DebunkEU.org analysts Aleksandra Michałowska-Kubś and Jakub Kubś conducted an analysis of social media posts related to sanctions imposed on the transit of goods to Kaliningrad. They used NamSor machine learning classification to assign a likely country of origin for the social account names (real or fake).
We’ve used a non-supervised clustering algorithm to identify Russian (Ivanov, Popov), Korean (Kim, Pak) and other names in Kazakhstan and differentiate them from Kakakh names (AZAMAT ABDRAKHMANOV, ERLAN AHMETOV, AIGUL ABDRAKHMANOVA / AYGUL ABDRAKHMANOVA …)
Romanization of Japanese names is easy, but translating a Japanese name back to its original form in Kanji with the correct probabilities is hard. There are many Kanji variants for a single Japanese name in its romanized form.
We’ve released a new version of the opensource Java Naive Bayes Classifier (JNBC), so it can now run on RocksDB