大众点评数据下的城市场所范围感知方法

“场所（place）”是具有特定语义和人文体验的地理位置，是虚拟地理环境中地理知识表达的核心要素，能够为深入理解与表达人们的地理空间认知，以及基于虚拟地理环境的分析模拟提供支持，因而对城市内部现有场所的识别与模糊建模是城市空间结构研究的基础之一。众源数据为提取和表达模糊场所提供了新的途径，然而目前研究多针对单个或少数几个场所进行建模。本研究针对城市多场所空间范围的多尺度建模，提出基于自适应核密度估计的模糊集方法，为进一步理解城市场所的模糊认知范围提供了可视化解决方案。并以北京市五环内场所为实例，采用大众点评网中商户自行上传的兴趣点（POI）数据，对其进行提取与表达。通过对比百度地图所展示的场所相应范围，发现该方法在大众点评数据集下的应用能更好地揭示商圈语义下的场所认知。

Representing Multiple Urban Places' Footprints from Dianping.com Data

A place is a geographical area with a particular semantic and humanistic experience.In virtual geographic environment studies,places play an important role in spatial knowledge representation,by providing the support for human-centroid understanding of geographic environment and VGE-based analysis and simulation.Identifying and modeling the existing vague places in cities are therefore fundamental to VGE studies.Crowd-sourced geo-data provide a new approach to extracting and representing vague places.The most of existing research,however,focused on modeling single or a few places but not considering the influence of threshold selection among multiple point sets of places in diverse scales.In order to model multiple places of the city at different spatial scales,we proposed the fuzzy set method based on adaptive kernel density estimation for generating the spatial footprints of places,which can provide a reasonable and efficient way to model multiple vague places.In the case study,POIs inside the 5^th Ring Road of Beijing,collected from Dianping.com,are used to visually represent footprints of places in the way of fuzzy sets and α-cuts,with the former one avoiding the over simplified representation of continuous surface while the latter focusing more on the crisp boundaries.By comparing the results and the corresponding places' scales showed by base maps,we found that the use of the dataset harvested from dianping.com could provide a better understanding of the places' footprints of commercial context.