Platial Science: A multi-dimensional data signatures approach to modeling Place

历史上，地理信息科学（GIScience）侧重于量化实体之间地理空间关系的方法和技术，无论是地籍调查还是人类流动模式。随着大数据分析和环境智能的最新进展，传感器移动技术和社交媒体的蓬勃发展，GIScience已经从传统的空间数据分析扩展到整合非地理空间上下文信息。这些数据的纳入使地理信息科学从传统的基于空间的方法转向新兴的“基于地点的信息科学”学科。“******我的大部分工作的应用领域是识别人类活动的模式，例如，人们在何时何地吃饭、购物等。从这个角度来看，地理空间数据提供了关于一个人的位置的有价值的信息，但很少提供对个人正在参与的活动的洞察。这样，空间位置只是复杂的多维空间概念中的一个维度。例如，了解一个人正在访问的地方（例如，梅尔在蒙特利尔的餐馆）比个人的地理坐标更有价值。某人的地点位置提供了与地点类型、个人活动、人口统计等相关的更多细节。通过确定一个人的“平面”位置，而不仅仅是他们的空间位置，我们可以更好地了解他们的活动、移动趋势等。因此，地点计算模型在地理信息学和信息科学中越来越重要，因为它们在从地理信息检索和决策支持系统到交通建模和城市规划等主题中发挥着主导作用。******传统上，这个地方的概念是从高度定性的角度来看待的。在我的研究计划中，我建议通过利用用户贡献数据的异质性来对地点进行计算建模。这些数据表现为在线评论、社交媒体帖子和地理社交签到等。具体来说，我把这个地方的概念分开，从三个不同的维度来看待它。一种空间维度——关注地点的地理位置及其与其他地理特征的空间关系。时间维度——展示地点如何随时间变化（例如，白天的公园和晚上的公园是不同的）；主题维度——描述场所的类型和场所提供的活动。通过识别这些维度中的独特模式，我正在进行的工作表明，地方可以被量化，邀请对差异和相似性进行评估。在与学生的合作中，我计划继续这一研究思路，并打算证明这些新颖的基于地点的签名可以用来回答一系列基于信息学的问题。

Historically, Geographic Information Science (GIScience) has focused on methods and techniques for quantifying geospatial relationships between entities, be they cadastral surveys or human mobility patterns. As recent progress in big data analytics and ambient intelligence is met with sensor-enabled mobile technology and the social media boom, GIScience has expanded from traditional focus on spatial data analysis, to incorporating non-geospatial contextual information. The inclusion of these data shifts GIScience away from traditional Space-based approaches towards the burgeoning discipline of “Place-based Information Science.”******The application domain of much of my work is on identifying patterns in human activities, e.g., where and when people eat dinner, shop, etc. From this perspective, geospatial data provides valuable information about a person's location, but rarely offers insight into the activities in which the individual is participating. In this way, spatial location is only one dimension in the complex multi-dimensional concept of Place. For instance, knowledge of the place that an individual is visiting (e.g., Mel's diner in Montreal) is of greater value than just the geographic coordinates of the individual. The place location of someone provides considerably more detail related to type of place, activities of the individual, demographics, etc. By identifying a person's “platial” locations, rather than just their spatial locations, we gain a better understanding of their activities, mobility trends, etc. Consequently, computational models of place are increasingly important in both geoinformatics and information science as they play a dominant role in topics ranging from geoinformation retrieval and decision support systems to transport modeling and urban planning.******Traditionally, this concept of place has been approached from a highly qualitative perspective. In my research program I propose to model place computationally by exploiting the heterogeneity of user-contributed data. These data, manifest as online reviews, social media posts and geosocial check-ins, to name a few. Specifically, I pull apart this concept of place and approach it from three different dimensions, 1. a spatial dimension- concerned with the geographic location of places and their spatial relationship to other geographic features, 2. a temporal dimension- demonstrating how places change depending on time (e.g., A park in the day is a different than a park at night) and 3. a thematic dimension- that describes the type of place and activities that are afforded by the place. Through the identification of unique patterns within these dimensions, my ongoing work demonstrates that places can be quantified, inviting the assessment of differences and similarities. In collaboration with my students, I plan to continue on this research thread and intend to demonstrate that these novel place-based signatures can be used to answer a range of informatics-based questions.