medianeuroscience / emfdscore

Fast, flexible extraction of moral information from textual input data.
GNU General Public License v3.0
103 stars 29 forks source link

Which text scoring option is most suitable for my research? #20

Closed MarleneJacobsen1 closed 1 year ago

MarleneJacobsen1 commented 1 year ago

Hey, I was just wondering if I could get some advice on which scoring option to choose for my research. I am analysing newspaper coverage about the issue of loss & damage (compensation for climate change-related damages).

I understand from the tutorial that it's not advisable to use the 'all probabilities' option as I want to determine which foundations are more or less represented in the articles. That leaves me with option 2 and 4. I'm not sure whether sentiment scores or vice-virtue scores will be more useful for my purpose.

So the sentiment scores range from -1 to 1. What about the vice and virtue scores? It looks like they're always positive. So do they range from 0 to 1?

What's the benefit of having a foundation probability in addition to the sentiment score? Why does the output for vice-virtue not also include foundation probabilities?

Lastly, if I want to calculate the average sentiment score per foundation per newspaper, would it be an issue that some are negative and others are positive? Would python or R automatically square them to prevent the scores from cancelling each other out?

Apologies for my absolute beginner questions. I'd be grateful for any advice and opinions. Thanks a lot!