Prompt Guided Copy Mechanism for Conversational Question Answering

Answer-BART[5] uses an end-to-end model to process the question and passage, generating potential Conversational Question Answering (CQA) is a challenging evidence and a natural answer. REAG[6] incorporates the task that aims to generate natural answers for conversational evidence extraction task into the transformer model's encoder flow questions. In this paper, we propose a pluggable approach to improve the natural answer's confidence. Unlike the above for extractive methods that introduces a novel prompt-guided methods, S-net[7] fuses the extraction and generation that it copy mechanism to improve the fluency and appropriateness of first uses the extraction model to collect the passage's mostimportant the extracted answers. Our approach uses prompts to link questions sub-text and then synthesize them into the final answer to answers and employs attention to guide the copy mechanism by the generative model.