2021/03/13 by Endri Kacupaj, Kacupaj, Endri, Barshana Banerjee +5 · 2 citations
Computer Science · #Computation and Language (cs.CL) #FOS: Computer and information sciences #Natural Language Processing Techniques #Speech and dialogue systems #Topic Modeling
paper · pdf · doi:10.48550/arxiv.2103.07771
openalex publication_date 2021/03/13 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
This paper presents ParaQA, a question answering (QA) dataset with multiple\nparaphrased responses for single-turn conversation over knowledge graphs (KG).\nThe dataset was created using a semi-automated framework for generating diverse\nparaphrasing of the answers using techniques such as back-translation. The\nexisting datasets for conversational question answering over KGs\n(single-turn/multi-turn) focus on question paraphrasing and provide only up to\none answer verbalization. However, ParaQA contains 5000 question-answer pairs\nwith a minimum of two and a maximum of eight unique paraphrased responses for\neach question. We complement the dataset with baseline models and illustrate\nthe advantage of having multiple paraphrased answers through commonly used\nmetrics such as BLEU and METEOR. The ParaQA dataset is publicly available on a\npersistent URI for broader usage and adaptation in the research community.\n