Tourist preferences are important for the high-quality planning and design of recreation spaces. The famous scenic locale of West Lake in Hangzhou, China, is used as an example in this study. Based on multi-source data (e.g., online comments, and tourist photographs), we used content analysis, kernel density estimation, and image semantic segmentation technology to determine the spatial distribution of tourists’ landscape preferences. We analyzed these spatial sight characteristics from the viewpoint, sight distance, and perspectives. The results show that tourists’ landscape preferences are mainly concentrated on landscape architecture for recreation. The viewpoints of these preferences are concentrated in the north-south embankment—far beyond that of the east-west embankment. The preferences also show a spatial sequence in terms of sight distance, and the best visual effect is the open platform at the north and south islands of Xiaoyingzhou. From the perspective of vision, the degree of spatial openness in the tourists’ landscape preferences is proportional to the distance of vision; the two factors have a convergent relationship. The discussion of the characteristics of tourists’ landscape preference and space sight creation in Xiaoyingzhou provides a reference for the quality improvement of island recreation space.