Reconnecting digital publications to the web using their spatial information

Digital publications can be packaged and viewed via the Open Web Platform using the EPUB 3 format. Meanwhile, the increased amount of mobile clients and the advent of HTML5's Geolocation have opened a whole range of possibilities for digital publications to interact with their readers. However, EPUB 3 files often remain closed silos of information, no longer linked with the rest of the Web. In this paper, we propose a solution to reconnect digital publication with the (Semantic) Web. We will also show how we can use that connection to improve contextualization for a user, specifically via spatial information. We enrich digital publications by connecting the detected concepts to their URIs on, e.g., DBpedia, and by devising an algorithm to approximate the location of any detected concept, we can provide a user with the spatial center of gravity of his reading position. The evaluation of the location approximation algorithm showed a high recall, and the high correlation between estimation error and standard deviation can provide the user with a sense of correctness (or spread) of an approximation. This means relevant locations (and their possible radius) can be shown for a user, based on the content he or she is reading, and based on his or her location. This methodology can be used to reconnect digital publications with the online world, to entice readers, and ultimately, as a novel location-based recommendation technique