Labs/Ubiquity/Meetings/2009-03-24 i18n Meeting

Details

IRC channel: #ubiquity
Dial in:
- +1 800 707 2533 (pin 369) Conf# 201 (US Toll Free/Skype)
- +1 650 903 0800 x92 Conf# 201 (US/International)
- +1 416 848 3114 x92 Conf# 201 (Canada)

meeting times: this seems to be a good meeting time for many people... the latest doodle vote raised no objections. We'll try to continue to meet at this time weekly or so.
the new parser: mitcho created a demo of the new parser design. (source available)
- The design pairs a universal parser class with individual languages' parameters, making it easy to add new languages.
- cers added a Danish parser on the fly (now added to the demo + bitbucket)
- Future plans:
  - Felipe may make a Portuguese parser
  - Aza feels it's a little slow... we should continue to optimize it.
  - We should incorporate this into Ubiquity proper as a Ubiquity parser plugin. This may break current commands, though, with its new way of specifying arguments by semantic role. Aza suggests the time to do this may be together with web-page commands.
thoughts on collecting data (discussion of these two blog posts)
- Observation: people are excited and want to contribute, but not all can code.
- Consensus: asking for translations is easier than asking for someone to write a parser.
- Idea: we can build a "command-bank" of translated Ubiquity input and maybe train parsers over it—this could give us a good stab at many languages very quickly.
- Future plans:
  - Discuss how best to collect these types of data with l10n folks
  - mitcho will explore tools/scripts to automatically train parsers based on this kind of data.