From Text-based to Image-based Named Entity Recognition: A Survey

Anonymous

Abstract: Named Entity Recognition (NER) is a task to recognize mentions of entities such as person, location, drug, time, biological protein, etc. NER serves as a key component for a number of Natural Language Processing applications including machine translation, entity linking, information retrieval, question answering, etc. Traditional NER is limited to identifying and categorizing entities in text-based data. In recent decades, as Document Image Understanding emerges as a new research area, recognizing entities from image-based documents becomes a new goal in Artificial Intelligence. This paper investigates both text-based and image-based NER through reviewing a series of significant and relevant tasks, datasets, methods, and evaluations, with the goal to present a clear overview of the field. Further, the survey provides a reflection on the field by discussing the challenges and future directions in NER.

Paper Type: long

Research Area: Information Extraction