Ilaria Bartolini's research activity primarily focuses on collaborative filtering, learning of user preferences, similarity search and browsing techniques for very large collections of “non-conventional” data, notably multimedia data such as text textual documents, images, video/audio streams, time series, and, more in general, any interesting “patterns”.
More in details, it concerns issues related to the definition of effective and efficient techniques for the retrieval of data of interest by exploiting both the automatic characterization of the content by means of low-level features (for example, the set of relevant words of a book, the color distribution of a photograph or of a key-frame of a video clip) and its semantics. She has developed and spread worldwide query models based on these characterizations, together with efficient and scalable query processing algorithms for highly differentiated users.
More recently, the same techniques have been successfully applied to the context of “authorship attribution” and, more in general, within the “cultural heritage” domain, and to the health context for the automatic detection of diseases.
