Information discovery with data mining in social media messages andpresenting in visual analytic environment: COVID-19 tweet dataset example

Nowadays, social media is the most effective tool for communication and building personal interests. With the widespread use of smart devices and social platforms, the importance of social media has increased and has become preferred by many people. Social media has become a source of Big data, unstructured, granular and large-scale digital data, supporting interaction between people. The COVID-19 pandemic is recognized as the most significant global crisis of our time. Although this pandemic initially started as a health crisis, it has rapidly evolved into a unique socio-economic and environmental crisis that has spread across the globe. Tracking spatial development constitutes a large part of pandemic management. There is a great need for spatial information to ensure an effective community response to control outbreaks, contact tracing and prevention of spread. Quality data and statistical results are essential for crisis management.Geovisual analytics utilizes different disciplines such as cartography, data management, data mining, interface design and cognitive science to help users detect patterns and predict future outcomes using spatial data. The main objective of this thesis is to develop a Web-based spatial visual analytics application that reflects the interactive nature of information science and cartography. In this context, the tweet data set created on the social media platform, formerly Twitter and now X, about vaccines during the COVID-19 pandemic period was used. Accordingly, the tweet data series created on the social media platform Twitter about vaccines during the COVID-19 pandemic period was used. In order to track the spatial evolution of the tweet data, location information extraction was performed. After that, the hidden patterns in the tweet data set were discovered. Using the new information discovered in the dataset, clustering analysis was performed using K-means and Ward methods. The similarities and differences of the countries where the messages were created were revealed. A Web-based interactive application has been developed in which the messages in the Tweet dataset, the new information discovered from the dataset, the information obtained as a result of the clustering analysis and the thematic maps produced interact with each other, filtering tools are included within the scope of the queries determined by the researcher, the spatial - temporal development of the messages in the dataset is followed on the same screen, and the results of the research are transferred to the user with different graphical tools.