Why-questions try to get to the bottom of the matter and ask for explanations. In this paper, we show how we can learn the complete structure of probabilistic rules for a question-answering system from example interpretations. These rules are then used by a meta-interpreter to find answers in the form of explanations for why-questions for a particular application domain.
|Title of host publication||IMA2019 First International Workshop on Interpretability: Methodologies and Algorithms (IMA2019)|
|Number of pages||9|
|Publication status||Published - 2019|
- probabilistic logic programming
- probabilistic rule learning