thor_cot_schema.json

{
  "name": "thor-cot-schema",
  "schema": [
    {"prompt": "Given the sentence '{sentence}'. Which specific aspect of '{entity}' is possibly mentioned?", "out": "r_aspect"},
    {"prompt": "Given the sentence '{sentence}'. The mentioned aspect is about '{r_aspect}'. Based on the common sense, what is the implicit opinion towards the mentioned aspect of '{entity}', and why?", "out":  "r_opinion"},
    {"prompt": "Given the sentence '{sentence}'. The opinion towards the mentioned aspect of '{entity}' is '{r_opinion}'. Based on such opinion, what is the sentiment polarity towards '{entity}'?", "out":  "r_polarity"},
    {"prompt": "Given the sentence '{sentence}'. The sentiment polarity is '{r_polarity}'. Based on these contexts, summarize and return the sentiment polarity only, such as: positive, negative, neutral.", "out": "r_label"}
  ]
}