Description
The files in the input directory must be one of the following supported types:
- ner_json
- IOB
- BILOU
Solution
Provide path to a directory for the Input Folder parameter containing supported files. The supported file formats are JSON or CSV. The following are the supported dataset types:
- ner_json—The folder must contain a .json file with text and the labelled entities.
- IOB—This format contains
the following tags: B -
'beginning', I -
'inside', O -
'outside'. The folder must
contain the following two .csv files:
- token.csv—Contains text as input chunks
- tags.csv—Contains IOB tags for the text chunks
- BILUO—This format contains
the following tags: B -
'beginning', I -
'inside', L - 'last', O -
'outside', U -'unit. The folder must contain
the following two .csv files:
- token.csv—Contains text as input chunks
- tags.csv—Contains IOB tags for the text chunks
For more information about these formats and labelling data in these formats, see the Labelling text using Doccano guide.