engleski

Analysis of the Textual Information Extracted from News Portals

The primary means of informing the population in modern society is through news portals. This paper analyses the characteristics and effects that such way of communication creates. The influence was studied in particular on an example of current global phenomenon “vaccination” (hr. cijepljenje). The research method follows the CRISP-DM process adapted to the digitalized form of textual data. The analysed corpus, in the form of natural spoken language, was scraped from Croatian news portals. The subsequent processing extracts information from unstructured textual sources and provides valuable insights, like how much a particular topic is represented in the article. Modeling is based on the application of multiple text mining algorithms, like Words Cloud, Topic Modelling, Concordance and Sentiment Analysis. The implemented model produces indicators for objective information interpretation. The findings suggest that the portals associated the notion of vaccination with the COVID-19 pandemic. Furthermore, this term was often used in a political context. The words used and predominantly negative character of texts dealing with vaccination has led to the transmission of negative emotions to readers. A significant aspect of the study is the fact that it was conducted on the corpus of texts written in Croatian – a relatively small and morphologically complex language.

Digital News, Information Extraction, Natural Language Processing, Text Mining, Web-scraping

nije evidentirano