Topic Modelling of ongoing conflict between Russia and Ukraine

Online news sites provide hotspots to extract popular ratings and opinions on a wide range of topics. Realizing what individuals are referring to and understanding their concerns and suppositions is exceptionally significant to organizations and political missions. Furthermore, it is incredibly difficult to physically peruse such enormous volumes of data and gather the themes. Keeping in mind the prevailing plight of war-torn nations such as the recent conflict between Russia and Ukraine. This study performs aims to perform topic modelling using LDA (Latent Dirichlet Allocation) and text analysis on datasets collected from various online news websites. To increase the accuracy and efficacy of the topic modelling, a comparative analysis is proposed that elevates the performance of machine learning models. This study also develops an algorithm where the entire process can be automated from the point of data collection to finding optimum array of topics in the given dataset. Searching for insights from the collected information can therefore become very tedious and time-consuming. Topic modelling was designed as a tool to organize, search, and understand vast quantities of textual information. The topic model using LDA was utilized to do a text analysis for this research. In the beginning, researchers have scraped a total of 1178 articles that covered the war conflict between Russia and Ukraine from December 1, 2021, to May 16, 2022. After that, researcher built the LDA model and modified hyper parameters based on the coherence score Cv that was used for the model evaluation technique. When using the most effective model, prominent topics, and representative documents pertaining to each topic, topic allocation among the documents, and potential enhancements are covered in the last section.