Chapter 5: Realizing a Metadata Framework 195 the processor may refuse to give the data to some applications it does not trust. Additionally, as well as accepting or refusing, the processor can modify the data. For instance, it can blur the property data before it is delivered to the application from the Metadata Engine. Blurring can be used to make some private data less private. For example, instead of giving the exact location where the picture was shot, only the country is returned to the requesting application. Similarly, instead of giving out names of the participants in a calendar invitation, only the number of participants is given out. Processors need to be written by those who create their own meta- data defi nitions. The framework will not associate a processor to any metadata property, but only to those that belong to a namespace that is owned by the processor writer. Yet again, this requires that the underlying operating system can enforce authentication of processor DLLs and every use of the Metadata Framework API calls. Another usage for the processor, beyond privacy control, is assist old applications to use new metadata defi nitions. Applications devel- oped prior to the releases of some new metadata defi nitions cannot use the new defi nitions, as they cannot know what they mean or how to use them. Since processors are basically just applications and their writers know the purpose of the new defi nitions, the processors may, for instance, use its own dialogues, displays, etc. to show meaning of the metadata if it makes sense. A simple example would be if someone creates a new type of location metadata. If an application has no idea how to use the new location defi nition, the processor attached to it might show the location using its own maps or other mapping applications. 5.10 Summary In the previous chapters, we discussed how metadata is essential for content management. In order to solve problems in content manage- ment, and especially those that distributed nature of mobile content brings, we also need to learn how to manage metadata. We showed the variety of the metadata ranging from simple tags to context and relationships. We discussed how it is important to make the metadata management framework something that the operating system takes care of, instead of the applications themselves, and why it is important to separate metadata from the content binaries. We described our extensible ontology and discussed how it is important to defi ne semantics of metadata as accurately and strictly as possible