A method of automatic generation of daily work report for power dispatching based on RoBERTa. Issue 1 (May 2020)

• Record Type: Journal Article
• Authors: Li, Jie; Zhang, Tao; Li, Hao; Lu, Jixiang; Lu, Jinjun; Li, Ziqi
• Is Part Of: IOP conference series. Volume 853:Issue 1(2020)
• Publication Date: 2020-05
• Subjects: Natural Language Processing -- Information Extraction -- Named Entity Recognition -- Pattern Matching -- RoBERTa Introduction; Materials science -- Periodicals; 620.1105
• DOI: 10.1088/1757-899X/853/1/012032
• Languages: English
• ISSNs: 1757-8981

Abstract

Abstract: Power dispatching operators need to read a large number of power dispatching logs every day and arrange them into daily work reports. It takes them a lot of time to do information extraction and transfer that information into work reports. In order to reduce the workload of power dispatching operators, we propose an intelligent method for automatic generation of daily work reports in this paper by adopting RoBERTa, which is an improved version of BERT. With the outstanding performance of RoBERTa in semantic analysis and entity recognition, the workload of power dispatching operators is reduced approximately 50% through our proposed method, which means the efficiency is greatly improved.