Thread by @hexapode

I just finished teaching a 2 months class on Computational law at Epitech (a french software engineering school). Here is how it went…

I start with serie of 3 half day workshop through video conference where I start by going briefly through the history of the field, starting with Layman E Allen attempt to formAlise legalise in 1957 digitalcommons.law.yale.edu/cgi/viewconten…

Then moving to Legol, legol 2 and prolog with M. J. SERGOT and team work on British Nationality Act as a logic program doc.ic.ac.uk/~rak/papers/Br…

I then finished the historical context introduction with recent work happening at CODEX, and especially the brilliant work of O. Goodenough on standarisation with LSP law.stanford.edu/publications/d…

We then focus on Prolog (Swi-Prolog) and how it can be used to model legal text. As I was teaching a class of software engineer and not jurist, a lot of effort was put on explaining legal concepts so that they could understand the text they were modeling.

For the rest of the 2 months, students were given an assignment, build a system allowing the modeling of legal rules, use it to model the British Nationality Act, and expose it to an end user through a web interface / chatbot.

After 2 months of efforts, some groups have success in building a system to encode the #RulesasCode, to model the full legislation and to make a usable interface to query it.

This proves that making rules as code useful is feasible. Even for software engineers with no prior experience in logic oriented programming or legal background. Future is bright.