摘要
电力企业作为国家支柱型企业面临着复杂的网络舆情管理形势,为了防止突发事件所产生的不良网络舆情信息对电力企业的运行造成影响,构建大数据舆情危机管理应用成为目前研究的热点。当前网络舆情管理平台不能随着互联网随时产生的海量数据实时更新模型,处理增量数据时效性较差。文章建议将Batch SVM增量算法与Boosting算法相结合,提出了一种增量网络舆情发现方法,并将该算法与流式计算云平台进行结合,构建了一套高效的基于流式云平台的在线增量舆情发现机制。在真实数据集上构建的实验验证了所提机制的准确性和效率,在保证准确度达到85%的前提下,所提机制的处理时延较现有算法降低50%以上,可以有效实现在线舆情发现,在构建电力网络舆情管理平台方面具有较大的应用价值。
As a national backbone enterprise, power enterprises face a more complex network public opinion management situation. In order to prevent the bad network public opinion information generated by the emergency from affecting the operation of the power enterprise, the construction of big data public opinion crisis management application has become a hot research topic. The current network public opinion management platform can not update the model in real time with the massive data generated by the Internet at any time, thus the timeliness of processing incremental data is poor. To solve this problem, using a combination of Batch SVM incremental algorithm and Boosting algorithm, this paper proposes an incremental network public opinion discovery method. The algorithm is combined with the stream computing cloud platform to build an efficient incremental public opinion discovery mechanism based on stream cloud platform. The experiments using the real data set verify the accuracy and efficiency of the proposed mechanism. Under the premise of ensuring the accuracy is 85%, the processing delay of the proposed mechanism is reduced by more than 50% compared with the existing algorithm, and online public opinion can be effectively found. It has great application value in building power network public opinion management platform.
作者
韩耀廷
赵一鸣
谢炯
陈晓宇
杨浒昀
HAN Yaoting;ZHAO Yiming;XIE Jiong;CHEN Xiaoyu;YANG Huyun(Inner Mongolia Electric Power Group,Mengdian Information&Telecommunication Co.,Ltd.,Hohhot 010000,China)
出处
《电力信息与通信技术》
2019年第11期70-75,共6页
Electric Power Information and Communication Technology
