The Multi-Intent Comprehension Model for Intelligent Dialogue Systems in Power Dispatch
Feng Wei, Dongqing Liu, Xun Zhang, Ke Wang, Yugang Zhang, Guoliang Liu, Hong Zhao
State Grid Corporation of China (China)
阅读操作
确认中在文库中上传 PDF 后可生成中文音频讲解。
摘要与影响
With the deepening digital transformation of the power industry, traditional manual dispatch modes struggle to meet the high efficiency and safety requirements of modern power grid operations. Addressing the complexity of natural language understanding in power dispatch scenarios, this paper proposes ELIDSNet, a deep learning-based joint learning model for multiintent recognition and entity extraction. This model incorporates four key improvements over the original LIDSNet architecture: it replaces the character-level CNN with the BERT-base-chinese pre-trained model, introduces a multi-layer feature fusion and residual connection mechanism, constructs a dual sequence modeling framework that synergizes BiLSTM and multi-head selfattention, and implements multi-task joint learning for intent classification and named entity recognition. Experimental results demonstrate that E-LIDSNet achieves a sample accuracy of 99.48 % on the intent recognition task, and a sample accuracy of 99.87 % on the entity recognition task, significantly outperforming traditional machine learning methods and baseline deep learning models. This model provides accurate and reliable natural language understanding capabilities for intelligent dialogue systems in power dispatch, effectively enhancing the intelligence level of power grid operations and the efficiency of operation and maintenance work.
逐年被引趋势
暂无年度引用数据
关键指标
同类平均 = 1
同领域 · 同年份 · 同类型
Google Scholar 与 OpenAlex 的被引统计范围不同,数值存在差异属正常。
AI 辅助阅读
依据:摘要
可就本文提问;依据不足时会说明。
学术脉络
学科主题
计算机 / AISpeech and dialogue systems
Topic Modeling · Natural Language Processing Techniques