Who’s (un)popular? When? Why?

Popularity of various public figures changes through time. It is possible to infer the popularity of a person by performing sentiment analysis on news articles. Visualising popularity can be very useful as it can provide insights to the causes of changes in popularity.

Some people are liked, respected, and followed by the community, while others are possibly equally intriguing but in a negative way. Who’s popular and who isn’t? How does popularity of a person change through time? What are the causes for changes in the popularity? Answering these questions can help us answer the most practical question of them all — what can we do to affect a person’s popularity rating?

Coming back to the area of NLP, one can’t help notice that news articles are usually full of sentiment towards a person giving a good hint about their current popularity. Slavomjer is a system that analyzes and aggregates opinions towards a person expressed in news articles. State-of-the-art methods for sentiment analysis are used for the analysis. Popularity timelines for people are plotted on a graph. This allows popularity analysis of a particular person, and checking the corresponding articles can determine the causes of popularity gain or loss. It also provides additional insights by comparing the popularity of several different people through a time period.

The system crawls news articles from several Croatian news portals and is available online.

Slavomjer was developed by a team of TakeLab students:

* Petra Almić
* Siniša Biđin (thanks for setting the system up online)
* Luka Krajcar

This work was done under the supervision of doc. dr. sc. Jan Šnajder and dr. sc. Goran Glavaš.