This document reviews approaches for user personalized tag-based image search. It discusses how tag-based image retrieval (TBIR) uses manually assigned tags to represent and search for images, as opposed to content-based image retrieval (CBIR) which matches images based on visual similarity. The document proposes a system that combines TBIR with CBIR, using both semantic tags and visual features to rank images for a user. It also discusses using clustering, common tags, and user profiles to filter images during the retrieval process. Experimental results showed that incorporating semantic and visual information improves diversity in the retrieved results compared to tag-based retrieval alone.