Open timgluz opened 7 years ago
@timgluz That sounds good. We could store the trained index in DB or in a git repo.
Currently doing research - going to checkout which tools are possible to use and which them are the best option;
Done: all the relating packages can be found here: https://github.com/Fosslim
Steps for refactoring:
gem "license_matcher", "~> 0.3"
into Gemfilei wrote a simple ROR app fosslim/lmweb
to demonstrate the usage
It's one of reasons why specs are slow - it must rebuild the token index each time and it takes time;
Fastest solutions is to move everything into own repo; But it doesnt speed up creation of index;
Better approach: use proper tools, refactor it into 2 processes:
training/match
, where the training-process will build and save model; and match-process will re-use already existing model;