Computational Models Enhancing Semantic Access to Digital Repositories

Abstract

The growing amount of heterogeneous digital repositories has created a demand for effective and flexible techniques for automatic multimedia data retrieval. While the primary type of information available in documents is usually text, other type of information such as images play a very important role because they pictorially describe concepts that are dealt with in the document. Unfortunately, the semantic gap separating the visual content from the underlying meaning is wide.

The main goal of the project concerns the investigation of machine learning approaches to improve the semantic access to multimedia repositories by combining information gathered from the textual content with the one coming from pictorial representation. Furthermore, they have to be scalable, efficient and robust with respect to the inborn high-dimensionality and noise in the data collection.