apertium/apertium-anaphora

Prototype anaphora in C++

Closed this issue · 6 comments

prototype exists in Python (coding challenge). Need to write it in C++

Code Written in C++.
Need to deal with ambiguity in biltrans (if any).

Completed Code - Deals with Escaped Characters and Multiple TL entries as well.
Python Code takes 0.7 seconds (500 sentences) and this code takes 0.4 seconds for the same input.
Note: this is time taken in bash, including stdout, etc.

Also compared outputs of both codes and diff returns nothing. Now we can improve upon this code.
Going to make a makefile because parsing and anaphora code are in different files.

If you're timing, you might want to do more than 500 sentences, there's often some warmup period. And pipe output to /dev/null, everything is slow if it has to be printed

@unhammer I was just looking at the time comparatively. This time I put the output in a file to reduce that time.

if you are comparing timings, the different implementations might have different warmup times (I don't doubt the C++ is faster, but by running on different sized input you'll get some idea of what the warmup-time is vs fully-warmed-up lines-per-second rate is)