【摘要】 如何使用海量社交媒体数据分析城市居民日常行为特征成为国内外学者广泛关注的问题。城市居民在不同时空间扮演的社会角色的不同将导致其出行行为特征随之变化，为此构建居民出行活动模式模型表征居民出行行为，引入自然语言处理领域内的标签狄利克雷分布模型Labeled-LDA完成群体分类，分析居民群体的出行行为特征，并讨论个体出行行为的不确定性。使用波士顿海量Twitter签到数据的实验表明，该方法能够有效区分典型的城市居民群体，并为居民在不同时空间表现出的不同出行行为特征提供概率解释。更多还原

【Abstract】 How to use massive social media data to analyze the daily behavior characteristics of urban residents has become a widespread concern of scholars at home and abroad. Urban residents’ travel behavior characteristics will change with their different social roles in different time and spaces. Therefore, human mobility model was constructed to represent the residents’ travel behavior, and Labeled-LDA in the field of natural language processing was introduced to analyze the travel behavior characteristics of urban residents in the group dimension, and discuss the uncertainty of individual travel behavior. Experiments in Boston with massive twitter check-in data show that this method can effectively distinguish the typical urban residents and provide a probability explanation for the different travel behavior characteristics of urban residents in different time spaces.更多还原