The Infrastructure tab is under construction in this moment. Thank you for your understanding.


Sentitopics is an R-package that implements joint sentiment-topic modeling for textual data. Traditional sentiment analysis dictionaries often overlook that a word’s sentiment can vary depending on its topical context. sentitopics addresses this by combining topic modelling with semi-supervised prior information from sentiment dictionaries, improving the accuracy of sentiment estimations. Validated with a multilingual dataset of parliamentary speeches, it demonstrates significant improvements over standard dictionaries without requiring human annotation. Moreover, sentitopics, compatible with R libraries like quanteda, tm, and tidytext, enables researchers to identify and monitor topic-specific sentiments in texts over time effectively. This approach not only refines sentiment analysis by incorporating contextual sensitivity but also offers a practical solution readily accessible to the research community.