Aspect-based sentiment analysis of financial texts for predicting corporate financial performance
Provider: Grantová agentura ČR
Programme: Standardní projekty
Implementation period: 01.01.22 - 31.12.24
Workplace:
Fakulta ekonomicko-správní - Centrum pro vědu a výzkum
Investigator: Hájek PetrTeam member: Munk Michal | Kovárník Jaroslav | Boďa Martin | Myšková Renáta
Description:
Sentiment analysis is attracting considerable interest in finance due to its capacity to provide additional insight into opinions and intentions of investors, managers and other stakeholders. A remarkable improvement in predicting corporate financial performance has been achieved when considering the textual sentiment. Previous approaches in this regard have been limited to document-level sentiment analysis of financial texts. However, a more fine-grained aspect-based sentiment analysis is presumably incrementally informative in predicting corporate financial performance because opinions expressed on different aspects of companies can be extracted. Here, we propose to investigate how aspect-based sentiment analysis of firm-related financial texts might predict corporate financial performance. To do this, we combine corporation-expressed, media-expressed, and internet-expressed aspect-based sentiments with quantitative financial information obtained from financial statements and financial markets.
Sentiment analysis is attracting considerable interest in finance due to its capacity to provide additional insight into opinions and intentions of investors, managers and other stakeholders. A remarkable improvement in predicting corporate financial performance has been achieved when considering the textual sentiment. Previous approaches in this regard have been limited to document-level sentiment analysis of financial texts. However, a more fine-grained aspect-based sentiment analysis is presumably incrementally informative in predicting corporate financial performance because opinions expressed on different aspects of companies can be extracted. Here, we propose to investigate how aspect-based sentiment analysis of firm-related financial texts might predict corporate financial performance. To do this, we combine corporation-expressed, media-expressed, and internet-expressed aspect-based sentiments with quantitative financial information obtained from financial statements and financial markets.
