Chapter 5: Realizing a Metadata Framework 141 1. Attributes, that is, tags 2. Context information 3. Associations or relationships between content objects 4. Usage history or events. Recall the discussion related to derived metadata (section 4.4), where derived metadata seeks to increase the abstraction level of the content. Of the above, attributes are metadata, whereas all the other classes are derived metadata. We see context information as an important source of metadata, especially as far as mobile use is concerned. Context usually does not describe the content object itself, but the situation where it was created or enjoyed. Context information plays an important role when creating automatic collections and playlists and trying to fi nd lost content objects. Section 4.7 provides more discussion on how the contextual metadata benefi ts the user. Associations are the kind of metadata that requires more than one content object. They describe how two or more objects relate to each other. A whole section about relationships will follow, as we consider them to be the most important form of metadata. Usage history is metadata describing how the content has been used and interacted with. It includes use count history, such as the playcount for a song; it also reveals when an image was edited, or how many multimedia messages were sent to a certain recipient. Among other tasks, usage metadata can determine what is the most important (read: most often used) content object in a mobile device. It is not always trivial to assign a given piece of metadata to one of the categories listed above, yet it is not particularly important, either. This division is included to show that a wide range of metadata, with a rich set of characteristics, exists beyond mere tags. In addition to media objects, applications also have metadata, such as version information, author, and compilation date. Metadata is generally understood as data about data; in this particular case, it is data about an application. 5.3.1 Tags Most common metadata are simple attributes, that is, tags. Some tags have more or less loosely defi ned semantics, such as ID3 tags for music fi les or IPTC tags for images. These attributes are often called (key, value) attributes since they are used in pairs. An example