This paper has finished a fine-grained sentiment analysis of micro-blog “Tree Hole” messages, and studied the distribution characteristics of emotional expression of the message makers in China's geospatial area, so as to provide support for the monitoring, early warning, intervention and rescue for patients with depression in the “Tree Hole” who have emotional deterioration and suicidal behavior. The Tree-Hole Intelligent Robot was used to capture the messages in the “Tree Hole”, and the standardized message data was obtained after data processed. Then an affective lexicon was set up together with the knowledge graph technology to count and analyze the emotional intensity scores of each province of China under the six emotions of joy, anger, sadness, disgust, fear and surprise, and the geospatial visualization results were presented. There are differences in the emotional expression of message makers in different provinces of China in “Tree Hole”. The emotional expression of the message makers in “Tree Hole” may be related to the economic condition, climatic condition, social background of the region, and may reflect the mental state of local netizens to a certain extent.