Futher information: http://code.google.com/p/pdfmeat/
We present a novel approach to retrieve metadata to scholarly papers stored locally as PDF files. A fingerprint is produced from the PDF fulltext to query an online metadata repository. The returned results are matched back to identify the correct metadata entry. These metadata can then be stored in the PDF itself, indexed for a desktop search engine, and collected in a user's or community's bibliography. We think this hitherto missing link but with our tool now available data eases the organization of scholarly papers, and increases accessibility to one's collected academic content. http://code.google.com/p/pdfmeat BTW2009 Best Demo Award -- Thank you!