laugustyniak / sentiment-aspects

Aspect-based Sentiment Analysis for News Bias - RENOIR Project

Aspect-based Sentiment Analysis

Introduction

Repository consist of works on aspect-based sentiment analysis, compared and visualized how the attitude of chosen aspects (especially persons, countries) changed over time. In the analysis aspect-based sentiment analysis tool was used; it was described and published in 2016 [1]. PWR used a subset of events related to Brexit gathered through the EventRegistry system http://eventregistry.org/. One of the unusual results of the analysis was in it showing that whether the term U.K. was used related attitudes was more negative, on the other hand, the term UK (without the dots) presented a much more positive sentiment. We investigated such phenomena and found that term U.K. is mostly used by United States news sources (such as USA Today). United Kingdom news sources use term UK instead.

Demo

The analysis of aspects over time, comparison of different timeframes and various aspects may be conducted on Dashboard. Credentials:

login: aspects

password: contact me to get password

References

[1] Augustyniak, Ł., Rajda, K., Kajdanowicz, T. Method for aspect-based sentiment annotation using rhetorical analysis, ACIIDS 2017.

##Acknowledgment:

The work was partially supported by The National Science Centre, decision no. DEC-2013/09/B, The National Science Centre Preludium grant and from the European Union’s Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No 691152 (RENOIR project). Research financed under the fund for supporting internationally co-financed projects in 2016-2019.