Abstract: Traditional park evaluations normally focus on individual parks with limited samples and survey time. The rapid development of big data analysis provides a new opportunity for park system evaluation. Based on the text analysis of social media data, this paper compares the evaluations of 50 parks in Guangzhou, and explores the relationship between park evaluation and demographic characteristics, geographic location, evaluation time, and park attributes. The study finds that: 1) The positive evaluations of the parks mainly focus on free facilities, environment and features, while the negative evaluations are mostly concerned with the management services such as sanitary conditions of parks, transportation costs, and parking. The management service of Guangzhou Park must be improved. 2) Women are more concerned than men in whether the parks can meet their social contact needs. Local tourists pay more attention to the service quality of the parks while tourists from other places pay more attention to the external attributes. 3) General parks should focus more on their accessibility and basic service functions. Characteristic parks should pay more attention to meet the needs and characteristics of specific groups of people. Natural parks should pay more attention to the ecological quality and undertake activities that require a larger space. 4) From 2010 to 2018, people paid more attention to the creation of park characteristics. The types of activities in the parks are more diverse. Modern park construction should focus on the improvement of cultural service quality. This research is significant as it has broken through the limitations of traditional park research methods by incorporating social media data in the evaluation of parks to establish a more universal and strategic park evaluation index system.