This research investigates the use of social network and semantic analysis methods on user-generated content from the TripAdvisor travel forum to forecast tourism demand. Analyzing over 2,660,000 posts from 147,000 users across seven major European cities over a decade, the study introduces a novel methodology for analyzing tourism-related big data. By integrating social network and semantic variables into Factor Augmented Autoregressive and Bridge models, the study found that these variables significantly improved forecasting performance compared to traditional univariate models and Google Trends data. Key factors such as forum language complexity and communication network centralization, particularly the presence of prominent contributors, were influential in forecasting international airport arrivals.