Logo Lanfrica

KameniAlexNea/qa_africa

Domaine:

natural language processing

Type de record:

dataset
Créateur:
Kam
Hôte:
African History QA Dataset # African History QA Dataset ## Overview The **African History QA Dataset** is a comprehensive collection of multiple-choice questions focused on various topics related to African history. The dataset was generated using advanced language model capabilities from **LLaMA 3.1: 8B** and consists of questions derived from a series of historical texts and books that cover different eras of African history. This dataset can be used for educational purposes, question-answering tasks, and historical knowledge assessments, and is also a valuable resource for training and evaluating machine learning models. ## Source Material The questions in this dataset were generated from a selection of historical books, covering a wide range of topics and periods in African history. Below is a list of the primary sources used to generate the questions: 1. **Le Cameroun une Afrique en miniature** 2. **Volume I** - Méthodologie et préhistoire africaine 3. **Volume II** - Afrique ancienne 4. **Volume III** - L'Afrique du VIIe au XIe siècle 5. **Volume IV** - L'Afrique du XIIe au XVIe siècle 6. **Volume V** - L'Afrique du XVIe au XVIIe siècle 7. **Volume VI** - Le XIXe siècle jusque vers les années 1880 8. **Volume VII** - L'Afrique sous domination coloniale, 1880-1935 9. **Volume VIII** - L'Afrique depuis 1935 Each book, available in both PDF and TXT formats, was used to extract relevant historical content. This content was then processed with **LLaMA 3.1: 8B** to generate questions. ## Dataset Structure The dataset consists of multiple-choice questions formatted as JSON objects, with the following fields: * **question_number** : A unique identifier for each question. * **question_text** : The main text of the question. * **answer_choices** : A list of possible answers, where each answer is represented by a letter (e.g., "A", "B", "C") and corresponding text. * **correct_answers** : The correct answer(s) for each question, stored as a list to support multiple correct answers when ap …