Application of Machine Writing Based on Artificial Intelligence in Meteorological Services
- Resource Type
- Conference
- Authors
- Zhang, Qi; Zhou, Yiping; Liu, Juan; Li, Wei; Li, Jian
- Source
- 2023 International Conference on Applied Physics and Computing (ICAPC) ICAPC Applied Physics and Computing (ICAPC), 2023 International Conference on. :509-513 Dec, 2023
- Subject
- Computing and Processing
Training
Transfer learning
Weather forecasting
Writing
Market research
Data models
User experience
weather service
machine writing
artificial intelligence
CNN
- Language
How to solve the problems of time-consuming, labor-intensive and subjective factors in manual writing in meteorological services is an important research topic. Machine writing based on artificial intelligence has become a way to improve the quality of meteorological information generation. This study explores the application of artificial intelligence-based machine writing in meteorological services by using a pretrained language model as the core technology. The pre-trained language model is trained on large-scale meteorological text data and has strong language understanding and generation capabilities. In the pre-training stage, the model learns language representation from unlabeled data through self-supervised learning. Then, the pre-trained model is applied to specific meteorological tasks through fine-tuning or transfer learning. Experimental results show that the prediction accuracy of this method in meteorological services is between 90% and 98%. Machine writing based on artificial intelligence has broad application prospects in meteorological services.