This post is a brief summary about the paper that I read for my study and curiosity, so I shortly arrange the content of the paper, titled UnifiedSKG: Unifying and Multi-Tasking Structured Knowledge Grounding with Text-to-Text Language Models (Xie et al., EMNLP 2022), that I read and studied.

They proposed the UNIFIEDSKG framework to standardize datasets, models, code, experiments, and evaluation metrics into a single framework by casting user requeests, structured knowledge, and outputs into the text-to-text format as follows.

Xie et al., EMNLP 2022

The following is the task unification, which is six task families , on UnifiedSKG into text-to-text format.

Xie et al., EMNLP 2022

  • Semantic Parsiong converts questions to logical forms.
  • Question Answering derives answers to natural language questions based on structured data.
  • Data-to-Text generation describes structured data in natural language.
  • Fact Verfication checks if a statement is true based on the structured data.
  • Conversational Task require understanding of not only the user’s last request but also the full interaction history between users and machines.
  • Formal Language to Text Translation describes formal language in natural language.

The all tasks above take as input a user request, a structured knowledge input, and an optional (dilaogue) context to predict an output y.

For detailed experiment and explanation, refer to the paper, titled UnifiedSKG: Unifying and Multi-Tasking Structured Knowledge Grounding with Text-to-Text Language Models (Xie et al., EMNLP 2022)

Reference