CAJ | 학술논문

微博热点预测是一类贫信息、小样本、不确定性的复杂预测问题。为了提高微博热点预测精度，提出一种基于支持向量机修正ARIMA误差的微博热点预测模型（ARIMA-SVM）。首先对微博数据进行预处理、提取主题构建网络微博热点时间序列，然后采用ARIMA建立网络微博热点预测模型，并采用支持向量机对ARIMA预测误差进行修正得到微博热点最终预测结果，最后模型性能进行仿真测试。结果表明，相对于传统预测模型，ARIMA-SVM提高了网络微博热点的预测精度，预测结果具有一定实用价值。
미박열점예측시일류빈신식、소양본、불학정성적복잡예측문제。위료제고미박열점예측정도，제출일충기우지지향량궤수정ARIMA오차적미박열점예측모형（ARIMA-SVM）。수선대미박수거진행예처리、제취주제구건망락미박열점시간서렬，연후채용ARIMA건립망락미박열점예측모형，병채용지지향량궤대ARIMA예측오차진행수정득도미박열점최종예측결과，최후모형성능진행방진측시。결과표명，상대우전통예측모형，ARIMA-SVM제고료망락미박열점적예측정도，예측결과구유일정실용개치。
Microblogginghottopicpredictionisthecomplexpredictionproblemwithpoorinformation,smallsampleanduncertainty.In order to improve the precision of microblogging hot topic prediction,we propose a microblogging hot topic prediction model which is based on correcting residual error of ARIMA by support vector machine (ARIMA-SVM).First,we pre-process the microblogging data and extract themes to construct the time series of microblogging hot topics;then we use ARIMA to build the prediction model of microblogging hot topics, and use support vector machine to correct the prediction errors of ARIMA to get the final prediction result of microblogging hot topics;at last we test the model performance by simulation.Simulation results show that the ARIMA-SVM improves the precision of microblogging hot topic prediction compared with traditional prediction model,and the prediction results has certain practical value.