Table 3. Examples of sentiment results for some tweets based on manual annotation.

Evaluations provided by Annotator A3 were excluded from the weighted average calculation as the number of outliers in its results exceeded 60%. Annotators with asterisks were given a double weight because they are vegans and are good at a context-specific lexicon.

#	Tweet	A1* w = 2	A2 w = 1	A3 (Excluded)	A4* w = 2	A5* w = 2	Weighted AVG	Annotation result	Sentiment
1	Moral consis...	1	3	5	1	4	2.14	1	Positive
2	I don’t care ...	−3	1	3	−2	−4	−2.43	−1	Negative
3	Coke Bottle ...	3	4	5	2	3	2.86	1	Positive
4	A great revie...	3	4	0	2	3	2.86	1	Positive
5	Over 110M a...	−3	0	5	−2	−3	−2.29	−1	Negative
6	What’s more...	−5	0	5	−2	−2	−2.57	−1	Negative