1.we construct a large-scale psychological counseling dialogue dataset, PsychDialog, collected from three Chinese online health community platforms. PsychDialog contains 3,268 multi-turn dialogues between doctors and patients together with a wide range of related metadata including visited hospitals, visited departments, the types of disorders and patient self-reports, which can advance the related research to build accurate and personalized dialogue models.
2.We show some sample datasets for this purpose:
{
"id": 0,
"Hospital": "北京中医药大学东直门医院",
"Department": "心理门诊",
"Disease": "焦虑症",
"Situation": "因体检出现乙肝表面抗体阳性...",
"Dialogue": [
"医生:...",
"病人:...",
"医生:..."
],
"Dia_turn": 2
}
We provide all the code needed for the experiment and store the information to be noted during the experiment in the README.md file under the specific file.