Jess,
I am not a linguist, and my answer will probably not satisfy you. I do know that I have asked a similar question about the "tones" of Toner Analyzer, and the "tones" of Natural Language Understanding (NLU). The tone algorithms and models used by the two services are different. Neither one is wrong, they were just trained differently, and have different underlying machine learning models.
What I usually tell customers who are looking for tone data is to try both. See what the cost differences are between the two of them, and see which model better fits the needs of your particular use case.
------------------------------
Daniel Toczala
Community Leader and Customer Success Manager - Watson
dtoczala@us.ibm.com------------------------------
Original Message:
Sent: Mon September 27, 2021 09:16 AM
From: Jess Shapero
Subject: Reconciling results within Tone Analyzer
Please can someone tell me how to reconcile Tone Analyzer's 'document tones' with the individual 'sentences tones'? I have read some of the documentation but I still cannot figure out how the calculations are made. Further to this, I have got some results from the Natural Language Understanding API - on the same texts - and they do not match the figures of the same 'tones' of the Tone Analyzer.
I would particularly welcome any response from any linguists here. Thank you in advance.
------------------------------
Jess Shapero
------------------------------
#EmbeddableAI
#WatsonAPIs