Denis Gordeev

Denis Gordeev, Alexey Rey, Dmitry Shagarov

Unsupervised cross-lingual embeddings mapping has provided a unique tool for completely unsupervised translation even for languages with different scripts. In this work we use this method for the task of unsupervised cross-lingual matching of product classifications. Our work also investigates limitations of unsupervised vector alignment and we also suggest two other techniques for aligning product classifications based on their descriptions: using hierarchical information and translations.

Rodmonga Potapova, Denis Gordeev

In this article we study verbal expression of aggression and its detection using machine learning and neural networks methods. We test our results using our corpora of messages from anonymous imageboards. We also compare Random forest classifier with convolutional neural network for "Movie reviews with one sentence per review" corpus.

Denis Gordeev

The problem of aggression for Internet communities is rampant. Anonymous forums usually called imageboards are notorious for their aggressive and deviant behaviour even in comparison with other Internet communities. This study is aimed at studying ways of automatic detection of verbal expression of aggression for the most popular American (4chan.org) and Russian (2ch.hk) imageboards. A set of 1,802,789 messages was used for this study. The machine learning algorithm word2vec was applied to detect the state of aggression. A decent result is obtained for English (88%), the results for Russian are yet to be improved.

Rodmonga Potapova, Denis Gordeev

This article deals with the analysis of the semantic content of the anonymous Russian-speaking forum 2ch.hk, different verbal means of expressing of the emotional state of aggression are revealed for this site, and aggression is classified by its directions. The lexis of different Russian-and English- speaking anonymous forums (2ch.hk and iichan.hk, 4chan.org) and public community "MDK" of the Russian-speaking social network VK is analyzed and compared with the Open Corpus of the Russian language (Opencorpora.org and Brown corpus). The analysis shows that anonymity has no influence on the amount of invective items usage. The effectiveness of moderation was shown for anonymous forums. It was established that Russian obscene lexis was used to express the emotional state of aggression only in 60.4% of cases for 2ch.hk. These preliminary results show that the Russian obscene lexis on the Internet does not have direct dependence on the emotional state of aggression.