Mechanical Tsar


RUSSE, the first international workshop on Russian semantic similarity evaluation conducted in 2014–2015, needed a human judgement dataset. In order to collect such judgements, a couple of scripts have been written in PHP and SQL.

RUSSE annotation in 2014.

In 2015, it has become necessary to use the similar approach for data collection, but for quite a different domain. In order to satisfy such requirements, Mechanical Tsar, an open source crowdsourcing engine for mechanized labor, has been written in Java.

Early version of Mechanical Tsar dashboard.