摘要
针对互联网突发事件监测快速响应场景下文本信息非结构化的特点,选择互联网众源突发事件文本数据,分析在突发事件监测信息分类中的朴素贝叶斯、BP神经网络、支持向量机和SGD支持向量机4种分类器的计算速度和分类效果。结果表明,支持向量机在突发事件监测分类问题上有更好的分类效果,综合计算时间随机梯度下降-支持向量机具有相对快速且较准确的分类能力。
Aiming at the unstructured characteristics of text information in the scene of rapid response to Internet emergencies monitoring,analyzing the calculation speed and classification effect of naive Bayes,BP neural network,support vector machine and SGD support vector machine in the classification of emergency monitoring information is the primary targets.The results indicate that the support vector machines have a better classification capability on monitoring emergency and processing classification.Considering the stochastic gradient descent of the calculation,time-support vector machines capable of doing fast and accurate classification.
作者
张昱
罗年学
ZHANG Yu;LUO Nianxue(School of Geodesy and Geomatics,Wuhan University,Wuhan 430079,China)
基金
国家重点研发计划资助项目(2018YFC0807000)
作者简介
张昱(1995-),女,黑龙江哈尔滨人,武汉大学测绘学院硕士研究生.
