Response to question 4

There is a great need in the world where I live for more sophisticated IDM tools. Most people in my world are stuck with Oracle, MS Access, Web text-retrieval, or, if they’re really sophisticated, MySQL. Little of the great work being done in the IDM research community is accessible to people in my world. The problem, I think, is the (very sensible) tendency for each database research group to create its own system embodying its own unique contributions. So, if I need a database system to support semi-structured data, object-oriented data, geo-spatial data, text, and images, I have to collect software from several research groups and somehow make it all work together. This is too hard for most people.

It would be a lot easier if someone, or the community as a whole, would take on the task of knitting together the best ideas from lots of different research groups into a single comprehensive (or more likely, highly modular and extensible) system.

I see little hope that the commercial DBMS vendors will push in this direction anytime soon. Relational, and now the minimally object-oriented systems called object-relational, does a great job of supporting typical business database needs. This is the overwhelming majority of the market.

The research community has to take the lead in getting research results out into the world. Perhaps the Linux phenomenon can provide a model for this sort of effort. Would it make sense to envision a Linux-like approach to IDM? What I have in mind is a body of very high quality, open source software that provide sophisticated IDM capabilities.

Jeff Ullman, in one of his position statements, noted that no one seems to want deductive databases. This is a sad state of affairs. I’ve seen lots of problems that cry out for deductive capabilities attached to databases. I suspect that the non-acceptance of deductive databases reflects the non-accessibility of database research to the world at large. If there were an easy way for people to add deductive capabilities to their databases, I suspect they would do so.