| 引用本文: | 王海瑶,王慧芳,胡俊华,徐积全,李建红,何奔腾.基于对话文本智能挖掘的电力设备供应商评价[J].电力自动化设备,2021,41(7): |
| WANG Haiyao,WANG Huifang,HU Junhua,XU Jiquan,LI Jianhong,HE Benteng.Evaluation of power equipment suppliers based on intelligent mining of electric power dialogue text[J].Electric Power Automation Equipment,2021,41(7): |
|
| 摘要: |
| 目前电网企业的电力设备供应商满意度评价主要依赖于人工统计和指标计算，其准确性受评价人员和评价内容的影响较大。以电力业务平台的对话文本为研究对象，在扩充已有电力本体词典的词条和属性的基础上，建立了基于文本挖掘技术的电力设备供应商评价模型。首先提出了基于Transformer的双向编码器下句预测与余弦相似度加权的单轮对话文本下句预测分析方法，建立了对话中断交叉处理流程和供应商识别规则，实现了电力对话文本主题归纳；然后考虑对话文本语义情感的复杂性，提出了对话情感分析规则，建立了供应商评价模型。最后通过算例验证了所提方法的准确性，结果表明基于对话文本智能挖掘的电力设备供应商评价，具有可行性和有效性，可以作为目前评价方法的有益补充。 |
| 关键词: 电力设备 文本智能挖掘 对话文本 下句预测 情感分析 供应商评价 |
| DOI：10.16081/j.epae.202104009 |
| 分类号:F274 |
| 基金项目: |
|
| Evaluation of power equipment suppliers based on intelligent mining of electric power dialogue text |
|
WANG Haiyao1, WANG Huifang1, HU Junhua2, XU Jiquan2, LI Jianhong3, HE Benteng1
|
|
1.College of Electrical Engineering, Zhejiang University, Hangzhou 310027, China;2.State Grid Zhejiang Electric Power Company, Hangzhou 310007, China;3.Zhejiang Huayun Information Technology Co.,Ltd.,Hangzhou 310012, China
|
| Abstract: |
| Currently the satisfaction evaluation of power equipment suppliers mainly relies on manual statistics and index calculation, and its accuracy is greatly affected by evaluators and evaluation contents. The dialogue text of the electric power service platform is used as the research object. Taking the dialogue text of power service platform as the research object, the evaluation model of power equipment supplier based on text mining technology is established on the basis of expanding the entries and attributes of the existing electric power ontology dictionary. Firstly, the next sentence predictive analysis method of single-round dialogue text based on BERT(Bidirectional Encoder Representations from Transformers) NSP(Next Sentence Prediction) and cosine similarity weighted is proposed. The dialogue interruption cross-handling process and supplier identification rules are established to realize theme induction of electric power dialogue text. Then, considering the complexity of the semantic sentiment of the dialogue text, the dialogue sentiment analysis rules are proposed for establishing the supplier evaluation model. Finally, the accuracy of the proposed method is verified by an example. Results indicate that the evaluation of power equipment suppliers based on the intelligent mining of dialogue text is feasible and effective, and can be used as a useful supplement to current evaluation methods. |
| Key words: electric power equipment intelligent text mining dialogue text next sentence prediction sentiment analysis supplier evaluation |
