All preferences are stored on your device.
2021
It is built from checkable properties of the record itself: whether the article has a DOI, how many independent sources confirmed its data, whether the required fields are filled in, and how recent the latest confirmation is. Disagreement between sources and a missing identifier lower it. A value of 80 or above means the record was published without further checking; 60 to 79 means it is published at reduced confidence, with a manual check advisable. A record verified by an editor by hand is never pushed below 90 by a recalculation. This scores our data about the article, not the article itself.
In this paper, part-of-speech tagging on Kazakh text has been implemented using an LSTM neural network with a bidirectional modifier. A quite simple and fast tagger has been built, tested and evaluated on the self-collected dataset of Kazakh sentences with an accuracy of 94 %. There was addressed basically the problem of tagging having a small number of training samples (around 100 Kazakh sentences). It has been shown that quite good accuracy can be achieved in this situation.