加速交替最小二乘法推荐系统优化设计

推荐系统帮助用户在海量数据中更便捷地找到他们最感兴趣的内容。但推荐系统存在可信度低、推荐结果的可解释性不足、可扩展性不好、随着用户数量的增大,计算时间增长且精度较低、数据稀疏性和冷启动等问题。为此提出基于交替最小二乘法(alternating least squares,ALS)的推荐系统优化算法,在ALS基础上对两个部分进一步优化:第一部分采用LBFGS (limited-memory broyden-fletcher-goldfarb-shanno)算法使搜索方向快速计算出来;第二部分采用阻尼牛顿法求解步长因子。在Spark平台上加以验证,取得较好效果。

The Optimal Design of an Accelerated Alternating Least Square Recommendation System

Recommendation system came into being in the era of big data, core meaning is to help users more convenient to find them in huge amounts of data are most interested in. Although recommendation systems application scenario is widespread, but there is insufficient credibility is low, and recommend the results can be interpreted, poor scalability, and the increase of the number of users, the computing time and low accuracy, data sparsity and cold start and other issues. This paper proposes a recommendation system optimization algorithm based on alternating least squares ALS, which further optimizes the two parts based on ALS: the first using L-BFGS(Limited-memory Broyden-Fletcher-Goldfarb-Shanno) algorithm allows the search direction quickly calculated; Secondly, the damped Newton method is used to solve the step factor.