Difference between revisions of "User talk:Jimregan"
| Line 4: | Line 4: | ||
:Probably for the bilingual dictionary, a probabilistic dictionary could be exactracted from the <s>EuroParl</s> JRC-Acquis corpus (see [[Corpora]]) using GIZA++ — of course it would need checking. - [[User:Francis Tyers|Francis Tyers]] 20:57, 6 October 2007 (BST) |
:Probably for the bilingual dictionary, a probabilistic dictionary could be exactracted from the <s>EuroParl</s> JRC-Acquis corpus (see [[Corpora]]) using GIZA++ — of course it would need checking. - [[User:Francis Tyers|Francis Tyers]] 20:57, 6 October 2007 (BST) |
||
Ok, well, for Irish-English, we might be able to get some help from Kevin Scannell, who has an unreleased Irish-Scots Gaelic pair, but who is very in favour of GPL software. Probably the pair with the best prospects is Polish-Russian, but its worth giving them all a go I reckon. My googletalk/gmail is <code>firstname.lastname@gmail.com</code>, ICQ is 28459314. Do you have a sourceforge account yet? We'll be moving to self-hosting at some point, but at the moment we're using sourceforge for SVN. |
|||
Regarding the "unsupported transducer type", what can you give me the versions of lttoolbox-unicode and apertium-unicode that you are using? I'll try your dictionary and see how it works. - [[User:Francis Tyers|Francis Tyers]] 21:54, 6 October 2007 (BST) |
|||
Revision as of 20:54, 6 October 2007
Wow awesome. Great work!
Which pair were you interested in Polish to/from ? Scratch that, I see Polish—English. If you want I can help set it up in our SVN. Do you have some kind of instant messaging ? - Francis Tyers 20:43, 6 October 2007 (BST)
- Probably for the bilingual dictionary, a probabilistic dictionary could be exactracted from the
EuroParlJRC-Acquis corpus (see Corpora) using GIZA++ — of course it would need checking. - Francis Tyers 20:57, 6 October 2007 (BST)
Ok, well, for Irish-English, we might be able to get some help from Kevin Scannell, who has an unreleased Irish-Scots Gaelic pair, but who is very in favour of GPL software. Probably the pair with the best prospects is Polish-Russian, but its worth giving them all a go I reckon. My googletalk/gmail is firstname.lastname@gmail.com, ICQ is 28459314. Do you have a sourceforge account yet? We'll be moving to self-hosting at some point, but at the moment we're using sourceforge for SVN.
Regarding the "unsupported transducer type", what can you give me the versions of lttoolbox-unicode and apertium-unicode that you are using? I'll try your dictionary and see how it works. - Francis Tyers 21:54, 6 October 2007 (BST)
