A System Review on Bootstrapping Information Extraction

Focusing on the supervision problems caused by high-cost and low-quality labeling in information extraction, we provided a detailed overview of the various approaches that were proposed to solve the sub-tasks of bootstrapping information extraction. We summarized current principal approaches and depicted the specific issues addressed in recent research. To provide inspiration and reference for similar studies in terms of mainstream data sources, evaluation specifications and applications, we summarized the relevant datasets, evaluation metrics, and systematic applications of bootstrapping information extraction. In addition, we reflected on the remaining problems of bootstrapping information extraction and highlighted some directions for future work.