Abstract: Users engage in all sorts of activities in various different urban functional zones in different roles. How to infer user roles based on urban functional zones or infer frequent urban functional zones based on user roles is a challenge in the research area of human mobility. This paper builds a mutual inference model for user roles and urban functional zones through extracting context awareness factors from users’ micro-blog, e.g. user ID, check-ins and GPS-coordinates. The experimental results show that the model achieves good performance both on the aspects of user roles inference and urban functional zones inference.

Key words: user roles, functional zones, user mobility, context factors

摘要： 用户以不同的角色在城市中不同的地域间从事各种活动，如何通过用户活动的地域推断他们的角色，以及通过用户的角色推断他们有可能频繁活动的地域结构，是用户移动性领域研究的挑战性问题。通过从用户所发微博中提取必需的情境要素，如用户ID、check-ins和GPS坐标等，对用户的移动规律建立用户角色与城市地域结构的互推断模型。实验结果表明，该模型无论在利用用户角色推断城市地域结构，还是在利用城市地域结构推断用户角色方面，均取得了较好的性能。

关键词: 用户角色, 地域结构, 用户移动性, 情境要素