Exception when using the confidence metric

[ali3assi]

Hi by adding:
nlp.add_pipe('dbpedia_spotlight', config={'process': 'candidates', 'confidence': 0.75})
the following exception is generated:

Traceback (most recent call last):
  File "C:\Users\Admin\Documents\codePython\dbpedia\main.py", line 14, in <module>
    doc = nlp(text)
  File "C:\Users\Admin\miniconda3\envs\projet1\lib\site-packages\spacy\language.py", line 1022, in 
__call__
    error_handler(name, proc, [doc], e)
  File "C:\Users\Admin\miniconda3\envs\projet1\lib\site-packages\spacy\util.py", line 1617, in raise_error
    raise e
  File "C:\Users\Admin\miniconda3\envs\projet1\lib\site-packages\spacy\language.py", line 1017, in 
__call__
    doc = proc(doc, **component_cfg.get(name, {}))  # type: ignore[call-arg]
  File "C:\Users\Admin\miniconda3\envs\projet1\lib\site-packages\spacy_dbpedia_spotlight\entity_linker.py", line 138, in __call__
    start_ch = int(ent['@offset'])
TypeError: string indices must be integers

Any suggestion why I get such an exception please?
Thanks

[MartinoMensio]

Hi @ali3assi,
Thank you for reporting the issue.
It was a problem of the candidates endpoint that:

• when a single candidate exists, it is inside a dict
• when multiple candidates exist, they are in a list

Version 0.2.2 of the package fixes it. Let me know if you have any problems.

Martino