Dear Nadilla,
thanks so much for your feedback and your questions! Your paper sounds really interesting as well but I'm still working my way through my CADAAD-watchlist on YouTube!
(1) the historical newspaper data were extracted from Nexis and Gale Primary Sources
(2) the comments/metadata are in a csv-File, and I mostly annotated using simply MS Excel - the comments themselves were also POS-tagged (UCREL CLAWS7 tagset, using using the free CLAWS web tagger) and then analysed with AntConc (e.g., for collocations of pronouns and easier viewing of concordance lines).
(3) apart from very few examples of typed ;) and :) to mark sarcastic/'humourous' comments, there weren't any emojis. The upvotes/downvotes would be interesting to look at in more detail, but so far I've basically only established that the negative comments always have more upvotes than downvotes (and vice versa).
I'm currently waiting for reviewers' responses to the submitted book chapter on this - I'm happy to send you the revised/accepted version once it's ready (September/October)!
Best wishes from Liverpool,
Ursula