Every country uses its resources for services such as defence, education, infrastructure, and financing the operations of ministries. The state budget is an act passed for each year and contains the balance of expenditures and revenues. The country’s economic and political situation is one of many factors that can have a high impact on determining the state budget expenditures. In this thesis, we explore the possibility of measuring and uncovering the true circumstances of the state budget by analysing news of the state administration website. It is possible to uncover potential connections between news and the state budget by examining changes between budgets, a deep processing of news, their number, and their presented sentiment. By comparing and analysing passed state budgets, changed state budgets, and amending budgets, we showed the differences between them and the potential causes for the changes. Deep-learning methods and natural language processing allowed us to recognize the sentiment of every news story. The results showed a connection between the state budget and news from the state administration website. The number of news consistently kept a similar pattern throughout the year. The share of news marked with a positive sentiment was correlated with the heights of actual expenditures from the state budget through the years. The higher the share of positive news, the smaller was the sum of actual state budget expenditures, and vice versa. The results were mostly unsurprising. Our analysis could be improved with the inclusion of other non-state news sources.
|
