search

normativeai / backend

GNU General Public License v3.0
3 stars 3 forks source link

Decoupling the formalization from the annotated text #41

Open shaolintl opened 5 years ago

shaolintl commented 5 years ago

Right now, every time the annotation changes (or even just the annotated text), the formalization is re-generated (upon clicking save). I see one main limit:

1) We cannot turn off formalized norms in order to debug the formalization

@Gauntlet173 , can you list further limitations of the current interface which will be solved by this decoupling?

Gauntlet173 commented 5 years ago

It is the formalization that does the "work". So if the user can't see the formalization, they can't see what they are doing. Right now they can see grey boxes, and they know there is some sort of connective there, but there is no way to look at it and see what connective it is. No way to change it if the connective is wrong, or if they have highlighted the wrong text.

We need something that is more WYSIWYG for logic.

There are other things I'm sure I just can't remember right now.

Separating the formalization from the text is one way to make sure the user can see what they are doing. Another way would be to change the way the annotations work so that everything you need to know is visible in the annotation window, but that is very difficult when you consider that many things not explicitly stated in the legislation may need to be formalized, so there is nothing to highlight, and the structure of the text doesn't follow the structure of the formalizations.

The downside of decoupling is that the formalization becomes your one source of truth, and you may need to figure out how to update the associated section of text when the formalization is changed. That's complicated.