Closed myrix closed 1 year ago
Implemented, now we can have instances from this corpus.
To ensure instance generation, it's better to create new entries, re-upload and re-parse documents, and then disambiguate two or three words in each new parser result. Then new instances would be generated after selecting the corpus at http://lingvodoc.ispras.ru/valency and clicking 'Update valency data' button.
We probably should make valency data updating better, in particular, finish #775 and and think about how we should do it in cases like this, when we change the algorithm and should re-parse some data, when data does not change.
On http://lingvodoc.ispras.ru/valency there's no instances from the http://lingvodoc.ispras.ru/dictionary/6236/1709/perspective/6236/1712/view corpus.
The problem is non-standard VBLEX and VBSER verb markers of the Apertium Russian parser, , incompatible with Lingvodoc's valency instance code which understands only V.
We should upgrade the code so that VBLEX- and VBSER-marked verbs are also recognized.