Multi-Level Visual Similarity Based Personalized Tourist Attraction Recommendation Using Geo-Tagged Photos

Geo-tagged photo-based tourist attraction recommendation can discover users’ travel preferences from their taken photos, so as to recommend suitable attractions them. However, existing visual content-based methods cannot fully exploit the user and information of photos extract features, do not differentiate significance different photos. In this article, we propose multi-level similarity-based personalized using geo-tagged (MEAL). MEAL utilizes contents interaction behavior data obtain final embeddings users attractions, which are then used predict visit probabilities. Specifically, by crossing define four similarity levels introduce a corresponding quintuplet loss embed addition, capture self-attention mechanism representations attractions. We conducted experiments on two datasets crawled Flickr, experimental results proved advantage method.