sgnlp.models.span_extraction.utils.get_all_evidence_utterance_from_conversation

get_all_evidence_utterance_from_conversation(emotion: str, conversation_history: List[str]) → Dict[str, List[str]]

Iterate through a conversation history to let each utterance be the evidence utterance. The last utterance is treated as the target utterance. Ouput dictionary is in a format which can be used with RecconSpanExtractionPreprocessor

Parameters

• emotion (str) – Emotion of the target utterance

• conversation_history (List[str]) – List of utterance in a conversation. The last utterance is used as the target utterance.

Returns

Dictionary in a format that can be used with RecconSpanExtractionPreprocessor

The dictionary looks like this: {‘emotion’: [‘happiness’], ‘target_utterance’: [‘……’], ‘evidence_utterance’: [‘……’], ‘conversation_history’: [‘……’]}

Return type

Dict[str, List[str]]