| 摘 要: 为解决人们“每天不知道吃什么”的烦恼,采用Spark分布式处理框架,结合分布式存储数据库(MongoDB)、日志收集系统(Flume)、分布式系统文件(HDFS)等工具,实现对菜品实时评分及特色化推荐。系统包含菜品评分及储存模块、评分数据处理模块、菜品推荐模块、推荐结果展示模块等,其中推荐模块采用协同过滤推荐算法。使用测评方法和指标验证3种推荐模型的有效性,根据测评结果设计并实现以基于物品的推荐模型为主、以基于Spark ALS的推荐模型为辅的智能菜品推荐系统。该系统能够快速准确地推荐顾客喜欢的菜品,提高了商家的服务效率和顾客的满意度,可用性较高。 |
| 关键词: 菜品推荐;Spark;协同过滤算法;HDFS |
|
中图分类号: TP311
文献标识码: A
|
| 基金项目: 河南省科技攻关项目“MOOC的大数据学习行为与成效分析技术研究”(222102320218);黄河交通学院2023年度校级教育教学改革研究项目(HHJTXY-2023jgyb20);黄河交通学院校级课程教学资源库建设项目“数据清洗与融合”(HHJTXY-2022kczyk101) |
|
| Design and Implementation of Intelligent Dish Recommendation System Based on Spark |
|
ZHOU Yangyue, LI Shifeng, LI Lin
|
(Huanghe Jiaotong University, Jiaozuo 454000, China)
yangyzhouzmxxn@163.com; 1006061024@qq.com; 18839508953@163.com
|
| Abstract: To solve the problem of people not knowing what to eat every day, this paper proposes an intelligent dish recommendation system by using the Spark distributed processing framework and tools such as MongoDB (a distributed storage database), Flume (a log collection system), HDFS (Hadoop Distributed File System), etc., to achieve real-time rating and personalized recommendation of dishes. The system includes a dish rating and storage module, rating data processing module, dish recommendation module, and recommendation result display module, etc. The recommendation module adopts a collaborative filtering recommendation algorithm. Evaluation methods and indicators are used to validate the effectiveness of three recommendation models. Based on the evaluation results, the intelligent dish recommendation system is designed and implemented, which is primarily based on item-based recommendation models and supplemented by Spark ALS-based recommendation models. The system can quickly and accurately recommend dishes that users like so as to improve the business service efficiency and customer satisfaction, with high availability. |
| Keywords: dish recommendation; Spark; collaborative filtering algorithm; HDFS |
