Diacritics correction in Turkish with context-aware sequence to sequence modeling

Digital texts in many languages have examples of missing or misused diacritics which makes it hard for natural language processing applications to disambiguate the meaning words. Therefore, restoration is a crucial step languages. In this study we approach problem as bidirectional transformation diacritical letters and their ASCII counterparts, rather than unidirectional diacritic restoration. We propose context-aware character-level sequence model transformation. The independent sense that no language-specific feature extraction necessary other utilization word embeddings directly applicable trained Turkish correction task assessment used tweets benchmark dataset. Our best setting proposed improves state-of-the-art results terms F1 score by 4.7% on ambiguous words 1.24% over all cases.