We just rolled out a major upgrade to the PDFs published in our journal Semantics and Pragmatics. We already had extensive hyper-linking in the articles and in the bibliographies. Now, our PDFs sport rich meta-data that will make it easier for all kinds of services to recognize crucial information about our articles. The meta-data are embedded in the PDF but not visible to the naked eye. You can inspect them by probing into the document properties with Adobe Reader or some other tool. But the main purpose is for automated access by search engines and bibliographic software. If you drag one of our PDFs into a database maintained by Papers, for example, the information about the article will automatically populate the fields in the database, with no need for manual filling in of fields. Similarly, if you use JabRef to maintain a BibTeX database, you can import our PDFs via “Import > XMP-annotated PDF” and have all the BibTeX fields filled in automatically.
Behind the scenes, we are using a customized version of the hyperxmp package to embed XMP meta-data in a number of standardized formats (Dublin Core, PRISM, BibTeXmp).
We are in the forefront of scientific journals in implementing these meta-data standards. To our knowledge only the Nature Publishing Group and Elsevier are also consistently providing rich meta-data in their PDFs. Certainly, none of the other journals in our field have moved to these standards. The upshot for S&P authors is that their work is made even more accessible and useful for readers, in a way that is far ahead of competing journals.
[We thank Uli Sauerland for asking us whether our PDFs could embed bibliographic information in addition to us making BibTeX entry information conveniently available from the online abstracts at the S&P site.]