WWT-QA: A Chinese question answering dataset for the wastewater treatment domain

Under WWT_QA, the policy_qa directory contains train.json, val.json, and test.json with 19,074, 2,119, and 2,355 policy QA samples, respectively. These samples cover key points and interpretations of policies, regulations, and standards. The knowledge_qa directory contains three duplicate files: 12,168, 1,513, and 1,578 knowledge QA samples, respectively. These samples cover high-frequency frontline topics, including process flows, equipment maintenance, and operational control. All samples follow an Alpaca-style instruction-tuning format. Each sample is a JSON object with three fields: instruction, input, and output. Here, instruction represents the question, output corresponds to the answer, and input provides optional supplementary context (set to the empty string "" for all samples in WWT_QA). The dataset can be used for fine-tuning and benchmarking LLMs for wastewater treatment.