| 摘要: |
| 公园是城市重要的公共空间，是居民
亲近自然、放松身心的主要场所。用户在访问公
园过程中，多种感知要素影响其体验和评价，通
过分析其影响机制和特征可以针对性地改善使
用体验，提高规划的人文关怀。本文以重庆36
个公园为例，通过网络爬虫技术采集海量用户点
评数据，使用百度深度学习模块进行文本情感分
析，共挖掘10个感知要素。通过分析要素感知频
率、感知倾向以及不同公园类型自身感知特征，
提取影响公园感知的6个关键因素，其中自然景
观、绿化覆盖、公园规模、声音气味4个要素可
以通过提高可视性等措施来增加积极感知，公
园服务设施、地形起伏2个要素是公园规划建设
中应重点改善的内容。本文希望通过使用感知
导向的主观评价，为人性化公园和宜居城市建
设提供指导。 |
| 关键词: 城市公园 使用感知 使用评价 文本情感分析 深度学习 |
| DOI：10.13791/j.cnki.hsfwest.20220420 |
| 分类号: |
| 基金项目:国家自然科学基金项目（51978091） |
|
| Research on User Perception Evaluation of Urban Parks Based on Text SentimentAnalysis: Taking 36 Parks In Chongqing as an Example |
|
YE Lin,JIANG Lun,HAN Guifeng
|
| Abstract: |
| With the rapid development of economy and urbanization, Chinese residents have
made significant improvements in the physical environment and living conditions, while their
spiritual needs still need more promotions. Urban parks play an active role in improving the
quality of human settlements, they are the main places for residents to get close to nature and
relax. In the process of visiting parks, various perception factors affect users’ experience and
their evaluation. By analyzing the influence mechanism and characteristics, planners can target
to improve users’ positive experience, and improve the refinement level of urban planning. In
recent years, research on users’ perception based on social media data has gradually emerged.
Using deep learning techniques to mine the perception information contained in comments,
which provides new ideas for a more comprehensive evaluation of users’ perceptions and park
operation.
In this paper, we select 36 parks in the core area of Chongqing city as a case study. All
of them have a high number of online comments. It covers seven types of parks, including
comprehensive park, ecological park, children’s park, sports park, memorial park, amusement
park and geopark. We collect 22,355 user comments through crawler technology, the perception
factors are divided into four major categories and ten minor categories. Baidu AI Cloud platform
is used to analyze the text sentiment, and SPSS is used to statistics the perception frequency
and perception tendency. The results show that park landscape and environmental factors can
be perceived easiest, while the infrastructure factors are usually ignored. Environmental factors
tend to give users the most positive experience, followed by landscape factors, while park size,
topography and facilities are easy to trigger negative perceptions. By analyzing the perceptual
characteristics of different park, it is found that there are differences in the degree of attention
caused by various factors. For example, in sports parks, people are more likely to perceive the
facilities, while in other types of parks, people pay less attention to them. In addition, users are
likely to have a positive perception in parks with beautiful scenery or comprehensive functions.
Moreover, we find that there are many factors affect the overall perception, among them,
those factors whose perception tendency changes rapidly with the increase of users’ attention
can greatly affect using experience, that is the key to this study. We analyze the correlationsbetween perception frequency and perception tendency. The results show that six factors significantly impact users’ perception, such as natural
landscape, green coverage, park size, sound and smell, service facilities and topographic relief. Based on the above findings, several suggestions
are proposed, such as improving the sophistication of park design, increasing green visibility on the main tour routes, enrich the variety of scented
flowers to prolong the viewing time. Design more parking facilities, rest facilities and catering facilities should also take into consideration, which
can reduce the negative perception impression.
The shortcomings of this study and future studies have also discussed. Firstly, in terms of methods, the assessment accuracy of users’ perception
are not perfect, there still need more work to improve the deep learning algorithm and enlarge the basic corpus, which can have substantial
improvements on the accuracy of results and the efficiency of analysis. Secondly, parks are diverse, their characteristics are influenced by various
factors, such as culture and topography. One suggestion can not fit all, it needs some adaptive changes from the actual situation. Thirdly, factors that
affect users’ perception are not limited to park characteristics, the users’ conditions, access methods, urban landscape and climatic should also taken
into consideration. Fourth, the assessing system can be divided into more categories. For example, Chongqing has a hot and rainy summer, cold and
wet winter, while the climate in spring and autumn is relatively pleasant. Therefore, the climate factor can be further subdivided into two features
(winter and summer, spring and autumn) in future studies, which can improve the research precision and the application of planning practice.
Overall, this study can enhance the measurements and evaluation on visitors’ perceptions of parks. In order to effectively improve users’
experience, satisfaction and happiness, it combines user experience with several specific practices, such as landscape construction, the maintenance
of infrastructure, and management. This study can also provide constructive guidance for humanized park and livable city construction. |
| Key words: Urban Park User Perception User Evaluation Text Sentiment Analysis Deep Learning |
