基于大数据的社交网络数据挖掘

发布时间：2018-03-07 22:00

本文选题：大数据　切入点：社交网络　出处：《苏州大学》2015年硕士论文　论文类型：学位论文

【摘要】：万事万物都有联系,人们与社交网络相互联系,与信息,组织,所在地相互联系,只对个体进行研究所获得成果是有限的,所以要对整个系统进行研究,而这个系统就是一个社交网络。社交网络分析由图论,数学以及社会学中的社交网络理论发展而来,在最近的这10-15年中,网络分析还借助计算机科学,物理学,生物学和经济学等不同领域的发展。与常规的网络数据分析相比,基于社交网络的数据分析,分析对象非常明确,分析的形式多种多样,分析的方法和内容更多。若将常规的分析方法应用到社交网络数据分析中,我们会发现分析的结果达不到我们的预期效果。就当前的数据分析方法、数据模型和社交网络数据,本文从数据的获取、数据的清洗加工、数据模型应用并在应用过程中对模型进行优化,形成的数据分析结果如何为管理者进行决策进行了一系列的研究。本文先从斯坦福大学的Stanford Large Network Dataset Collection中下载Amazon networks数据,探讨了从抓取数据、清洗数据、预处理数据、探索数据、选择合适的算法、建立相关模型到最终分析评估的全过程。先用Python对预处理数据,导入数据库,并用Clementine 12.0分析其中客户购买产品后评价的有效性,并评估此评价是否会对其他客户购买该产品产生影响,采用数据挖掘技术可挖掘出数据背后的信息。数据挖掘中的聚类分析是将相似度较高的用户聚成一个类,然后再进行数据分析。本文对客户评价的次数,评价的时间,评价的频次,评价的得分进行聚类分析,最后对分析过程中运用的两种聚类方法取得的结果进行比较,构建客户分级体系。通过分析结果可对当前亚马逊的购买客户进行分级维护,对于最有价值客户需要给予重点关注,将社交网络大数据与微营销结合起来,提升亚马逊的销售业绩和影响力,从原先的单一购物商城模式,发展成为综合电商平台,并培育自己的客户群。
[Abstract]:Everything is connected, people are connected to social networks, they are connected to information, organizations, locations, and the results of individual research are limited, so we have to study the system as a whole. And this system is a social network. Social network analysis has evolved from graph theory, mathematics, and sociological social network theory, and in the last 10-15 years, network analysis has also relied on computer science, physics. Development in different fields, such as biology and economics. Compared with conventional network data analysis, social network-based data analysis has very clear objects and many forms of analysis. If we apply the conventional analysis method to the social network data analysis, we will find that the results of the analysis are not as good as we expected. For the current data analysis method, the data model and the social network data, In this paper, the data acquisition, data cleaning and processing, data model application and optimization of the model in the application process, The result of data analysis makes a series of researches on how to make decisions for managers. Firstly, this paper downloads Amazon networks data from Stanford Large Network Dataset Collection of Stanford University, discusses how to capture data, clean data, preprocess data and explore data. Select the appropriate algorithm, establish the relevant model to the whole process of final analysis and evaluation. First, use Python to pre-process the data, import the database, and use Clementine 12.0 to analyze the validity of the evaluation after the customer buys the product. And evaluate whether this evaluation will affect other customers to buy the product, the data mining technology can be used to mine the information behind the data. The clustering analysis in data mining is to cluster the users with high similarity into a class. Then the data are analyzed. The paper makes cluster analysis on the frequency, time, frequency and score of customer evaluation. Finally, it compares the results obtained by the two clustering methods used in the process of analysis. Build a customer rating system. Through the analysis of the results of the current Amazon customers can be graded maintenance, for the most valuable customers need to give priority attention, the social network big data and micro-marketing combined, To boost Amazon's sales and influence from the original single shopping mall model to an integrated ecommerce platform and to nurture its own customer base.
【学位授予单位】：苏州大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TP311.13

【相似文献】