A Survey on Deep Learning Techniques for Joint Named Entities and Relation Extraction

Abstract

Named Entity Recognition (NER) and Relation Extraction (RE) are two principal subtasks of knowledge-based systems that extract meaningful information from unstructured text. With Recent advances in Deep Learning techniques, new models use Joint Named Entities and Relation Extraction (JNERE) techniques that simultaneously accomplish NER and RE subtasks. These models avoid the drawbacks of using the traditional pipeline method. As contributions of our study to the other related works, we specifically survey JNERE techniques. The reason for not focusing on pipeline methods or other older techniques is the recent advances of JNERE methods in achieving the state-of-art results for most databases. Additionally, we provide a comprehensive report on the embedding techniques and datasets available for this task. Finally, we discuss the approaches and how they imnpoved the results.