Location-Aware Tag Recommendations for Flickr

Flickr is one of the largest online image collections, where shared photos are typically annotated with tags. The tagging process bridges the gap between visual content and keyword search by providing a meaningful textual description of the tagged object. However, the task of tagging is cumbersome, therefore tag recommendation is commonly used to suggest relevant tags to the user. Apart from textual tagging based on keywords, an increasing trend of geotagging has been recently observed, as witnessed by the increased number of geotagged photos in Flickr. Even though there exist different methods for tag recommendation of photos, the gain of using spatial and textual information in order to recommend more meaningful tags to users has not been studied yet. In this paper, we propose novel location-aware tag recommendation methods and demonstrate the effectiveness of our proposed methods.