基于移动终端的高校设备报修重复信息抽取研究  

Research on repetitive information extraction of university equipment repair based on mobile terminal

作　　者：李晖 秦广久[2,3] 王明刚[1] 王立辉 谭启忠 LI Hui;QIN Guangjiu;WANG Minggang;WANG Lihui;TAN Qizhong(Office of Logistics Management,Shandong Jianzhu University,Jinan 250101,China;Office of Asset Management,Shandong Jianzhu University,Jinan 250101,China;School of Mechanical and Electrical Engineering,Shandong Jianzhu University,Jinan 250101,China)

机构地区：[1]山东建筑大学后勤管理处,山东济南250101 [2]山东建筑大学资产管理处,山东济南250101 [3]山东建筑大学机电工程学院,山东济南250101

出　　处：《电子设计工程》2025年第7期81-84,89,共5页Electronic Design Engineering

基　　金：2023年山东省学校后勤协会项目(HQXH2023-007,HQXH2023-087);2023-2024年中国教育后勤协会项目(YBKT 2024104)。

摘　　要：高校设备报修数据具有多模态和复杂性的特征,当数据量过大而不能存储在单个节点时,需要对数据增量压缩,以实现对故障信息的多层级转达,但增量压缩会产生重叠增量,只关注关键词频次而不考虑零散化的多维度特征,会导致重复信息抽取的F1分数较低。故提出基于移动终端的高校设备报修重复信息抽取研究。设计一种高效的数据传输协议,该协议通过报修信息格式标准化和压缩处理,以解决数据量不断增大的问题,优化移动终端与服务器之间的数据传输过程。将压缩后的报修信息从移动终端传输至服务器,提取并整合包括关键词频次、文本长度、平均句子长度以及语义特征在内的多维度零散化报修信息特征,并输入到SVM模型中,实现重复信息抽取。测试结果显示,应用所设计方法后报修重复信息抽取的F1分数最大值达到了0.98。The repair data of university equipment is characterized by multi-mode and complexity.When the amount of data is too large to be stored in a single node,incremental compression of the data is required to achieve multi-level transmission of fault information.However,incremental compression will produce overlapping increment,and only focusing on keyword frequency without considering the multidimensional characteristics of fragmentation will lead to a low F1 score of repeated information extraction.Therefore,the research of repeated information extraction of college equipment repair based on mobile terminal is proposed.An efficient data transmission protocol is designed,which can solve the problem of increasing data volume and optimize the data transmission process between mobile terminal and server through standardization and compression of repair information format.The compressed repair information was transmitted from the mobile terminal to the server,and the multi-dimensional fragmented repair information features,including keyword frequency,text length,average sentence length and semantic features,were extracted and integrated,and input into the value SVM model to achieve repeated information extraction.The test results show that the maximum F1 score extracted from the repeated information after the application of the design method reaches 0.98.

关 键 词：报修信息 高校设备 特征提取 重复信息抽取 移动终端 

分 类 号：TN925[电子电信—通信与信息系统]