Researchers use social media data to analyse public sentiment about Coronavirus measures

Project funded by eScience Center

For the past months, many of us have experienced all kinds of feelings about the safety measures against the Coronavirus. The announcement of some new measures may have us feeling timid and resigned, or on the other hand, rebellious and insubordinate. Understanding the public sentiment on a national level may help policy makers make better decisions, and even predict people’s adherence to current and future measures. With this goal, computer scientist Shihan Wang will use social media data for sentiment analysis around Coronavirus measures.

Social media on smartphone

Shihan Wang explains: “We select Twitter data about a certain topic using relevant hashtags and keywords, and then we use machine learning and natural language analysis to figure out whether people have a positive or negative opinion towards the current policy measures.”

Her project, ‘Dutch Public Reaction on Governmental COVID-19 Measures and Announcements (PuReGoMe)’, was funded by the Netherlands eScience Center last month. In this project, Wang works together with two colleagues from the Intelligent Systems group, Marijn Schraagen and Prof Mehdi Dastani.

Specific measures

Wang and her colleagues will do the data analysis in two parts. “First of all, we look at the overall period, to see what the trending topics are and how they change. For example, does anything change when RIVM announces their daily statistics around 2pm? Secondly, we do case by case analyses, where we study the public sentiment around specific measures. This is mostly relevant around the press conferences in which new measures are announced.”


The project has only just started, and will run for six months in total. “We are still in the building stage,” says Wang, “but we’re trying to have the first results as soon as possible. We are collaborating with RIVM, who can perhaps use this data to help predict how well the public will adhere to the current measures. Besides the real-time monitoring aspect, we also hope that this project will contribute in the longer term to insights into social media behaviour during a pandemic, both from a computer science perspective and for policy making.”

The researchers do admit that Twitter data in itself may not be completely reliable. “We do validate our results with data from other social media. And of course this should not be the only source of information to adjust measures. But it could provide a cost-effective and efficient way to predict public reactions in a timely manner.”

eScience Center

In addition to this project, the Netherlands eScience Center also funded a project by mathematics researcher Martin Bootsma, who is working on calculations around the exit strategy in the Netherlands. Read more in the eScience Center news article.