Abstract
Selecting the right set of descriptors for the annotation of a specific dataset can be a hard problem in research data management. Considering a dataset in an arbitrary domain, an application profile is complex to build because of the abundance of metadata standards, ontologies and other descriptor sources available for different domains. We propose to partially automate the process of data description by generating application profile recommendations based on a research data asset knowledge base. Our approach builds on existing technologies for exploring linked data and results in a process which can be tightly coupled with the research workflow, giving researchers more control over the description of their data. Preliminary experiments show that we can build on state-of-the-art technologies for search indexes, graph databases and triple stores to explore existing sources of linked data for our profile generation.

Abstract
We build and analyze a coreference network based on entities from photo descriptions, where nodes represent personalities and edges connect people mentioned in the same photo description. We identify and characterize the communities in this network and propose taking advantage of the context provided by community detection methodologies to improve text illustration and general search. © 2012 Springer-Verlag Berlin Heidelberg.

Abstract
Text illustration is a multimedia retrieval task that consists in finding suitable images to illustrate text fragments such as blog entries, news reports or children stories. In this paper we describe a crossmedia retrieval system which, given a textual input, selects a short list of candidate images from a large media collection. This approach makes use of a recently proposed method to map metadata and visual features into a common textual representation that can be handled by traditional information retrieval engines. Content-based analysis is enhanced by visual abstraction, namely the Anisotropic Kuwahara Filter, which impacts feature information captured by the Joint Composite and Speeded Up Robust Features visual descriptors. For evaluation purposes, we used the well-established MIRFlickr photo collection, with 25,000 photos and user tags collected from Flickr as well as manual annotations provided as image retrieval groundtruth. Results show that image abstraction can improve visual retrieval as well as significantly reduce processing and storage requirements, even more when paired with Google's WebP image format. We conclude that applying a visual rerank after an initial text retrieval step improves the quality of results, and that the adopted text mapping method for visual descriptors provides an effective crossmedia approach for text illustration. © 2012 Springer-Verlag Berlin Heidelberg.

Abstract
The Web is being increasingly used by health consumers to search for health information. In this domain, the quality of the retrieved contents is crucial to avoid healthcare hazards. To address this problem and help the user identify reliable and credible contents, initiatives have appeared that certify the compliance of health websites to quality standards. In this work we explore the impact of medical certification on several aspects of health information retrieval performance. Moreover, we analyze the usefulness of certification categories to the personalization of the search experience. Our findings suggest that medical certification might be incorporated as a ranking criterion. We conclude that the medical accuracy of the resulting knowledge is enhanced by the use of certified information and depends on the users' comprehension of the document. In general, we also conclude that there is space for personalization in search by health consumers. © 2011 Springer-Verlag Berlin.

Abstract
The University of Porto has a well-established set of specialized libraries serving the research and student population of its 14 schools. Thematic digital libraries can be valuable for organizing specific collections and for supporting emergent communities. This work focuses on two case studies, one in the area of the Fine Arts and the other in the area of Food and Nutrition. For building both digital libraries we propose to use the existing university repository infrastructure and to establish a metadata workflow that makes use of available descriptions in the library catalogues and in the university information system. We expect that such an approach, which takes into account the institutional context and resources, can be used in other collections at our university and inspire similar initiatives elsewhere.

Abstract