In the past several months, we have witnessed a non-precedent disease which is COVID-19,
known initially as Corona Virus Disease of 2019 and later declared as a pandemic.
It has resulted in unprecedented pressures on each country for controlling the population
by properly utilizing available resources. And it is now becoming a source of depression,
stress, and anxiety because of information posted on social media. In this project,
we designed and developed a web application for mining large amounts of Twitter data
for pandemic crisis management. The twitter text data is targeted specifically for
the COVID-19 pandemic. We analyzed twitter text data using mining techniques such
as natural language processing (NLP) to find the sentiment of the text. The data was
passed through several processing steps, starting from collecting the relevant messages
and tweets and storing them in data files for further processing. We then performed
filtering based on user inputs such as keywords, date ranges, to retrieve tweets from
the data files based on the filtering terms.
We then combined the analysis results with the associated metadata, such as geolocation, for high-level analysis and visual inspection. For example, users can view sentiments of a given keyword (e.g., "Vaccine") for a given date frame on the world map. Thus, the web application did provide a user-friendly and interactive graphical interface. We believed that by correlating tweet sentiments with corresponding users' geolocations, we could identify and locate crises in different geographic locations and the severity. Therefore, we hope the proposed technique can answer questions like, "Which city is facing the highest level of difficulty in the food supply?" or "what is the most pressing issue of a particular area?" and so on. We believed the proposed work would have a significant impact on pandemic crisis management. It will be helpful for different entities, such as government or non-government organizations, who strive to provide maximum support to the citizens to help them during the crisis. We already performed extensive testing of the application and obtained interesting results. For example, for different search terms(e.g., Coronavirus , Vaccine etc.) we have identified the top 10 most negative and top 10 most positive countries and cities. Also, we have demonstrated how the running time varies with different search terms due to the frequency of the term found in the Tweets.