Hits:
Date of Publication:2018-01-01
Journal:计算机工程与设计
Affiliation of Author(s):经济管理学院
Volume:39
Issue:3
Page Number:752-757,791
ISSN No.:1000-7024
Abstract:To solve the problem of data sparsity and algorithm scalability in collaborative filtering algorithm in cloud environment,a collaborative filtering algorithm based on limited-memory BFGS optimization was proposed.The features of user and item were combined to reduce the impact of data sparsity.L-BFGS algorithm was designed to train factorization machine for rating prediction and recommendation.To evaluate the performance of the proposed algorithm,a set of experiments on different datasets was conducted in the Spark.The results show that the proposed algorithm outperforms other methods in precision,and has good scalability.The efficiency and computing resources can better adapt to real-time requirements of recommendation system in cloud environment.
Note:新增回溯数据