We utilized Twitter's Stream API function to collect our dataset. First, we collected Covid19 related Twitter data using https://github.com/computermacgyver/twitter-python Github repo. The keywords we used to feed Twitter's Stream API can be found here: covid19_keywords.txt. Next, we performed a GREP function on our dataset using the keywords found here: vaccine_keywords.txt. Then we used this filtered dataset to perform our analysis.
For reproducibility, we suggest using Hydator found here: https://github.com/DocNow/hydrator. We provided the Twitter Tweet Ids and Twitter User Ids from our filtered dataset to feed into Hydrator here: /data/twitter_id_str.txt. Note Twitter actively removes and/or suspends users/tweets from its platform, and this may or may not cause omissions from your dataset.
This repo contains four subfolders:
Data
The Data folder contains a list of Twitter User IDs and Twitter ID strings.
Scripts
The Scripts folder contains scripts we used to data munge. These scripts require the data to be in a specific format, which is not provided in the Data folder. The Scripts folder does not contains the necessary collection and processing scripts that enable the analysis scripts. In order to replicate findings, please navigate outside of this repo and move into the Covid repo, which will provide the collection and processing steps.
Analysis
The Analysis folder contains the scripts we used to create our analysis.
Results
The Results folder contains our findings in this study.