Annotated Uzbek Sentences Dataset for Binary Sentiment Classification

As part of this study we compiled the first open corpus of short Uzbek sentences annotated for binary sentiment classification. Corpus size and composition - 4 676 sentences in total: 3 042 — Positive, 1 634 — Negative. - All texts are written in the modern Uzbek Latin alphabet and consist mainly of short everyday utterances (average length ≈ 6.0 tokens). Raw class files: UZ_positive.txt and UZ_negative.txt (one sentence per line). Upd (Version 2): Now it is available to download .csv format file, which has both negative and positive sentences and phrases.