The semantic enterprise

It's interesting to see, what vast amounts of data any given enterprise gathers. Even within a small or midsize company, a lot of different, mostly single-tasked and integrated databases, are created on an ad-hoc basis as the need arises.
In some applications, the fact, that a small, single-tasked database is included in the product, is pretty well disguised.
That's where portal technology comes to mind, that's where service oriented architectures could play the joker card and integrate those various data sources into a single, concentrated dashboard.
At first, however, those different applications need a semantic coherency to exchange and interpret contextual data. The danger with customized database applications is, that in most cases, customization only focused on the single task. A CRM application, that models a complete view of enterprise, clients and persons and their relationships is not so happy with a simple addressbook application that only focused on person names and phone numbers.
An ontology, documenting the complete data objects of the enterprise, is a way to build a semantic layer, interconnecting the different applications within a single language.
In my experience it has always been a hard task, to persuade the client of the necessity to focus on a semantic understanding of the data stored, retrieved and processed - even if it's just a simple form at first glance. Hope that changes with more enterprise level SOA.