learning a causal graph from text

[juliensiebert]

Dear all,

thanks for this awesome repo. I have not seen (maybe I have missed it) any work on learning a causal graph from text where the input would be some text and the output to be predicted (i.e., the labels) would be a corresponding causal graph.
A very simple example would be:

• input: "Ada hits the ball. The ball breaks the window."
• output: (kick the ball) ---> (window breaks)

Are you aware of such kind of work?

Best regards,

Julien