| 基于DBM的电力投诉工单分类的应用研究 |
投稿时间：2019-05-20 修订日期：2019-07-24  点此下载全文 |
| 引用本文： |
| 摘要点击次数: 173 |
| 全文下载次数: 0 |
|
| 基金项目:国家自然科学基金资助项目(No.51277015)。 |
|
| 中文摘要:近年来，随着经济的发展与社会的进步，用户寻求不断完善的用电服务以及对维权意识不断提高，供电企业应对用户投诉行为进行深度分析，从而发现供电业务的不足，进而完善供电服务，满足用户的各种需求。因此，提出了基于深度玻尔兹曼机的电力投诉工单识别分类模型。首先对投诉工单数据进行数据清洗，对处理后的数据使用Jieba进行分词并制作字典，再运用BoW模型对对所分词向量化处理提取文本特征。进一步地，通过TF-IDF算法找出关键词以及余弦相似度计算训练、测试文档间的相似度；最后使用深度玻尔兹曼机对投诉工单进行分类。实验证明，分类的准确度达到80%，有效地缓解电力部门的工作压力，提高工作效率。 |
| 中文关键词:投诉 TF-IDF DBM 文本分类 |
| |
| Application Research on Classification of Power Complaint Work Order Based on DBM |
|
|
| Abstract:In recent years, with the development of the economy and the progress of society, users are seeking continuous improvement of electricity service and awareness of rights protection. The power supply enterprise should conduct an in-depth analysis of the user's complaint behavior, so as to discover the shortage of the power supply business， and then improve the power supply service to meet the various needs of users. Therefore, a classification model based on the deep boltzmann machines for power complaint work order identification is proposed. First, the data of the complaint work order data is cleaned and use the Jieba algorithm to segment the processed data, and create a dictionary. Then the BoW model is used to extract the text feature from the segmentation vectorization process. Further, the TF-IDF algorithm is used to find keywords and cosine similarity calculations to calculate the similarity between training and test documents. Finally, the deep boltzmann machine is used to classify the complaint work orders. The experiment proves that the accuracy of the classification reaches 80%, effectively alleviating the work pressure of the power sector and improving work efficiency. |
| keywords:complaint TF-IDF Deep Boltzmann Machine (DBM) Text classification |
| 查看全文 查看/发表评论 下载pdf阅读器 |
位访问者,今日访问
