Omission Detection

Omission detection focuses on identifying missing information in various data modalities, including text, images, and translations. Current research employs diverse approaches, ranging from probing encoder outputs in transformer-based models to utilizing computer vision and optimal transport algorithms for word alignment in machine translation and map analysis. This field is crucial for improving the reliability and trustworthiness of AI systems, particularly in safety-critical applications like medical summarization and causal inference, where missing information can have significant consequences. The development of robust omission detection methods is driving improvements in natural language generation, machine translation, and visual question answering.