Petr Sojka: Position Statement
Mathematical community dreams about WDML already for a long time. Partial successes are already there as the `local' DMLs as NUMDAM, Euclid, DML-CZ, CEDRAM, RusDML and many others. This is, however, far from possibilities available today technologically on a global scale, as e.g. Google Scholar (or PubMed Central) shows.
From the business point of view, I suggest to proceed analogically as biomedical community and NiH when they created tremedously successful PubMed Central library and take their experience when building similar library for mathematical sciences. The differences are there (mathematical community is orders of magnitude smaller, it is more heterogenous, math notation needs special tools as math OCR, math formulae indexing and search, etc.), but this example shows a success story to follow (setup policy, heavy financial support of technical developments and tools, publishers' involvement and "motivation" by achieving critical mass). It seems that in mathematics there is no strong player like NiH to take a lead.
On the technological side, there are specific challenges in lowering semantic gap between available DML content (80% are images, and for the rest the are mostly only PDFs) and a representation of mathematical knowledge needed to fulfill modern digital library services (math-aware search, classification, similarity search etc.). These areas should be researched and appropriate tools developed to serve math community needs as part of WDML. One significant cause of PubMed Central success was initial investment of NLM DTD and its suite of programs to ease the input barrier for publishers and reach the critical mass of content quickly.
I am keen to work towards the goals above (as I have been doing in DML-CZ and EuDML projects sofar).
