TurEngMix: A Text Corpus and Benchmark for Turkish-English Code-Mixed Language Identification and Named Entity Recognition
This work introduces TurEngMix, a corpus of 5.5K noisy, naturally occurring social media posts rich in Turkish-English code-mixing, and constructs a new Turkish-English benchmark for code-mixed language identification and named entity recognition, comprising 15K expert-annotated tokens.
Ilayda Dogan, Phuong-Anh Nguyen-Le, Julia Mendelsohn
· 0 citations