社交网络中,消息的爆发预测属于社交网络流行动态分析的范畴,是社会计算领域的研究热点之一.通过利用基于深度循环神经网络对社交消息的传播过程进行建模,提出了SMOP(social messages outbreak prediction model based on recurrent neu...社交网络中,消息的爆发预测属于社交网络流行动态分析的范畴,是社会计算领域的研究热点之一.通过利用基于深度循环神经网络对社交消息的传播过程进行建模,提出了SMOP(social messages outbreak prediction model based on recurrent neural network)模型.与传统的基于机器学习的模型相比,SMOP直接对消息转发的到达过程进行建模,避免了传统方法中繁琐的特征工程;与基于点随机过程的模型相比,SMOP可以自动学习消息传播过程的速率函数,不需要手动定义消息传播速率的特征函数,具有较强的数据场景适应性.另外,SMOP采用了时间向量和用户向量的输入表示方法,将时间的周期性和用户的兴趣偏好建模到传播过程之中,提升了SMOP的预测效果.在Twitter和新浪微博数据集上的实验结果均表明,SMOP具有优良的数据适应能力,可以在消息传播的早期(0.5h),以较高的F1值预测某条社交消息是否爆发,验证了模型的有效性.展开更多
Rock burst is a severe disaster in mining and underground engineering,and it is important to predict the rock burst risk for minimizing the loss during the constructing process.The rock burst proneness was connected w...Rock burst is a severe disaster in mining and underground engineering,and it is important to predict the rock burst risk for minimizing the loss during the constructing process.The rock burst proneness was connected with the acoustic emission(AE) parameter in this work,which contributes to predicting the rock burst risk using AE technique.Primarily,a rock burst proneness index is proposed,and it just depends on the heterogeneous degree of rock material.Then,the quantificational formula between the value of rock burst proneness index and the accumulative AE counts in rock sample under uniaxial compression with axial strain increases is developed.Finally,three kinds of rock samples,i.e.,granite,limestone and sandstone are tested about variation of the accumulative AE counts under uniaxial compression,and the test data are fitted well with the theoretic formula.展开更多
文摘社交网络中,消息的爆发预测属于社交网络流行动态分析的范畴,是社会计算领域的研究热点之一.通过利用基于深度循环神经网络对社交消息的传播过程进行建模,提出了SMOP(social messages outbreak prediction model based on recurrent neural network)模型.与传统的基于机器学习的模型相比,SMOP直接对消息转发的到达过程进行建模,避免了传统方法中繁琐的特征工程;与基于点随机过程的模型相比,SMOP可以自动学习消息传播过程的速率函数,不需要手动定义消息传播速率的特征函数,具有较强的数据场景适应性.另外,SMOP采用了时间向量和用户向量的输入表示方法,将时间的周期性和用户的兴趣偏好建模到传播过程之中,提升了SMOP的预测效果.在Twitter和新浪微博数据集上的实验结果均表明,SMOP具有优良的数据适应能力,可以在消息传播的早期(0.5h),以较高的F1值预测某条社交消息是否爆发,验证了模型的有效性.
基金Project(2010CB226804)supported by the National Basic Research Program(973 Program)of ChinaProject(11202108)supported by the National Natural Science Foundation of ChinaProject(BK20130189)supported by the Natural Science Foundation of Jiangsu Province,China
文摘Rock burst is a severe disaster in mining and underground engineering,and it is important to predict the rock burst risk for minimizing the loss during the constructing process.The rock burst proneness was connected with the acoustic emission(AE) parameter in this work,which contributes to predicting the rock burst risk using AE technique.Primarily,a rock burst proneness index is proposed,and it just depends on the heterogeneous degree of rock material.Then,the quantificational formula between the value of rock burst proneness index and the accumulative AE counts in rock sample under uniaxial compression with axial strain increases is developed.Finally,three kinds of rock samples,i.e.,granite,limestone and sandstone are tested about variation of the accumulative AE counts under uniaxial compression,and the test data are fitted well with the theoretic formula.
