ''
+    else:
+        after = after_tmp[0] if after_tmp[0].startswith(' ') else ' {}'.format(after_tmp[0])
+    source_text = '{0}{1} {2} {1}{3}'.format(before, HIGHLIGHT_TOKEN, example['answer'], after)
+    example['sentence_answer'] = re.sub(r'\s+', ' ', source_text)
+
+    return example
+
+
+if __name__ == '__main__':
+    os.makedirs('./data/processed', exist_ok=True)
+    for i in ["books", "electronics", "grocery", "movies", "restaurants", "tripadvisor"]:
+        for s in ["dev.csv", "test.csv", "train.csv"]:
+            df = pd.read_csv(f'SubjQA/SubjQA/{i}/splits/{s}')
+            df = df[[x != 'ANSWERNOTFOUND' for x in df['human_ans_spans']]]
+            df['review'] = [x.replace('ANSWERNOTFOUND', '') for x in df['review']]
+            output = []
+            for _, _g in df.groupby('q_review_id'):
+                if any(i == 'ANSWERNOTFOUND' for i in _g['human_ans_spans']):
+                    continue
+                # if len(_g["human_ans_spans"].unique()) != 1:
+                #     continue
+                # _df = _g.iloc[0]
+                _len = [len(i) for i in _g["human_ans_spans"]]
+                _df = _g.iloc[_len.index(max(_len))]
+                start, end = eval(_df['human_ans_indices'])
+                # if re.sub(r'[\s\W]', '', _df['review'][start:end]) != re.sub(r'[\s\W]', '', _df["human_ans_spans"]):
+                #     input(f"{_df['review'][start:end]} != {_df['human_ans_spans']}")
+                #     continue
+                out = process_single_data(question=re.sub(r'\s+\?', '?', _df['question']),
+                                          answer=_df['review'][start:end],
+                                          paragraph=_df['review'])
+                out['question_subj_level'] = int(_df['question_subj_level'])
+                out['answer_subj_level'] = int(_df['answer_subj_level'])
+                out['paragraph_id'] = _df['review_id']
+                output.append(out)
+            with open(f'./data/processed/{i}.{s.replace(".csv", ".jsonl")}', 'w') as f:
+                f.write('\n'.join([json.dumps(i) for i in output]))
+
+

qg_subjqa.py

@@ -0,0 +1,69 @@
+import json
+import datasets
+
+logger = datasets.logging.get_logger(__name__)
+_DESCRIPTION = """[SubjQA](https://github.com/megagonlabs/SubjQA) dataset for question generation (QG) task."""
+_URL = 'https://huggingface.co/datasets/asahi417/qg_subjqa/raw/main/data/processed'
+_DOMAINS = ["books", "electronics", "grocery", "movies", "restaurants", "tripadvisor"]
+
+
+class QGSubjQAConfig(datasets.BuilderConfig):
+    """BuilderConfig for SquadQG"""
+
+    def __init__(self, **kwargs):
+        """BuilderConfig for SquadQG.
+        Args:
+          **kwargs: keyword arguments forwarded to super.
+        """
+        super(QGSubjQAConfig, self).__init__(**kwargs)
+
+
+class QGSubjQA(datasets.GeneratorBasedBuilder):
+
+    BUILDER_CONFIGS = [QGSubjQAConfig(name=i, description="SubjQA from domain of `{}`.".format(i)) for i in _DOMAINS]
+
+    def _info(self):
+        return datasets.DatasetInfo(
+            description=_DESCRIPTION,
+            features=datasets.Features(
+                {
+                    "answer": datasets.Value("string"),
+                    "question": datasets.Value("string"),
+                    "sentence": datasets.Value("string"),
+                    "paragraph": datasets.Value("string"),
+                    "sentence_answer": datasets.Value("string"),
+                    "paragraph_answer": datasets.Value("string"),
+                    "paragraph_sentence": datasets.Value("string"),
+                    "paragraph_id": datasets.Value("string"),
+                    "question_subj_level": datasets.Value("int32"),
+                    "answer_subj_level": datasets.Value("int32")
+                }
+            ),
+            supervised_keys=None,
+            homepage="https://github.com/asahi417/lm-question-generation"
+        )
+
+    def _split_generators(self, dl_manager):
+        downloaded_file = dl_manager.download_and_extract({
+            'train': f"{_URL}/{self.config.name}.train.jsonl",
+            'dev': f"{_URL}/{self.config.name}.dev.jsonl",
+            'test': f"{_URL}/{self.config.name}.test.jsonl"
+        })
+        return [
+            datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={"filepaths": downloaded_file["train"]}),
+            datasets.SplitGenerator(name=datasets.Split.VALIDATION, gen_kwargs={"filepaths": downloaded_file["dev"]}),
+            datasets.SplitGenerator(name=datasets.Split.TEST, gen_kwargs={"filepaths": downloaded_file["test"]}),
+        ]
+
+    def _generate_examples(self, filepaths):
+        _key = 0
+        for filepath in filepaths:
+            logger.info("generating examples from = %s", filepath)
+            with open(filepath, encoding="utf-8") as f:
+                _list = f.read().split('\n')
+                if _list[-1] == '':
+                    _list = _list[:-1]
+                for i in _list:
+                    data = json.loads(i)
+                    yield _key, data
+                    _key += 1