Search

• Geotagged photographs are used to investigate cultural ecosystem services (CESs) to the coastal areas of Hong Kong. • Visiting sentiments are accessed through user-generated textual content as a proxy in the analysis of the geotagged photographs. • Both cultural attractions and protected areas are typical hotspots associated with CESs. • The relationships between CESs-related sentiments and critical sociodemographic and temporal factors are discussed. Ecosystem service assessments have been conducted for coastal management with a primary focus on tangible natural capital. Thus far, the explicit variation of cultural ecosystem services (CESs) and particularly the sentiments of CES-related ecotourism are not well understood. This paper takes advantage of big social media data to unravel the patterns of CESs and visiting sentiments in coastal areas of Hong Kong. Through machine-learned keyword labels, we employed content analysis to derive visual information for geotagged photographs. Applying natural language processing to apps with machine learning in cloud computing, we derived the sentiments based on user-generated textual content associated with coastal ecotourism. Based on regression analysis and multiple comparisons analysis, we identify the association between critical demographic and temporal factors with CES and related visiting sentiments. Referring to previous studies, we identified the main intangible benefits into six basic divisions based on 424 keyword labels for coastal areas. Our results show that hotspots of CESs are spatially concentrated in both cultural attractions and protected areas, which are critical for coastal ecosystem management and protection. More specifically, these areas of high CES value have good spatiotemporal accessibility, high infrastructure coverage, and spatially explicit population and economic growth. Furthermore, we discover that sentiments related to coastal CESs vary based on social media characteristics. Our study renews the indicators of quantitative CES evaluation based on crowdsourcing geospatial data. [ABSTRACT FROM AUTHOR]