A Deep Transfer Learning Toponym Extraction and Geospatial Clustering Framework for Investigating Scenic Spots as Cognitive Regions

In recent years, the Chinese tourism industry has developed rapidly, leading to significant changes in relationship between people and space patterns scenic regions. To attract more tourists, surrounding environment of a region is usually well developed, attracting large number human activities, which creates cognitive range for region. From perspective tourism, tourists’ perceptions tourist attractions are located city differ from objective spots. Among them, social media serves as an important medium tourists share information about spots potential learn spot information, it interacts influence people’s destination image. Extracting names data exploring their spatial distribution basis research on attractions. This study takes Hangzhou, well-known China, case explore its popular First, we propose attraction name extraction model based RoBERTa-BiLSTM-CRF extract data. Then, use multi-distance clustering method called Ripley’s K filter extracted names. Finally, combine road network polygons generated using chi-shape algorithm construct vague regions each spot. The results show that classification indicators our proposed significantly better than those previous toponym models algorithms (precision = 0.7371, recall 0.6926, F1 0.7141), also generally conform habitual cognition.