微型电脑应用

标签体系涉及不同领域、不同行业、不同平台，存在差异性，因此对其通用性、可扩展性要求较高。数据质量、数量、多样性等因素也导致用户族群画像构建具有较高难度。为进一步降低画像平均绝对误差，设计一种标签体系下的用户族群画像动态构建方法。整合用户族群内部资源，并以此建立客户标签体系，进一步采集小数据，按照相关性将主题下的单词实现动态排序，并确定标签维度，采用矩阵分解法构建用户族群画像。实验结果表明，研究方法平均绝对误差较小，用户族群画像动态构建效率较高，满足用户族群画像设计需求。

Since the label system involves the differences among different fields, different industries and different platforms, the requirements for its strong versatility and good scalability are high. Factors such as data quality, quantity and diversity also make it difficult to construct user group portrait. In order to further reduce the average absolute error of portrait, a dynamic construction method of user group portrait based on label system is designed. The method integrates the internal resources of the user group, and establishes the customer label system based on this, further collects small data, dynamically sorts the words under the theme according to relevance, determines the label dimension, and uses the matrix decomposition method to build the user group portrait. The experimental results show that the average absolute error of the research method is small, and the dynamic construction efficiency of user group portrait is high, which can meet the needs of user group portrait design.