街道环境是城市居民最常接触到的环境，街道品质优劣将对周边居民的情绪产生不同程度的影响。为了探究城市街道环境与情绪体验的关系，以广州市北京路历史文化街区为例，对街道情绪进行 预测。首先利用开放街道地图、百度街景和高德 POI，提取多种街道特征；接着使用腾讯云语义分析工具，获取微博位置签到数据的正面情绪概率，将其作为部分街段的情绪值；并通过相关性 分析，研究多种街道特征与情绪的关系；最后利用人工神经网络回归模型，训练和预测了研究范围内所有街段的情绪值。结果表明：建筑可见率与情绪值呈正相关，植被和汽车可见率与情绪值呈负相关；北京路文化核心区内的单位长度街道情绪预测值最高，随着街道与北京路中轴距离的 增加，情绪值呈下降趋势；距离中轴较远的西南部出现较多负面情绪，建议通过沿街建筑立面修缮、引入高端业态、优化人车流线规划等方式对街道进行品质提升。

As the most common environment for urban residents to contact, the quality of the street will bring different degrees of emotional experiences to the surrounding residents. Therefore, it’s essential to explore the relationship between the urban street environment and emotional experience. With a case study conducted in Beijing Road Historical and Cultural Blocks, Guangzhou, this paper aims to study the relationship between multiple street characters and sentiments and predict street sentiments. Multiple street characters were extracted through Open Street Map, Baidu Street View, and Gaode POI datasets. The Tencent Cloud Semantic Analysis Tool is used to obtain the positive sentiment probability of check-in data at each location, which was partially used as the sentiment value of streets. Correlation analysis was used to study the relationships between multiple street characters and sentiments. The artificial neural network regress model was trained to predict the sentiments in all streets in the study area. The results show that the visibility of buildings has a positive relationship with sentiment values, while the visibility of vegetation and cars has a negative relationship with sentiment values. Unit-length street sentiments are the highest in the core area of Beijing Road cultural blocks, which decline as the streets depart from the central axis. More negative sentiments appear in the southwest area, far from the central axis, where restoration and improvement are needed. It is recommended that the quality of the street be upgraded through the repair of building facades along the street, the introduction of high-end businesses, and the optimization of pedestrian and vehicular flow planning.

TU986