Sercan Karakaş

Fine-tuned BERTurk models outperform prompted large language models in Turkish sentiment analysis, especially when distinguishing subtle neutral sentiments.

Despite their proficiency in many language tasks, LLMs fail to leverage plausibility cues for syntactic disambiguation in the same way humans do, revealing a critical gap in commonsense reasoning.

Even a small, targeted dataset can bridge the gap in cross-dialect transfer learning for low-resource languages, significantly boosting dependency parsing accuracy.