Research on street space quality on the basis of the semantics of people—taking two Suzhou streets as examples

As cities enter the stage of stock planning, the quality of street space has attracted increasing attention. The development of big data-related technologies provides opportunities for large-scale measurement and urban space analysis, but existing research on urban spatial quality focuses on environmental indicators and large-scale urban problems, with less quality exploration for street-scale users. This paper takes two typical streets in the new and old cities of Suzhou city as the research objects, constructs a street space perception index system based on the perspective of users, uses a deep learning model to collect and analyse the comment text, and analyses the correlations between different perception elements and street quality. The results of the semantic access perception quality data show that the two streets differ in multiple items of security, comfort, convenience and recognition. The new district street has good perception in walking space, the degree of greenness and road traffic, while the old city street shows strength in terms of public facilities and cultural display. The results of the street quality perception correlation show that Shiquan Street has a strong correlation between the overall space quality and multiple perception indicators, while Ligong Dike has correlation only between functional sub-items and some perception indicators. The overall classification results and the sub-item results of the two streets vary, and the sub-item results can better reflect the difference of existing conditions. This paper provides a reference for refined data analysis at the street scale and targeted optimization strategies for street space quality.