由于数据规模的快速增长,高效用序列模式挖掘算法效率严重下降.针对这种情况,提出基于Map Reduce的高效用序列模式挖掘算法Hus Ma R.算法基于Map Reduce框架,使用效用矩阵高效地生成候选项;使用随机映射策略均衡计算资源;使用基于领域...由于数据规模的快速增长,高效用序列模式挖掘算法效率严重下降.针对这种情况,提出基于Map Reduce的高效用序列模式挖掘算法Hus Ma R.算法基于Map Reduce框架,使用效用矩阵高效地生成候选项;使用随机映射策略均衡计算资源;使用基于领域的剪枝策略来防止组合爆炸.实验结果表明,在大规模数据集下,算法取得了较高的并行效率.展开更多
电商网站的兴起与用户在线购物习惯的形成,带来了海量的在线消费行为数据.如何从这些行为数据(如点击数据)中建模用户对相似产品的比较和选择过程,进而准确预测用户的兴趣偏好和购买行为,对于提高产品的购买转化率具有重要意义.针对这...电商网站的兴起与用户在线购物习惯的形成,带来了海量的在线消费行为数据.如何从这些行为数据(如点击数据)中建模用户对相似产品的比较和选择过程,进而准确预测用户的兴趣偏好和购买行为,对于提高产品的购买转化率具有重要意义.针对这一问题,提出了基于用户行为序列数据和选择模型的在线购买预测解决方案.具体而言,1)使用行为序列效用函数估计用户在购买周期(session)中的最佳替代商品,然后对购买商品和最佳替代商品建立基于潜在因子的选择模型(latent factor based choice model,LF-CM),从而得到用户的购买偏好,实现对用户购买行为的预测.更进一步,为了充分地利用用户在每个购买周期的所有选择和比较信息,提高预测精度;2)提出了一种可以作用于购买周期内所有商品的排序学习模型(latent factor and sequence based choice model,LFS-CM),它通过融合潜在因子和行为序列的效用函数,提高了购买预测的精度;3)使用大规模真实数据集在分布式环境下进行了实验,并与参照算法进行了对比,证实了所提出的2个方法在用户在线购买预测上的有效性.展开更多
The role of biophysical variables in constructing community structure changes with the time since fire.The major objective of this study is to verify the transition stage and its underlying variables for the postfire ...The role of biophysical variables in constructing community structure changes with the time since fire.The major objective of this study is to verify the transition stage and its underlying variables for the postfire forest and soil microbial function in the boreal forested area of China.A 50-year fire chronosequence was presented,and biomass of forbs,shrubs and woody plants was separately weighted to assess their contribution to the whole community with the year since fire(YSF).Simultaneously,soil biophysical properties were measured for stands in different time periods after fire.Soil microbial functions,i.e.growth efficiency(GE)and carbon use efficiency(CUE),were calculated based on ecoenzymatic and soil nutrient stoichiometry.In terms of vegetative structure,forbs’proportion decreased from 75%to 1.5%,but the proportion of woody plants increased from 0.04%to 70%across this fire chronosequence.GE and CUE of soil microorganisms averaged 0.242 and 0.236 and were significantly higher in 9,15 and 31 YSF than in 2 and 3 YSF.Soil metal content was significantly increased at the late stage of this fire chronosequence,and soil calcium content showed a positive correlation with woody plant biomass and a negative correlation with soil microbial function.Overall,the present work highlights that the time period of 15 and 31 YSF is a hallmark stage for aboveground vegetative structure and soil microbial function to change in different trends and that the calcium content may partly account for these two divergent trajectories.展开更多
文摘由于数据规模的快速增长,高效用序列模式挖掘算法效率严重下降.针对这种情况,提出基于Map Reduce的高效用序列模式挖掘算法Hus Ma R.算法基于Map Reduce框架,使用效用矩阵高效地生成候选项;使用随机映射策略均衡计算资源;使用基于领域的剪枝策略来防止组合爆炸.实验结果表明,在大规模数据集下,算法取得了较高的并行效率.
文摘电商网站的兴起与用户在线购物习惯的形成,带来了海量的在线消费行为数据.如何从这些行为数据(如点击数据)中建模用户对相似产品的比较和选择过程,进而准确预测用户的兴趣偏好和购买行为,对于提高产品的购买转化率具有重要意义.针对这一问题,提出了基于用户行为序列数据和选择模型的在线购买预测解决方案.具体而言,1)使用行为序列效用函数估计用户在购买周期(session)中的最佳替代商品,然后对购买商品和最佳替代商品建立基于潜在因子的选择模型(latent factor based choice model,LF-CM),从而得到用户的购买偏好,实现对用户购买行为的预测.更进一步,为了充分地利用用户在每个购买周期的所有选择和比较信息,提高预测精度;2)提出了一种可以作用于购买周期内所有商品的排序学习模型(latent factor and sequence based choice model,LFS-CM),它通过融合潜在因子和行为序列的效用函数,提高了购买预测的精度;3)使用大规模真实数据集在分布式环境下进行了实验,并与参照算法进行了对比,证实了所提出的2个方法在用户在线购买预测上的有效性.
基金supported jointly by the Key Project of National Key Research and Development Plan(grant no.2017YFC0504002)the Fundamental Research Funds for the Central University(grant no.2015ZCQ-LX-03).
文摘The role of biophysical variables in constructing community structure changes with the time since fire.The major objective of this study is to verify the transition stage and its underlying variables for the postfire forest and soil microbial function in the boreal forested area of China.A 50-year fire chronosequence was presented,and biomass of forbs,shrubs and woody plants was separately weighted to assess their contribution to the whole community with the year since fire(YSF).Simultaneously,soil biophysical properties were measured for stands in different time periods after fire.Soil microbial functions,i.e.growth efficiency(GE)and carbon use efficiency(CUE),were calculated based on ecoenzymatic and soil nutrient stoichiometry.In terms of vegetative structure,forbs’proportion decreased from 75%to 1.5%,but the proportion of woody plants increased from 0.04%to 70%across this fire chronosequence.GE and CUE of soil microorganisms averaged 0.242 and 0.236 and were significantly higher in 9,15 and 31 YSF than in 2 and 3 YSF.Soil metal content was significantly increased at the late stage of this fire chronosequence,and soil calcium content showed a positive correlation with woody plant biomass and a negative correlation with soil microbial function.Overall,the present work highlights that the time period of 15 and 31 YSF is a hallmark stage for aboveground vegetative structure and soil microbial function to change in different trends and that the calcium content may partly account for these two divergent trajectories.
