融合Fisher线性判别分析的多维特征融合情景感知推荐方法

发布时间：2018-11-18 09:12

【摘要】：针对采用单维特征建立用户的偏好模型所导致的推荐结果无法有效覆盖用户潜在偏好特征而影响推荐质量的问题,提出了一种基于Fisher线性判别分析的多维特征融合情景感知推荐方法。该方法建立了包含时间衰减度、属性偏好、偏好可影响程度等多维特征的偏好样本空间;采用特征融合、投影变换等方法,在最佳鉴别矢量空间基于Fisher判别准则融合用户的多维特征;采用拉格朗日乘子法求解最优投影方向,建立起多维特征优化的偏好获取模型。在BookCrossing与Netfilix数据集上的实验结果表明:与现有方法相比,所提方法的推荐准确率平均提高了16.61%,多样性平均提高了约38.01%,能够有效地覆盖用户的潜在偏好特征,并取得更好的推荐质量。
[Abstract]:To solve the problem that the user's preference model based on one-dimensional feature can not effectively cover the user's potential preference characteristics, the recommendation quality is affected. A multi-dimensional feature fusion scenario perception recommendation method based on Fisher linear discriminant analysis is proposed. In this method, preference sample space is established, which includes multi-dimensional features, such as time attenuation, attribute preference, preference influence degree and so on. Using the methods of feature fusion and projection transformation, the multi-dimensional features of users are fused in the optimal discriminant vector space based on Fisher discriminant criterion, and the Lagrange multiplier method is used to solve the optimal projection direction, and a preference acquisition model for multi-dimensional feature optimization is established. The experimental results on BookCrossing and Netfilix datasets show that compared with the existing methods, the recommended accuracy of the proposed method is improved by 16.61 on average, and the diversity is increased by about 38.01, which can effectively cover the potential preference features of the users. And to achieve better quality recommendations.
【作者单位】：西北大学信息科学与技术学院;山西水利职业技术学院信息工程系;
【基金】：国家自然科学基金资助项目(61373176)
【分类号】：TP391.3

【参考文献】