Authors: Ghada Ahmed and Fatma Meawed
Crowd-sourced location-based information is rich of local and socially relevant content in different areas around the world. Despite its appeal, utilizing such content for mobile augmented reality browser results in an overwhelming experience for the users. The Augmented reality view becomes cluttered with content making it difficult for users to discover what can be truly interesting in their surrounding area. In this work, we present a method for POI recommendation according to users' topical interests. We discover users' topics of interests from their twitter feeds by classifying them against a trained model to build a topical user profile. Afterwards, the selected topics are matched with the relevant categories of the crowdsourced POIs. We discuss the results of our topical profiling of users and how it is integrated in an augmented reality browser. Our method for user profiling can be applied for the recommendation of any categorized items.
Keywords: User profiling, Social media analysis, Location-based recommendation, Information filtering, classification, personalization, Mobile Augmented Reality