论文部分内容阅读
【目的】为有效解决微博客户特性的表示问题,以更好地实施企业微博客户细分。【方法】借助微博平台上客户的个人和社会关系特性,利用客户及其好友的自定义标签表示客户的特性,采用基于非负矩阵分解的文本聚类方法,提出一种面向企业微博的客户细分框架。【结果】实验结果表明,基于非负矩阵分解的方法取得约86.130%的asw指标平均值,远远超出基于K-means和层次聚类的方法。【局限】只通过融合微博客户个人及其关注好友的标签表示微博客户特性的方法不能够全面刻画客户特征。【结论】能够为企业微博客户细分中的客户特性的表示、细分、评价及结果可视化等问题提供参考和借鉴。
【Objective】 In order to effectively solve the problem of expressing the characteristics of Weibo customers and to better implement the enterprise Weibo customer segmentation. 【Method】 Based on the personal and social relationship of customers on Weibo platform, this paper uses the custom labels of customers and their friends to represent the characteristics of customers. Based on the text clustering method based on non-negative matrix factorization, Customer segmentation framework. 【Result】 The experimental results show that the method based on non-negative matrix factorization achieves an average of about 86.130% asw indices, far exceeding the methods based on K-means and hierarchical clustering. [Limitations] The method of expressing the Weibo customer’s characteristics only by integrating the Weibo customer’s individual and their followers’ tags does not fully characterize the customer. 【Conclusion】 It can provide reference and reference for the representation, segmentation, evaluation and visualization of customer characteristics in the customer segmentation of enterprise Weibo.