osunlp/KBQA-Agent

Introduction In traditional knowledge base question answering (KBQA) methods, semantic parsing plays a crucial role. It requires a semantic parser to be extensively trained on a vast dataset of labeled examples, typically consisting of question-answer or question-program pairs. However, the rise of LLMs has shifted this paradigm. LLMs excel in learning from few (or even zero) in-context examples. They utilize natural language as a general vehicle of thought, enabling them to actively navigate… See the full description on the dataset page: https://huggingface.co/datasets/osunlp/KBQA-Agent.