一种基于中文分词和数据聚合的餐饮行为特征挖掘方法

摘要

针持卡人使用银行卡进行日常交易时，通过商户类别码（MCC）可以判断交易商户是否为餐饮商户，但是无法进一步细分交易商户所属菜系。为了分析持卡人餐饮行为特征，提出了一种银联数据与外部数据相结合的数据聚合方法。选择百度地图餐饮数据作为外部数据，对不同菜系的餐馆名进行中文分词、清洗，形成区分不同菜系的关键词组。以银联大数据平台为基础，百度数据与银联数据通过MapReduce技术进行聚合。百度菜系的关键词组对银联餐饮商户进行菜系划分，带有菜系标签的银联餐饮商户数据与银联持卡人日常交易数据进行聚合，挖掘出持卡人餐饮行为特征。%When cardholder using bank card for daily transaction, the merchant can be judged whether it is a restaurant merchant by MCC, but the merchant cannot be divided into detailed cuisine. For analyzing cardholder restaurant behavior characteristics, this paper proposes a method of data aggregation for combining UnionPay data with external data. Baidu map restaurant data is chosen as external data, restaurant names of every cuisine are segmented and cleaned, and key words of every cuisine are built. Based on UnionPay big data platform, Baidu data and UnionPay data are aggregated by MapReduce. UnionPay restaurant merchants are divided into different cuisines by Baidu cuisine key words, UnionPay restaurant merchants data with cuisine labels is aggregated with UnionPay cardholder daily transaction data, cardholder restaurant behavior characteristics are mined.