Closed sandroacoelho closed 7 years ago
Hi all, this may be related to the moving of the FREME framework installations to the Adapt centre. I assume that the moving does not include data sets that are currently being uploaded. @jnehring can maybe say more (he will be back next week, I assume).
Hi @fsasaki , Thank you. I will check with him
NER consists of spot, classify and link. The dataset you provided only adds the term "ankerloze spouwmuur" to the linking phase. Maybe the term is not recognized by entity spotting? Or it is wrongly classified?
Also you specified input format = plaintext. But the text you submitted does not look like plaintext. You can experiment with different input formats, maybe XML is a better choice?
If you like we can discuss this issue on wednesdays developers call.
Hi @jnehring, I have answered @Gustavo with the same explanation on sunday. Thanks for your answer.
We can close it
@Gustavo Publio says:
Hi all,
As you may already know, we are using the FREME NIF service API to annotate some Dutch texts.
But we found out that the service is missing some annotations.
For instance:
We are using the dataset sbr2 that I've already uploaded into the service This dataset contains, among other terms, several references to the term "ankerloze spouwmuur":
Even so, if we submit a text with this term, it is simple ignored. For instance, here is a sample curl call:
As we are relying in the data provided by the NIF service, could you please check what is going on? Whether it could be something related to the language, or maybe the way we are doing it.