Christina Zheng

@matt0fer Can you double check if the app is still broken? And can you try our /message API to see if the utterance works? https://wit.ai/docs/http/20220503/#get__message_link

@matt0fer Can you double check if the wit.ai app is really 2614205462213696? Since the wit.ai app works well with /message API, if the NLP webhook setting is correct, it shouldn't...

At first, can you go to the Messenger webhook settings page and double check if the nlp setting looks good? And if the setting looks good, can you try to...

The app's training status shows that the training was done. SO it is not the training issue. Can you describe the problem with more information? Ex. the sample utterances and...

@brennadoyle Checking the utterances tab, the utterance "How do I find a job" is annotated with intent "Jobs". Can you double check all of the utterances' annotations?

@brennadoyle What I noticed is that the data annotation is incorrect. The screenshot I shared is not from the prediction result, but from the Utterances page. The data annotation in...

@anjaly I noticed that, in your app there are many utterances with word "test" and annotated as OutOfScope. This may be the reason for "Test 2" not working. Can you...

@anjaly Removing the examples won't make them considered OutOfScope. Removing utterances and adding them back won't affect the model quality. If the utterances and annotations are the same, the model...

The problem about only 25 keywords is just a UI display problem. The keywords still exist in the backend. The UI fix is on the way.

"freetext & keyword" type supports nesting. If you modify the entity "for_or_against" to "freetext & keyword", and annotate a new utterance, the nesting will happen. Any reason why you don't...