Hi,
I'm sorry I haven't been in touch this week. The weekend was busy with full day placement trainings.
I've
managed to draft an algorithm and implement it (partially). Varnam can
now use any stemrule but for now all the stemrules belong to a single
level. I've created a separate repo for designing the algorithm. You
will be interested in the file "03algorithm" :
https://github.com/lonesword/mlstemmerI've added some things (like intermediate word learning) that might be of good use to varnam in the algorithm.
Maybe I can work on providing better and less cluttered documentation on the algorithm design towards the end of the program.
I will implement the 3 tier algorithm within 2 days, make a commit and inform you :)
A few things :
1. Ruby has issues
printing a c string. Apparently, ruby strings are not '\0' terminated
and this causes some trailing garbage when using varnamc with the -m
option (to stem from command line). A work around is to write a print
function, attach it to ruby using ffi and call it - but that sounds like
a rather unclean way of doing it. Any ideas?
2. I tried compiling argtable statically
from source and link it. It worked. However, I can't use #include
"libargtable.h" because it says file not found. Even without the
#include the functions are working but I cannot access the members of a
struct directly (which is necessary). Also, adding the libargtable
source to libvarnam adds another 6.0 mb. Maybe I can get back to this
later :)
3. GSoC midterm evaluations open on June 23rd and closes by June 27th. Just a reminder :)
Also, can we have a meeting sometime this week?
thanks,
Kevin.