桌面云坐席系统电力客户工单信息精准查询方法  

Accurate Query Method of Power Customer Work Order Information in Desktop Cloud Agent System

作　　者：杨维 张浩 张才俊 曹璐 曾月阳 徐强 YANG Wei;ZHANG Hao;ZHANG Caijun;CAO Lu;ZENG Yueyang;XU Qiang(State Grid Customer Service Centre,Tianjin 300300,China;Beijing China-Power Information Technology Co.,Ltd.,Beijing 100085,China)

机构地区：[1]国家电网有限公司客户服务中心,天津300300 [2]北京中电普华信息技术有限公司,北京100085

出　　处：《微型电脑应用》2023年第10期172-175,共4页Microcomputer Applications

摘　　要：为了提升特征提取关联度和查询支持度,实现电力客户工单信息高精度高加速比查询,提出了桌面云坐席系统电力客户工单信息精准查询方法。该方法采用条件随机场分词算法实现电力客户工单信息文本的自动分词,并通过改进卡方特征选择算法获得最佳特征,然后将特征作为VSM的列矩阵变量,并依据TF-IDF算法计算特征向量的权重,最后整合特征的先验概率与重要度作为朴素贝叶斯分类器输入值,实现信息精准分类,最终完成电力客户工单信息精准查询。经实验验证:该方法提取特征具有较高的词汇关联度,查询支持度较高,遗漏概率低,能够实现精准的电力客户工单信息查询。In order to improve the relevance and query support of feature extraction,and achieve high-precision and high speedup query of power customer work order information,a precise query method of power customer work order information in desktop cloud agent system is proposed.In this method,conditional random field segmentation algorithm is used to realize automatic word segmentation of power customer work order information text,and the best feature is obtained by improvingχ2 feature selection algorithm.Then,the feature is taken as the column matrix variable of VSM,and the weight of feature vector is calculated according to TF-IDF algorithm,and the prior probability and importance of the last integrated feature are taken as the input value of naive Bayesian classifier to realize information segmentation information accurate classification.This finally completes the power customer work order information accurate query.The experimental results show that the feature extraction method has high lexical relevance,high query support,low omission probability,and can achieve accurate power customer work order information query.

关 键 词：桌面云坐席系统 电力客户 工单信息 精准查询 分词算法 

分 类 号：TP399[自动化与计算机技术—计算机应用技术]