This repository contains the data and materials related to Twitter Deliberative Politics and the CLAPTON (Corpus for the Linguistic Analysis of Political Talk ONline) datasets of social media posts annotated for facets of discussion quality.

Dashboard for model inspection: https://share.streamlit.io/sriramelango/nus-political-discourse-quality/app.py

The dashboard is for illustrative purposes, and the text pre-processing may not exactly follow what was reported in the paper for the sake of speed. We are working on optimizing the dashboard to accurately reflect the models in the paper.

Models used in the dashboard: Are in the pickles folder

Other resources corresponding to this paper are at: https://osf.io/28esd/

Which variables to use? Please refer to the most recent results discussed at https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3870554

#################################################Folder structure data/trainingdata/jaidka.csv comprises the raw annotated data

supplement/codingscheme.docx describes the coding scheme

Each observation was labeled for each of the deliberative qualities with a 0, 1, or NULL depending on whether there was agreement between at least 3 if not 4 annotators out of 4 total annotators.

###################################################Variable names

message_id and message: an id and the text of the tweet relevance: see coding scheme, codes whether the tweet is relevant to politics.

positive: whether the tweet shows empathy and respect (note that "positive" means empathy and respect) uncivil: any kind of incivility, whether, abuse, threat, or exaggeration uncivil_abuse: incivility that is abusive (see coding scheme) uncivil_threat: incivility that involves threats (see coding scheme) uncivil_exag: incivility that involves exaggerated claims (see coding scheme) reciprocal: deliberation which involves a genuine question (supposed to encode engagement) justif: justification whether internal or external. justif_int: internal justification justif_ext: external justification const: constructiveness whether fact-checking, solution-finding, or common-ground-finding const_fact: constructiveness involving fact-checking behavior const_com: constructiveness involving common ground finding const_sol: constructiveness involving solution finding len: length of tweet (is not used as an iv)

########################References

Jaidka, K. (Under review). Talking politics: Building and evaluating data-driven lexica to measure political discussion quality. SSRN.

Jaidka, K. (2022). Developing a multilabel corpus for the quality assessment of online political talk. In Proceedings of the 13th language resources and evaluation conference (Vol. 13).

Jaidka, K., Zhou, A., & Lelkes, Y. (2019). Brevity is the soul of Twitter: The constraint affordance and political discussion. Journal of Communication, 69(4), 345-372.