To be able to compile and run:
- Java JDK 8 (javac and java commands).
- make and curl commands.
- Connected to the internet (to download the corpus)
- Alot of RAM (~8 GB) to be able to preprocess the data.
Run "make init" command. This should take about 5 minutes.
If you want to use a pre made n-gram file, rum "make init-fast".
Run "make run". The program should take ~10 sec to start.
Run "make run-grammar". The program should take ~10 sec to start.
There are three types of results that can be generated: "make eval-nogrammar" "make eval-grammar" "make eval-userlearning" This takes about 20 mins to complete and the output is in the res/evaluation folder, where each type of result get an own result output folder.