| import pandas as pd | |
| import random | |
| RANDOM_SEED = 1510 | |
| random.seed(RANDOM_SEED) | |
| input_file = "passages_and_queries.csv" | |
| output_file = "query_and_100_passages.csv" | |
| data = pd.read_csv(input_file) | |
| output_data = [] | |
| for idx, row in data.iterrows(): | |
| correct_passage = row['Text'] | |
| query = row['Generated_Query'] | |
| all_passages = data['Text'].tolist() | |
| all_passages.remove(correct_passage) | |
| distractors = random.sample(all_passages, 99) | |
| passages = distractors + [correct_passage] | |
| random.shuffle(passages) | |
| correct_index = passages.index(correct_passage) + 1 | |
| new_row = { | |
| "id": idx + 1, | |
| "query": query, | |
| "correct_passage_index": correct_index | |
| } | |
| for i, passage in enumerate(passages): | |
| new_row[f"passage_{i+1}"] = passage | |
| output_data.append(new_row) | |
| output_df = pd.DataFrame(output_data) | |
| output_df.to_csv(output_file, index=False) | |