This paper utilizes web crawler in network text mining technology to obtain the Jiangsu garden review data on several OTAs and tourism UGC platforms, such as Baidu Tourism, Ctrip.com, and Ma’s Nest. The data preprocessing work such as text de-weighting, word frequency classification, and parameter solving is carried out on the document dataset by using the lexical algorithm, and the preprocessed text data is subjected to TF-IDF keyword extraction and LDA topic extraction. Starting from the characteristics of the cultural lineage, the intrinsic attraction elements affecting the commercial transformation of tourist attractions are explored, and taking the planning of the industrial, cultural tourism town in Ezhou as an example, the process of planning and designing from the establishment of tourism planning concepts, and thus the completion of the visual IP image, is elaborated. The tourism image perception of the four major gardens in Jiangsu is examined from both cognitive and emotional perspectives using the results of TF-IDF and LDA extraction. The results show that the trend of the frequency of each fraction of the emotion of the Humble Administrator’s Garden, Liouyuan Garden, Lion Grove, and Canglang Pavilion scenic spots is roughly the same, and the mean value of emotion is 0.758, 0.851, 0.828, and 0.845, respectively, reflecting the tourists’ average emotional perception of the individual scenic spots. This study provides new methods and ideas for promoting the integrated development of culture, commerce and tourism in small and medium-sized cities so that the culture and tourism industry can be developed sustainably.