Closed fishfree closed 3 years ago
@fishfree sorry for the slow response.
In general, the vice-virtue scoring option should be used if you want eMFDscore to return the average degree to which each of the moral foundations is upheld (virtue) or violated (vice) in a text.
As described in the tutorial:
The vice-virtue split is accomplished by considering the average sentiment of each foundation of each word, and then assigning this word to "virtue" if the foundation sentiment is positive, or to "vice" if the sentiment is negative.
Let's look at an example: Assume the word "kill" has the following probabilities and sentiments:
care_p fairness_p loyalty_p authority_p sanctity_p 0.4 0.174757 0.240741 0.202532 0.217822
care_sent fairness_sent loyalty_sent authority_sent sanctity_sent -0.695627 -0.796594 -0.668615 -0.751787 -0.684514
Using the emfd bow single vice-virtue
option, the word kill would be assigned a SINGLE care.vice score of 0.4, because a) the probability for the care foundation is highest compared to the other foundations, and b) the CARE sentiment for the word kill is negative.
Using the emfd bow all vice-virtue
option, the word kill would be assigned care.vice=0.4, fairness.vice=0.17, loyalty.vice=0.24, authority.vice=0.20, and sanctity.vice=0.21 score; because the sentiments are ALL negative for each foundation. IF the word kill had a positive sentiment score for any of the foundation, then the probability would be assigned to a .virtue column.
Please let me know if this clarifies matters and if you have further questions.
@fhopp Thank you very much! Understood now. Sorry for replying you late.
In the Jupyter note:
I cannot understand, could you please give an example? Many thanks!