# Automatic verbal aggression detection for Russian and American imageboards

**Authors:** Denis Gordeev  
**arXiv:** [1604.06648](https://arxiv.org/abs/1604.06648)  
**Date:** 2016-04-22  
**Summary page:** https://www.summarizepaper.com/en/paper/1604.06648v1/

## Key points

- Aggression in online communities, especially on anonymous imageboards, is a growing concern
- A study was conducted to automatically detect verbal expressions of aggression on American and Russian imageboards
- The study used a set of 1,802,789 messages and applied the machine learning algorithm word2vec to detect aggression
- The algorithm achieved a decent level of accuracy for English (88%), but improvements are needed for Russian language detection
- Developing effective methods for identifying and addressing aggressive behavior in online communities is important for ensuring safety and security on these platforms.

## Summary

The problem of aggression in online communities is a growing concern, particularly on anonymous forums known as imageboards. These platforms are notorious for their aggressive and deviant behavior, even when compared to other online communities. In response to this issue, a study was conducted to explore ways of automatically detecting verbal expressions of aggression on the most popular American (4chan.org) and Russian (2ch.hk) imageboards. The study utilized a set of 1,802,789 messages and applied the machine learning algorithm word2vec to detect the state of aggression. The results showed that the algorithm was able to achieve a decent level of accuracy for English (88%), but further improvements are needed for Russian language detection. This research highlights the importance of developing effective methods for identifying and addressing aggressive behavior in online communities in order to ensure safety and security on these platforms.