基于社交媒体数据的城市公园景感评价

城市公园是城市居民与自然生态系统互动的主要场所,为居民提供重要的生态系统服务。从景感生态角度和方法对城市公园文化服务进行评价,需要掌握大量居民个体对城市公园的感知、感受信息。随着现代信息技术和人们生活方式的发展,社交媒体成为大众获取信息、发布观点的主要方式之一。社交媒体数据具有数据易于获取、时间跨度长、数据量大等优势,同时,通过对社交媒体信息进一步进行文本提取与分析,还可以获取用户对特定活动场所更为具体的感知和感受,因此,社交媒体数据为城市公园景感评价提供了有价值的新的数据来源和手段。构建了基于社交媒体数据的城市公园景感评价框架和流程、评价指标以及信息提取方法,包括对获取数据进行预处理;构建感官感知及公园要素词库;通过情感分类,形成积极和消极情感文本,表征用户对文本关联景感要素的满意度;将各类情感文本分别进行要素及其对应感知感官的统计,以频率衡量景感要素对用户感知的重要性;在此基础上,基于重要性-绩效分析(IPA)方法构建了感知评价模型。最后,以北京玉渊潭公园为例,进行了实证研究。结果表明:视觉在居民对城市公园的感知体验中占据主导地位,触觉感知次之,听觉感知普遍不被重视,而嗅觉感知评价...

Landsense assessment on urban parks using social media data

The urban park is the place where dwellers interact with nature, contributing to urban sustainability by providing cultural ecosystem services. To evaluate the cultural ecosystem services of an urban park from the perspective of landsenses ecology, numerous information on visitors'' perceptions and experiences is required. With the development of modern information technology as well as dwellers'' lifestyles, the social media platform becomes one of the main places for the public to gain information and show opinions. Social media data is characterized by easy to obtain, long span of time, and a large quantity for landsense assessment of an urban park. By extracting and analyzing it, users'' specific perceptions and experiences of the given place can be obtained. This study established a framework of landsense assessment of an urban park using social media data and proposing a method for information extraction and indicator design. First, we preprocessed the data, and developed the corpus of senses and park elements, respectively. Then, the texts of positive and negative were divided by sentiment classification to measure users'' satisfaction on the relative landsense elements. Based on the above, the perception assessment model was established using the Importance-Performance Analysis (IPA) method. Finally, we conducted a landsense assessment of Yuyuantan Park in Beijing and suggested ways to improve visitors'' perception. The conclusions are as follows: vision is a strongly dominant component in people''s perception, following touch demission, while perceptions on the hearing dimension are relatively unimportant, and the indicator on smell dimension is still limited. ''Vision of plants'', ''vision of humans'', ''visual identification'', ''vision of water'' are the most important landsense elements. Through the perception assessment model, we did not identify the indicator that should be improved urgently, and the indicators that need to be improved secondly include ''vision of humans'', ''touch of roads'', ''food available'', ''feel of wind'', ''vision of animals'', ''sound of voice'', ''sound of plants'', and ''vision of roads''.