SAOKE stands for Symbol Aided Open Knowledge Expression. It is a manually annotated dataset containing more than 40,000 Chinese sentences and corresponding facts in SAOKE form. It is the largest publicly available manually annotated dataset for open domain information extraction tasks.

This dataset has the following advantages:

• The data is authentic and open to use: following the OIE system concept, using original sentences to express knowledge
• Compatible with all types of knowledge: Provides a unified view of four types of knowledge (relationships, attributes, descriptions, and concepts)
• Accurate expression: Ability to accurately express facts using discrete relational phrases, missing information, hidden information, etc.