kmjohnson/twitter-morals

U.S. Congressional Tweets Dataset used in the ACL 2018 paper for predicting Moral Foundations of the tweets of U.S. politicians.

twitter-morals's Issues

Clarification on first and second moral branch

Hi!

I am curious how the data was obtained for the first and second moral branch columns... in your paper you state that a subset of tweets were doubly-annotated by one liberal and one conservative annotator. Does this correspond to the first and second moral branch? If so, which one was the liberal annotator and which one conservative? Is it possible to indicate which annotator annotated each of the remaining examples?

Any clarification is much appreciated!

Clarification on Moral Foundations

Is the following mapping from numbers 1 to 11 correct?

The Moral Foundations labels (1..11) are created like this: 
    if branch == 1 and polarity == 1:
		mf = 1 **CARE**
	elif branch == 1 and polarity == 2:
		mf = 2 **HARM**
	elif branch == 2 and polarity == 1:
		mf = 3 **FAIRNESS**
	elif branch == 2 and polarity == 2:
		mf = 4 **CHEATING**
	elif branch == 3 and polarity == 1:
		mf = 5 **LOYALTY**
	elif branch == 3 and polarity == 2:
		mf = 6 **BETRAYAL**
	elif branch == 4 and polarity == 1:
		mf = 7 **AUTHORITY**
	elif branch == 4 and polarity == 2:
		mf = 8 **SUBVERSION**
	elif branch == 5 and polarity == 1:
		mf = 9 **SANCTITY**
	elif branch == 5 and polarity == 2:
		mf = 10 **DEGRADATION**
(If cell 9 has the value of 6, then its MF = 11 **NONMORAL**)

On the exhaustiveness of branches & polarity

Hi Kristen,

I have two questions regarding the moral foundation branches and polarity.

First, for tweets that are identified with 2 branches, shouldn't they contain two values for polarity as well? Or does the polarity only correspond to the first branch?

My current understanding is if a tweet's label is (branch1, branch2, polarity) = (2, 3, 1), this translates to fairness (branch=2, polarity=1) and loyalty (branch=3, polarity=1). Could you confirm whether this understanding is correct?

Second, the paper states that a "secondary foundation" is allowed to be annotated for a tweet. Does it mean that if a tweet contains two branches, those are the only labels it should have, according to the annotators?

I ask the second question because I'm interested in the absence of moral foundations in a tweet as well. However, the restriction to a maximum of 2 possible branches does not seem to guarantee that branches 3, 4, 5 aren't at all possible.

Your response is much appreciated!