基于Spark的实时情境推荐系统关键技术研究

随着移动互联网的飞速发展，人们面临的信息过载的问题日益严重，大数据场景下对用户的推荐面临着巨大困难。为了解决推荐时效性、准确度、大数据量，提出了一种基于Spark的实时情境推荐算法。该算法在协同过滤的基础上融合了情境过滤，以Kafka作为实时消息收发器，以Spark Streaming来处理实时流数据，增强了算法的准确性和时效性。实验证明，该算法和传统协同过滤算法相比，准确率和时效性更高，且在大数据场景下更有优势。

The Research of Real-time Recommendation System on Spark

With therapid development of mobile Internet,the problem cased by information overload is becoming increasingly serious.The recommendation is facing enormous difficulties in the situation of big data.In order to solve the problems cased by big data,the real-time recommendation system is designed in this paper.The algorithm combines the context filtering based on collaborative filtering,using Kafka as a real-time message transceiver,using Spark Streaming to handle the real-time streaming data,which enhances the accuracy and effectiveness of the algorithm. Experiments show that this algorithm is more accurate and timeliness.The advantages is more obvious in the situation of big data.