<p>The growth of location-based social networks and user-generated content has increased the demand for more effective tourism recommender systems. Many existing methods rely mainly on textual reviews or ratings, which limits their ability to capture user preferences. To overcome this limitation, this study proposes a hybrid location-based recommender system that improves the accuracy, personalization, and diversity of travel recommendations. The system leverages geo-tagged photos and their metadata along with textual reviews and demographic information. These data sources are combined using content-based, collaborative, and demographic filtering. The Bat Algorithm is further applied to optimize feature weighting and cluster users and locations, enhancing recommendation quality. the framework models semantic interests from reviews and captions for content-based matching, learns neighborhood patterns from the user–item interaction matrix for collaborative filtering, and applies a demographic classifier to mitigate sparsity for users with limited histories. The three signals are fused via a linearly weighted ensemble whose coefficients are optimized by the Bat Algorithm. In addition, Bat-based geo-clustering reduces the candidate space and improves locality-aware ranking without compromising relevance. Evaluations on real-world Yelp data show that the proposed approach outperforms reference models, including traditional CF and CB/CF hybrids as well as optimizer-based hybrids on precision, recall, and F1, while also yielding recommendation lists with higher diversity and novelty. These results indicate that combining geotagged photos and their metadata with textual reviews and demographic data, and using the Bat Algorithm to optimize model components and support geotag-based clustering, can substantially enhance tourism recommenders and improve user experience.</p>