基于现实挖掘的人群多尺度时空特征及复杂活动模式研究

By observation and analysis of characteristic and pattern of human behaviors, researchers exploit scientific understanding of society and human behavior, and improve the management of society. As a result of the recent explosion of sensor-equipped mobile phone market, the "digital footprints" left by people while interacting with cyber-physical spaces are accumulating with an unprecedented breadth, depth and scale.Leveraging the capacity to collect, mine and analyze the massive data of digital footprints of individual, reality mining has had a transformative in?uence on revealing the patterns of individual, group and societal behaviors, which are difficult to be identified by traditional methods..Based on the modeling and nonlinear dimensionality reduction of massive data of digital footprints, this project aims to advance scientific understanding, modelling and predicting the multi-scale characteristic of individual and group behaviors. The project concentrates on: (1)Quantitative analysis on large-scale trend and small-scale temporospatial variance of human behavior; (2) Revealing implicit patterns of human behavior by trajectory clustering; (3)Mining, quantitative analysis and prediction of frequent patterns of individual, group and societal behaviors at different time scale..The research of multi-scale temporospatial characteristic of human behaviors opens the way to the scientific understanding, analysis and prediction of patterns and characteristic of urban behaviors. It has also practical applications for urban planning, traffic planning, better using public resources, and better public emergency management.

通过观测、分析人群日常活动所蕴含的特征和规律，为更好地理解和管理人类社会提供科学依据，是当前人类活动研究领域的重大研究课题。而利用挖掘手段对海量现实活动数据进行挖掘和分析，揭示用传统方法难以揭示的人群活动特征、模式和规律是研究这一问题的有效途径。本项目以海量的移动现实数据为基础，以人群活动的多尺度时空特征为研究对象，在对海量现实活动数据进行多维建模、非线性降维和求解的基础上，开展：(1)活动的大尺度时空趋势与小尺度时空变异的定量分析；(2)人群活动的多尺度时空聚类特征及其演化规律；(3)不同时空尺度上的活动频繁模式挖掘以及分类活动模式的定量分析与预测等科学问题的研究。本项目的理论和工程意义在于探索现实挖掘在人群活动的研究和应用模式，为更好地理解、测度和分析人群活动的多尺度时空特征和活动模式提供理论和方法指导，为城市空间优化、交通规划和突发公共事件处理等实际应用的管理和决策提供科学判据。

本项目以海量的移动现实数据为基础，以人群活动的时空特征为研究对象，开展对基于道路匹配的轨迹高效压缩，基于网格相对密度的热点路径探测，在移动轨迹时间维度、空间维度、时空组合维度下的移动模式挖掘问题，以及基于交通轨迹的城市环境下群体移动行为动力学与移动出行模式挖掘进行了研究。本项目的理论和工程意义在于探索现实挖掘在人群活动的研究和应用模式，为更好地理解、测度和分析人群活动的多尺度时空特征和活动模式提供理论和方法指导，为城市空间优化、交通规划和突发公共事件处理等实际应用的管理和决策提供科学判据。