Class Incremental NER

Class incremental named entity recognition (NER) focuses on adapting NER models to continuously learn new entity types without forgetting previously learned ones, mirroring real-world scenarios where entity categories expand over time. Current research emphasizes improving robustness to noisy data, particularly in distantly supervised settings, and developing methods that effectively leverage limited labeled data for new entity types, including techniques like in-context learning and data augmentation with large language models. These advancements are crucial for building more adaptable and scalable NER systems, improving performance in domains with evolving entity categories and reducing the reliance on extensive, manually annotated datasets.