基于B2C电子商务数据仓库的研究与设计

发布时间：2018-11-03 13:33

【摘要】：B2C电子商务网站系统每天都会生成大量的产品交易数据和访问日志数据,这些数据中蕴含着大量有价值的信息,例如,订单的来源、客户的行为、访问者的兴趣等。分析这些数据,不仅能帮助企业决策层指导B2C电子商务网站的运营,吸引更多用户,提高销售利润还能够反映出企业在市场营销、销售推广、售后服务和财务管理等各个方面的状况。总之,对这些数据进行深入有效的分析能够帮助管理者改善客户关系、提高企业各个方面的竞争力。本文在描述了数据仓库的相关理论,包括数据仓库的概念、基本特征、系统结构,B2C电子商务的概念、OLAP多维数据分析基础上,提出了一个较完善B2C电子商务数据仓库模型,论文主要工作包含以下内容： 1.分析B2C电子商务数据仓库的用户需求,提出B2C电子商务数据仓库多层次概念模型,设计了相关的维度模型和事实集,基于该模型完成了部分维度表和事实表的物理设计。 2.分析B2C电子商务数据仓库数据源,讨论半结构化的数据源处理,针对Web访问日志,结合半结构化数据的预处理方法,提出了改进的页面媒体类型时间阈值会话识别算法。通过对不同的URL页面类型,采用不同的页面时间阈值计算方法。相对于已有的对所有用户访问页面使用单一的先验阈值和现有动态阈值计算,该方法能够更真实的反映用户会话的情况,且识别的准确率有了较大提高,为后续的分析提供高效准确的数据。 3.基于本文提出的B2C电子商务数据仓库模型,构建一个实验性B2C电子商务数据仓库项目。以Zen Cart网站系统为例,确定分析主题并基于多维建模思想建立各种不同粒度的维度、数据集市,并设计一个ETL体系结构,包括ETL调度方案,数据预处理方法等。最后对订单数据进行联机分析,展示B2C电子商务数据仓库的价值。本文提出的B2C电子商务数据仓库模型有以下特点： 1.该模型有针对性、实用性较强的特色,它涉及对内、外电子商务贸易活动中企业的各个主要方面,包括页面点击、产品销售、订单、用户对产品的评论、销售利润、仓库、订单产品、物流配送等。 2.该模型采用多层次维度设计,通过合理有效的概念分层,为企业决策层提供更好的分析视角。最后,本文通过实验验证了模型的有效性。
[Abstract]:B2C e-commerce website system generates a large amount of product transaction data and access log data every day, which contains a lot of valuable information, such as the source of orders, the behavior of customers, the interest of visitors and so on. The analysis of these data can not only help the decision makers to guide the operation of B2C e-commerce website, attract more users, but also can reflect the marketing and sales promotion of enterprises. After-sales service and financial management and other aspects of the situation. In a word, in-depth and effective analysis of these data can help managers to improve customer relations and enhance the competitiveness of all aspects of the enterprise. On the basis of describing the relevant theories of data warehouse, including the concept, basic characteristics, system structure, concept of B2C electronic commerce and OLAP multidimensional data analysis, this paper puts forward a perfect data warehouse model of B2C electronic commerce. The main work of this paper is as follows: 1. Based on the analysis of user requirements of B2C e-commerce data warehouse, a multi-level conceptual model of B2C e-commerce data warehouse is proposed, and the related dimension model and fact set are designed. Based on the model, the physical design of some dimension tables and fact tables is completed. 2. The data source of B2C e-commerce data warehouse is analyzed and the semi-structured data source processing is discussed. An improved session recognition algorithm of page media type time threshold is proposed for Web access log combined with the pre-processing method of semi-structured data. Through different URL page types, different page time threshold calculation method is adopted. Compared with the existing user access pages using a single prior threshold and the existing dynamic threshold calculation, this method can more truly reflect the user session, and the recognition accuracy has been greatly improved. Provide efficient and accurate data for subsequent analysis. 3. Based on the B2C e-commerce data warehouse model proposed in this paper, an experimental B2C e-commerce data warehouse project is constructed. Taking the Zen Cart website system as an example, the analysis topic is determined and based on the idea of multidimensional modeling, different grained dimensions, data marts are established, and a ETL architecture is designed, including ETL scheduling scheme, data preprocessing method and so on. Finally, the online analysis of order data is carried out to show the value of B 2 C e-commerce data warehouse. The B2C electronic commerce data warehouse model proposed in this paper has the following characteristics: 1. The model has the characteristics of pertinence and practicability. It involves all the main aspects of the enterprise in both internal and external e-commerce trade activities, including page clicks, product sales, orders, users' comments on products, sales profits, warehouses, etc. Order products, logistics distribution, etc. 2. The model adopts multi-level dimension design and provides a better perspective for enterprise decision making through rational and effective conceptual stratification. Finally, the validity of the model is verified by experiments.
【学位授予单位】：广东工业大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TP311.13

【参考文献】