\section{Conclusion and future developments}
\label{sec:conclusion}

\begin{conclusion}
In this paper we presented an approach for dealing, at least to some extent, with the information overload issue both on the Web and on the Desktop, and having as target early stage researchers. Our solution, inspired from the typical process of getting familiarized with a particular domain, combines elements from the Linked Web of Data and the Semantic Desktop, using semantic metadata as a common denominator. The result consists of three steps (\emph{extraction} -- \emph{expansion} -- \emph{integration}) that incrementally enrich the semantic metadata describing a publication, from no metadata to a comprehensive model, linked and embedded within the personal information space.
\end{conclusion}

\begin{conclusion}
Each step has associated a series of open challenges that we intend to address as part of our future work. As currently the \emph{extraction} works only on publications published as PDF documents, and formatted preferably with the LNCS and ACM styles, we plan to improve extraction algorithms to accommodate any formatting style, as well as develop new extraction modules for other document formats, such as Open Document formats. At the moment, the \emph{expansion} uses only two Linked Data repositories, i.e. the Semantic Web Dog Food Server and the Faceted DBLP. Future developments will include also other repositories, in addition to means for creating ad-hoc mash-ups between them, thus allowing the user to see data coming from different sources in an integrated and uniform view. Last, but not least, we plan an even tighter \emph{integration} within the Semantic Desktop, therefore enabling more meaningful queries and a richer browsing experience, and ultimately a complete automatization of the process, thus reducing the overhead to the minimum possible.
\end{conclusion}