I am using IBM Watson tool to determine tones (https://tone-analyzer-demo.mybluemix.net/) and personality scores (https://personality-insights-livedemo.mybluemix.net/) on different files containing natural language. What would be the recommended strategy and rationale for choosing between:

  1. Getting scores from different files and averaging/manipulating them.
  2. Getting one raw score on aggregated data and using it.
  • $\begingroup$ I love this question if only to get to know the tools you mentioned. It can be improved by telling your goal. $\endgroup$
    – Pieter21
    Commented Oct 24, 2017 at 21:45

1 Answer 1


I think this is 'mood' vs. 'character'. And as such, it depends on the question you ask. E.g. did I make my customer upset? Or is my customer loyal?

Though mood and character correlate, as someone with a certain character will be more likely to be in some mood, there is also variation.

I think you will measure 'mood' with the 'tone' analyzer, and 'character' with the 'personality' analyzer.

For the 'personality' you will need all the data you can get to get the 'average', for the 'mood', a single data point can be sufficient.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.