基于改进FP-growth的多品类打包推荐算法  

A multi-category bundling recommendation algorithm based on the improved FP-growth algorithm

作　　者：李雄清 李永 王骏飞 臧凌 刘德志 卞宇轩 柴阅林 李卓潇 刘云韬 Li Xiongqing;Li Yong;Wang Junfei;Zang Ling;Liu Dezhi;Bian Yuxuan;Chai Yuelin;Li Zhuoxiao;Liu Yuntao(Beijing Engineering Research Center of Civil Aviation Big Data,Beijing 101318,China;Travelsky Technology Limited,Beijing 101318,China;Beihang University,Beijing 100191,China)

机构地区：[1]北京市民航大数据工程技术研究中心,北京101318 [2]中国民航信息网络股份有限公司,北京101318 [3]北京航空航天大学,北京100191

出　　处：《网络安全与数据治理》2025年第3期47-53,共7页CYBER SECURITY AND DATA GOVERNANCE

摘　　要：多品类打包推荐是现代推荐系统中的重要任务,旨在通过组合不同类别的产品,向用户进行一站式推荐,以满足用户的多样化需求并提升用户体验。目前,该任务面临即时响应需求高、数据规模庞大、数据稀疏性高等挑战,现有打包算法难以应对。为解决上述问题,提出了一种基于改进FP-growth算法的多品类打包推荐算法,即在FP-growth算法的基础上,对产品属性间的关联关系进行挖掘,并根据属性关联规则匹配出符合度最高的打包产品,有效缓解了产品间的数据稀疏性问题。在基于航空旅游零售领域的数据集上,本方法相较于基准方法显著提高了打包质量和打包效率。Multi-category bundling recommendation is a critical task in modern recommender systems,which aims to combine products from various categories into a one-stop recommendation to meet users′diverse needs and enhance their experience.This task faces challenges such as high demand for real-time responses,large data scale and high data sparsity,which existing bundling algorithms struggle to address.This paper proposes a multi-category bundling recommendation algorithm based on an improved FP-growth algorithm,which mines associations between product attributes.The algorithm matches the most relevant bundling products based on attribute association rules,which effectively alleviates data sparsity under the product view.Experimental results on a dataset from the aviation travel retail sector show that the proposed method significantly improves both bundling quality and efficiency compared to baseline methods.

关 键 词：多品类打包 推荐系统 关联规则挖掘 FP-GROWTH算法 航空旅游零售 

分 类 号：TP311.1[自动化与计算机技术—计算机软件与理论] TP274[自动化与计算机技术—计算机科学与技术]