GSoC/GCI Archive
Google Code-in 2014 Apertium

Create a convertor for events for a MaxEnt POS tagger

completed by: Stan K.

mentors: Kevin Brubeck Unhammer, Francis Tyers

Convert an event table in the following format:

 

Output                | Input        | Features

this<det><dem><sg>    |  this        | (1, "pos", "<adj>")

this<det><dem><sg>    |  this        | (1, "pos", "<n>") (-1, "pos", "<BOS>") (2, "pos", "<adv>")

this<prn><tn><mf><sg> |  this        | (1, "pos", "<vblex>") (-1, "pos", "<BOS>") (2, "pos", "<pr>")

 

To the following format supported by YASMET:

input.yasmet:

2

0 #  1:0 # 1:1

0 #  2:0 3:0 4:0 # 2:1 3:1 4:1

1 # 5:0 3:0 6:0 # 5:1 3:1 6:1

 

input.features:

1   (1, "pos", "<adj>")

2   (1, "pos", "<n>")

3   (-1, "pos", "<BOS>")

4   (2, "pos", "<adv>")

5   (1, "pos", "<vblex>")

6   (2, "pos", "<pr>")

 

-----------------