云环境下针对企业营销的个性化智能推荐研究

发布时间：2018-04-04 16:03

本文选题：数据挖掘　切入点：客户模型　出处：《浙江理工大学》2017年硕士论文

【摘要】：网络营销模式是企业了解顾客的需求后作出企业利润最大化的销售策略。随着数据量的不断增加,个性化的销售模式也变得日趋重要,企业应对市场、用户需求的不断变化需做出更加敏捷的反映,从而增加企业与顾客直接交易的机会。个性化的推荐促使企业的营销模式更加成熟,从而快速提高企业的销售量和销售金额。这种推荐模式主要采用协同过滤技术、基于内容的推荐技术。协同过滤技术最简单的思路就是搜索与目标对象相似兴趣的邻居用户,并将相似用户的偏好列表推荐给目标用户,基于内容的推荐算法是根据用户之前喜欢的商品推荐相似产品。本文针对协同过滤技术、基于内容的推荐算法两种推荐技术存在的一些缺陷,考虑情景状态下消费者的兴趣指标和兴趣偏好,从大量数据中获取消费者兴趣特征值,利用获取的特征值来创建兴趣模型,实现基于消费者模型的个性化智能推荐。主要工作如下:(1)针对商品的特点,基于全国零售户销售数据,利用数据挖掘技术获取用户之间的关联关系,构建零售业的客户价值指标。聚类算法将获取的样本文件进行分类,从而提取的数据可作为企业中的个性化销售的基础数据。(2)基于情景的用户偏好的分析以及商品属性的分类。本文采用基于情境下的建模推荐模式,从商品的自有属性和用户的兴趣偏好的角度出发,建立用户兴趣的权重值。根据权重值大小进一步分析用户偏好的侧重方向。文中情景因素主要含有位置、时间段、季节、伙伴等特定情境,用来预测用户对商品资源的选择行为。(3)个性化推荐算法中冷启动问题和数据稀疏问题的研究。本文针对协同过滤算法中冷启动的问题提出了基于用户、商品属性、浏览时间的个性化推荐技术,解决个性化推荐中存在的缺陷。基于新用户好友关系解决数据稀疏问题。文中通过实验数据,对于企业中个性化销售中冷启动问题以及数据稀疏性进行了细致研究,从而有效的提高了企业中的销售数量和销售额度。
[Abstract]:The network marketing mode is the sales strategy that the enterprise makes the profit maximization after knowing the customer's demand.With the increasing amount of data, the individualized sales model is becoming more and more important. Enterprises should respond to the market and the changing needs of customers more quickly, thus increasing the opportunity of direct transaction between enterprises and customers.Individualized recommendation makes the marketing model more mature, thus increasing the sales volume and sales amount quickly.This recommendation mode mainly adopts collaborative filtering technology and content-based recommendation technology.The simplest idea of collaborative filtering is to search for neighbor users with similar interests to the target object, and recommend the preference list of similar users to the target users.Content-based recommendation algorithms recommend similar products according to the products users like before.In this paper, aiming at some defects of collaborative filtering technology and content-based recommendation algorithm, considering the interest index and interest preference of consumers in the situation, we obtain the characteristic value of consumer interest from a large amount of data.The interest model is created by using the obtained eigenvalues, and the personalized intelligent recommendation based on the consumer model is realized.The main work is as follows: (1) according to the characteristics of commodities, based on the national retail sales data, using data mining technology to obtain the relationship between users, build the retail customer value index.The clustering algorithm classifies the sample files so that the extracted data can be used as the basic data of personalized sales in enterprises. 2) the analysis of user preferences based on scenarios and the classification of commodity attributes.In this paper, based on the context-based modeling recommendation model, the weight of user interest is established from the point of view of the commodity's own attributes and the user's interest preference.The emphasis direction of user preference is further analyzed according to the weight value.The situational factors include location, time, season, partner and so on, which are used to predict the user's choice behavior of commodity resources. 3) the cold start problem and data sparsity problem in personalized recommendation algorithm.In order to solve the problem of cold start of collaborative filtering algorithm, this paper proposes a personalized recommendation technology based on user, commodity attributes and browsing time to solve the shortcomings of personalized recommendation.Solve the data sparse problem based on the new user's friend relationship.Based on the experimental data, this paper makes a detailed study on the cold start problem and data sparsity of individualized sales in the enterprise, which effectively improves the sales quantity and sales quota in the enterprise.
【学位授予单位】：浙江理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.3

【参考文献】