HQA-Data: A historical question answer generation dataset from previous multi perspective conversation

Abstract

This data article contains a quality assurance dataset for training the chatbot and chat analysis model. This dataset focuses on NLP tasks, as a model that serves and delivers a satisfactory response to a user's query. We obtained data from a well- known dataset known as “The Ubuntu Dialogue Corpus” for the purpose of constructing our dataset. Which consists of about one million multi-turn conversations containing around seven million utterances and one hundred million words. We derived a context for each dialogueID from these lengthy Ubuntu Dialogue Corpus conversations. We have generated a number of questions and answers based on these contexts. All of these questions and answers are contained within the context. This dataset includes 9364 contexts, 36,438 question-answer pairs. In addition to academic research, the dataset may be used for activities such as constructing this QA for another language, deep learning, language interpretation, reading comprehension, and open-domain question answering. We present the data in raw format; it has been open sourced and publicly available at https://data.mendeley.com/datasets/p85z3v45xk.

Keywords

• Natural language processing
• Deep learning
• BERT
• Machine learning
• Question answering generation
• Ubuntu dialogue corpus