大数据平台下的推荐算法研究

Research on Recommendation Algorithm based on Big Data Platform

随着Web2.0的到来,数据呈现爆炸式增长,数据过载问题引起越来越多的关注,推荐系统应运而生。通过对用户的历史交互信息进行分析抽取,得到用户偏好,最终将用户最感兴趣的内容推荐给用户。用户和标的物的交互信息较少,产生了数据稀疏的问题,当有新的用户或者新的标的物进入系统时,系统无法对其进行精确的推荐。近年来,深度学习的浪潮正传播到各个研究领域中,在过去的几年里,深度学习逐渐被应用到各个领域,推荐系统从单一的传统推荐算法开始过渡到基于深度学习的推荐算法,对于解决数据稀疏性和冷启动问题颇有成效。本文对近几年来的推荐系统算法进行了研究,分析了各个类别的推荐系统的特点,最后,分析了推荐系统未来的研究方向。

With the advent of Web 2.0,the problem of data explosion and data overload attracts more and more attention.As a result,recommendation system(RS)emerges as the times require.In RS,users'historical interaction information is analyzed and extracted to obtain users'preferences,and the content that users are most interested in is finally recommended to users.The shortage of interaction information between users and items results in the problem of data sparseness.When a new user or a new item enters the system,the system cannot accurately recommend it.In recent years,the wave of deep learning is spreading to various research fields.In the past few years,deep learning has been gradually applied to various fields,and the RS has been transformed from a single traditional recommendation algorithm to recommendation algorithms based on deep learning,which is quite effective in solving the problems of data sparsity and cold start.This paper studies the algorithms of RS in recent years,analyzes the characteristics of each category of RS,and finally analyzes the future research direction of RS.