There is now a first version of database filtering:

Using a XML dump file to extract the raw text of Wikipedia articles seems very hard. Correctly expanding templates and handling links to produce text corresponding to what is seen in the web browser is problematic.

WikiExtractor.py on github seem to work, as long as you don't use the latest commit, which is broken. You need to use the --templates option.

There is something called gensim.

Somehow, the "same* script had probably been downloaded using different methods:

The component for generating manuscripts for recording sentences (rather than full articles):

The work of adding MySQL/MariaDB support is about halfway through. There is a temporary branch with a version of pronlex that runs on MariaDB, passing all relevant test of the original Sqlite3 version.

Declined due to notification of too long inactivity on open task.

Declined due to notification of too long inactivity on open task.

I don't think we have any objections to the name, so please go ahead.

Almost done, but a few minor things need some more attention.

Fixed on not yet merged branch (dbapi.Query.TagLike)

Fixed on not yet merged branch

I think incorrect json might map into an empty or incomplete entry struct on the server side? We should take a look at this... maybe there need to be a separate validation step.

There is now a first version of API call to the lexicon server for deleting an entry, never to be seen again:

Joakim (KTH) mentioned the Montreal Forced Aligner:

No changes in the API have been made, but inserting/updating an lex.Entry should now work with the lex.Entry.Tag field.

This will probably break things in other places...

Has now been merged with pronlex master branch. This will probably result in errors in other places...