Sentiment Analysis of Italian and English Corpora of Internet News: A Comparison with Some Economic Trends

Computational linguistics, Sentiment analysis, Internet media.

Authors

  • Luca Pavan
    luca.pavan@flf.vu.lt
    Institute of Foreign Languages, Vilnius University, Vilnius, Lithuania; Language Studies Center, Faculty of Creative Industries, Vilnius Tech, Vilnius, Lithuania; Department of Foreign Languages, Literary and Translation Studies, Vytautas Magnus University, Kaunas, Lithuania.
May 13, 2022

Downloads

In this article, the sentiment analysis of several large Internet corpora made of Italian and English news is performed using a software written by the author, showing a possible connection with some economic trends. In this research, the news includes different topics (not necessarily financial news), and they are extrapolated from a large number of Internet newspapers. The software, already used in a previous article by the same author, is lexicon-based and makes use of scale points ranging from 0 to 100 to calculate an index of positivity in a text. The variation of sentiment tendency in the news corpora, calculated for a time period of several years, is later compared with some graphs showing some parameters of some economic trends, including the gross domestic product (GDP). It is found that the sentiment tendency of the news seems to have a relationship with the tendency of some economic trends that span the same time period. Positive growth of the economy per year seems connected with a positive variation in the index of positivity. Inversely, for a negative trend in the economy, the variation in the index of positivity is also negative. The article shows that, for various news topics, sentiment analysis can be useful to better understand some economic trends. For financial news, many studies show the possibility of predicting GDP growth through sentiment analysis. In this article, it is hypothesized that a prediction based on large news corpora including various topics could also be possible.