I gave a presentation at the InfoTech for Pharma and Biotech conference last week, in a symposium dedicated to text mining, titled "Bringing a publisher's content to the Life Science researcher". It seems relevant to post this now with all the talk of WikiProfessional's WikiProteins* this week, so here are the slides. The most important point is that we need to develop a standard annotation format, so that document-specific annotations contributed to one store (Zotero, say, or any individual publisher or service that provides document annotation) can be connected together with everyone else's annotations.
WikiProfessional doesn't have to deal with sharing document-specific annotations and their positions, because it's not storing document-specific annotations. Also, I don't think it differentiates between different meanings of the same word/phrase, which is a bit of a problem.
* The WikiProfessional annotation of the WikiProteins paper is a nice demonstration of community annotation.