Using ZGen

From Apertium
Revision as of 22:24, 4 December 2016 by Bujol (talk | contribs) (Created page with " == '''Requirements''' == - gcc 4.5 (or above) - boost 1.57 (or above) - cmake 2.8.10 == '''Input format''' == The best way to do is to have your input for both training ...")
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigation Jump to search

Requirements[edit]

- gcc 4.5 (or above)

- boost 1.57 (or above)

- cmake 2.8.10


Input format[edit]

The best way to do is to have your input for both training and using in the (partial) Dependency Tree structure, which has four entries. If some fields are empty, the _ (underscore) is used.

[forms] [postag] [head] [deprel]

e.g. 

__Ms.Haag__ NP 1 _
__Elianti__ NP 1 _
plays VBZ -1 _
. . 1 _

Please note that noun phrases should have __ prefix and __suffix

If your file isn't in the correct format, you'll have to write a small script to translate it.

When it comes to field [head], the tree is 0-indexed, with -1 as the head. So we'll have:


the _ 1 _
boy _ 2 _
went _ -1 _
home _ 2 _

-1                went
                      |
         |-----------------------|
         |                          |
2      boy                   home
         |
1      the

Which means that 'went' is the head, 'the' is a son of 1 ('boy'), 'boy' and 'home' are children of 2 ('went')

Configuring[edit]

First use makefile:

./configure
make