针对传统协同过滤(CF)存在的数据稀疏和冷启动的问题以及在矩阵分解方法生成结果矩阵的过程中由于各种变换产生误差的问题,提出一种混合信息增强的低秩稀疏矩阵分解(LSMF)论文推荐方法。首先,利用预训练的文档级表示学习和引文感知转换...针对传统协同过滤(CF)存在的数据稀疏和冷启动的问题以及在矩阵分解方法生成结果矩阵的过程中由于各种变换产生误差的问题,提出一种混合信息增强的低秩稀疏矩阵分解(LSMF)论文推荐方法。首先,利用预训练的文档级表示学习和引文感知转换器SPECTER(Scientific Paper Embeddings using Citation-informed TransformERs)学习论文的表示,计算并构造文章之间的相似度矩阵,将相似度矩阵与引文矩阵相加得到一个混合信息矩阵;其次,通过矩阵乘法将内容相似信息与引用信息融入到论文-作者矩阵中;最后,利用LSMF模型分解论文-作者矩阵以得到推荐列表。在ACL文集网络(AAN)和DBLP数据集上的实验结果表明,所提方法取得了较好的推荐性能,且所提方法引入内容信息与引用信息的方式同样适用于其他矩阵分解模型。对于非负矩阵分解(NMF)、奇异值分解(SVD)、低秩稀疏矩阵补全(LSMC)和去分解(GoDec),利用混合信息后的模型比未利用混合信息的原模型在2个数据集上的前30个推荐结果的召回率(R@30)分别提升了18.72、7.43、11.53、14.62和20.58、2.11、7.91、5.01个百分点。展开更多
To study the problem of knowledge translation in fuzzy approximation spaces, the concept of rough communication of crisp set in fuzzy approximation spaces is proposed. In a rough communication of crisp set in fuzzy ap...To study the problem of knowledge translation in fuzzy approximation spaces, the concept of rough communication of crisp set in fuzzy approximation spaces is proposed. In a rough communication of crisp set in fuzzy approximation spaces, the problem of uncertainty exists, for each agent has a different language and cannot provide precise communication to each other. By means of some concepts, such as CF rough communication cut, which is a bridge between fuzzy concept and crisp concept, cut analysis of CF rough communication is made, and the relation theorem between CF rough communication and rough communication of crisp concept is obtained. Finally, in order to give an intuitive analysis of the relation between CF rough communication and rough communication of crisp concept, an example is given.展开更多
将传统的两端直流系统短路比延伸到多馈入直流系统,建立了多馈入短路比(multi-infeed short circuit ratio,MSCR)表达式。基于CIGRE模型建立了三馈入直流输电模型,分析了与MSCR相关的变量,并结合换相失败免疫因子得出结论:增大多馈入短...将传统的两端直流系统短路比延伸到多馈入直流系统,建立了多馈入短路比(multi-infeed short circuit ratio,MSCR)表达式。基于CIGRE模型建立了三馈入直流输电模型,分析了与MSCR相关的变量,并结合换相失败免疫因子得出结论:增大多馈入短路比能够降低发生当地换相失败的风险。同时从多馈入交互作用因子的角度分析了发生同时换相失败的规律:减小交互作用因子能够降低同时换相失败的风险。展开更多
文摘针对传统协同过滤(CF)存在的数据稀疏和冷启动的问题以及在矩阵分解方法生成结果矩阵的过程中由于各种变换产生误差的问题,提出一种混合信息增强的低秩稀疏矩阵分解(LSMF)论文推荐方法。首先,利用预训练的文档级表示学习和引文感知转换器SPECTER(Scientific Paper Embeddings using Citation-informed TransformERs)学习论文的表示,计算并构造文章之间的相似度矩阵,将相似度矩阵与引文矩阵相加得到一个混合信息矩阵;其次,通过矩阵乘法将内容相似信息与引用信息融入到论文-作者矩阵中;最后,利用LSMF模型分解论文-作者矩阵以得到推荐列表。在ACL文集网络(AAN)和DBLP数据集上的实验结果表明,所提方法取得了较好的推荐性能,且所提方法引入内容信息与引用信息的方式同样适用于其他矩阵分解模型。对于非负矩阵分解(NMF)、奇异值分解(SVD)、低秩稀疏矩阵补全(LSMC)和去分解(GoDec),利用混合信息后的模型比未利用混合信息的原模型在2个数据集上的前30个推荐结果的召回率(R@30)分别提升了18.72、7.43、11.53、14.62和20.58、2.11、7.91、5.01个百分点。
基金supported by the Natural Science Foundation of Shandong Province (Y2006A12)the Scientific Research Development Project of Shandong Provincial Education Department (J06P01)+2 种基金the Science and Technology Foundation of Universityof Jinan (XKY0808 XKY0703)the Doctoral Foundation of University of Jinan (B0633).
文摘To study the problem of knowledge translation in fuzzy approximation spaces, the concept of rough communication of crisp set in fuzzy approximation spaces is proposed. In a rough communication of crisp set in fuzzy approximation spaces, the problem of uncertainty exists, for each agent has a different language and cannot provide precise communication to each other. By means of some concepts, such as CF rough communication cut, which is a bridge between fuzzy concept and crisp concept, cut analysis of CF rough communication is made, and the relation theorem between CF rough communication and rough communication of crisp concept is obtained. Finally, in order to give an intuitive analysis of the relation between CF rough communication and rough communication of crisp concept, an example is given.
文摘将传统的两端直流系统短路比延伸到多馈入直流系统,建立了多馈入短路比(multi-infeed short circuit ratio,MSCR)表达式。基于CIGRE模型建立了三馈入直流输电模型,分析了与MSCR相关的变量,并结合换相失败免疫因子得出结论:增大多馈入短路比能够降低发生当地换相失败的风险。同时从多馈入交互作用因子的角度分析了发生同时换相失败的规律:减小交互作用因子能够降低同时换相失败的风险。
