This era of COVID-19 has brought with it many things, like social distancing, working from home, and a whole lot of data. Data about the origins of COVID-19, its spread, prevention, reactions and responses, all of which is being constantly studied and analyzed all over the globe.
A good amount of this data comes from social media, given that we live in the social media age. Social media is a data source that is increasingly popular in Uganda, as is shown here. In fast-changing situations, it allows for understanding public perception and the evolution of public discussions. It also sheds light on public misconceptions and can even be a way to assess levels of public engagement.
At Sunbird AI, we are making our contribution to the study of this data by working on a social media analysis project. In this project, we analyze the Ugandan public response to the COVID-19 era along with the new policies it comes with.
Our focus so far has been on Twitter, which provides a special kind of data because people voice their thoughts and reactions to events in real-time. Twitter data can capture what exactly people feel about something even as it is unfolding.
Case study: masks
An example of the analysis we did was a case study on the reactions and discussions about the issue of compulsory face masks. This analysis was carried out in the days right before and right after it was announced that a free face mask would be availed to every Ugandan.
Here is a graph showing the tweets about it over a number of days:
We also analyzed the mask-related tweets by general subtopic, as shown below:
From this analysis, we were able to clearly gauge where the major interests of the public lay in relation to the issue of masks. As shown in the above image, the bulk of the masks discussion was about the implementation: how the exercise of distributing masks would be carried out. There were also concerns about the use of masks, i.e whether it is compulsory to wear a mask, which kinds of masks to wear and how to wear them, as well as a few political concerns.
The image below shows the major themes from the tweets about masks:
Our implementation of this project consisted of writing a Python script that sends requests to Twitter’s API and retrieves tweets along with other related information like likes, replies, and hashtags. Then there was a data visualization step using multiple visualization libraries in Python.