The Widely Interpretable Semantic Representation (WISeR) aims to capture abstract and concrete concepts, along with their relations, within plain text. A key objective of this project is to develop a meaning representation that can be seamlessly transformed into semantic graphs, enabling probabilistic reasoning in the context of dialogue. WISeR differentiates itself from the broader concept of Abstract Meaning Representation (AMR) through its unique features and characteristics such as:
- It provides deeper and richer levels of abstraction.
- It is Interpretable by non-experts in computational structures.
- It covers relations across utterances (e.g., referents, external arguments).
- WISeR GUidelines
- wiser-amr3: the AMR 3.0 corpus converted into WISeR
- wiser-dialogue: the new dialogue corpus annotated in AMR and WISeR.