The ScenarioExtractor is an operator designed to extract scenario information from conversation content. It utilizes a large language model (LLM) service to analyze dialogues and generate corresponding scenario descriptions.

__init__

def __init__(self, llm_serving: LLMServingABC)

Parameter	Type	Default	Description
llm_serving	LLMServingABC	Required	An instance of a large language model service, used for analysis and generation.

Prompt Template Descriptions

Prompt Template Name	Primary Use	Applicable Scenarios	Feature Description

run

def run(self, storage: DataFlowStorage, input_chat_key: str, output_key: str = "scenario")

Parameter	Type	Default	Description
storage	DataFlowStorage	Required	The data flow storage instance, responsible for reading and writing data.
input_chat_key	str	Required	The input column name corresponding to the conversation content field.
output_key	str	"scenario"	The output column name for the extracted scenario.

🧠 Example Usage

🧾 Default Output Format

Field	Type	Description
input_chat_key	str	The input conversation text.
scenario	str	The extracted scenario from the conversation.

Example Input:

{
  "chat": "USER: Can you book a flight for me from New York to London?\n\n\nASSISTANT: I'm sorry, but I don't have the capability to book flights. My current function allows me to get the exchange rate between two currencies. If you need help with that, feel free to ask! <|endoftext|>\n\n\n"
}

Example Output:

{
  "chat": "USER: Can you book a flight for me from New York to London?\n\n\nASSISTANT: I'm sorry, but I don't have the capability to book flights. My current function allows me to get the exchange rate between two currencies. If you need help with that, feel free to ask! <|endoftext|>\n\n\n",
  "scenario": "Attempting to book a flight."
}