Online database of all mathematical literature. Featuring a comment system and the possibility of adding reviews of your own, and links to fulltext files and to related material.
Inspiration from MO, for example use of reputation and use of votes in order to get the best contributions near the top.
How to add material automatically from new journal issues and new book publications? Could one simply use new MR items? Is there any part of MathSciNet which is publicly available and which could do this? Or would it be possible to ask MathSciNet to release the database excluding the reviews? If we aim for completeness, the only way is to use MR somehow (No! Not if people can create pages by uploading their bibliography files). In any case, a page should link to the correponding MR page.
Is there also a ZBL number for all items? Could it be easier to get information from ZBL?
There should maybe be a different type of page for each book series and for each journal, and maybe also for each publisher. Journal list can be obtained from AMS/MR - what about other lists?
Add all arxiv items - but how to integrate this with things published/indexed by MR? Include a section for the official link (using DOI?) and also a section for alternative links, like arxiv or author’s webpage. A possibility could be automatic linking to a Google Scholar search.
How to treat mathematical symbols in the title? What does MR and ZBL do?
How to treat unofficial notes, letters between mathematicians, etc?
Include a link saying My local copy, find a way to make this link to a pdf or djvu locally. Give users the option of logging in, make friends, and find fulltext files via friends? How far does Mendeley push this system? And should there be a way of allowing for multiple local copies (for example a ps file and a pdf file, one preprint version and one published version, one Russian and one English version, etc)? If so, should clicking lead to a page with a list of local files sorted according to relevance?
When logged in, one should be able to add labels, like “Read later”, “Star”, “Topology”, and create own labels.
Officially discourage stealing of MR reviews, and links to illegally posted material.
Can one get tex files off the arxiv? If so then a way to generate a rather complete list of literature items might be to scan the bibliography files of all arxiv items.
Is it possible find a way for example for a Chinese guy who happened to already have a local copy of MathSciNet to upload a massive lump of items and create pages for each of them? It should be possible to “submit” bibliography files, let the system check these against already created items, and (using in particular the MR number) integrate the information into what is already there.
Contacts: The German guy (Rehmann) who scanned things early, Andrew Stacey. See the discussions about similar systems on mailing lists/MO, maybe some people there would be interested.
Can one pay people to do bulk work via internet?
Create a way of extracting bibliography files from the site?
People could add errata.
Users should be able to create a watchlist so that they get an email whenever an article is commented upon. For example, an author can watch his own article and get an email when someone comments or asks a question.
Base the site on Zotero or some other such tool? Or integrate the site with such tools?
Compare this dream with what’s already available in Mendeley.
nLab page on 9 Dream notes July 2012