电网人机对话系统短文本用户意图自动识别方法  

Automatic identification method of user’s short text intention in power grid human⁃machine dialogue system

作　　者：马占海 张俊超 田光欣 MA Zhanhai;ZHANG Junchao;TIAN Guangxin(State Grid Qinghai Information&Telecommunication Compan,Xining 810008,China)

机构地区：[1]国网青海省电力公司信息通信公司,青海西宁810008

出　　处：《电子设计工程》2023年第17期128-131,136,共5页Electronic Design Engineering

摘　　要：为避免电量信号出现过度消耗的情况,从而实现对用户主机短文本信息意图的精准识别与分析,提出电网人机对话系统短文本用户意图自动识别方法。借助Hadoop框架结构,定义与短文本字符串相关的匹配条件,根据电量信号的传输意图分析结果,完成电网人机对话系统中的短文本用户意图处理。在此基础上,确定用户意图信息的相关度估算结果,分别计算识别窗口的宽度与长度数值水平,实现电网人机对话系统短文本用户意图自动识别方法的顺利应用。实验结果表明,与多级负荷聚类型识别方法相比,自动识别方法能够准确面对用户意图,在控制电量信号消耗量的同时,得到准确的短文本信息识别与分析结果,更符合实际应用需求。In order to avoid excessive consumption of power signal and realize accurate identification and analysis of user’s short text message intention,an automatic identification method of user’s short text intention in power grid human⁃machine dialogue system is proposed.With the help of Hadoop framework,the matching conditions related to the short text string are defined.According to the analysis results of transmission intention of electric quantity signal,the short text user intention processing in the human⁃machine dialogue system of power grid is completed.On this basis,the correlation estimation results of user intention information are determined,and the numerical levels of the width and length of recognition Windows are calculated respectively to realize the smooth application of automatic user intention recognition method in short text of power human⁃machine dialogue system.The experimental results show that compared with the multi⁃stage load clustering recognition method,the automatic recognition method can accurately face the user’s intention,and obtain accurate short text information recognition and analysis results while controlling the power signal consumption,which is more in line with the practical application requirements.

关 键 词：电网系统 人机对话 短文本 用户意图 自动识别 Hadoop框架 

分 类 号：TN919[电子电信—通信与信息系统]