94 Personal Content Experience: Managing Digital Life in the Mobile Age image belongs to is derived metadata, as is the listening history of an audio fi le. Derived metadata seeks to raise the level of abstraction of the data in such a way that some of its aspects are closer to human thinking, as opposed to computer data structures. Today, such abstraction rou- tinely takes place in pattern recognition, image analysis, speech analy- sis, statistical analysis, or any processing that translates raw data into meaningful indicators. As abstraction methods improve, we expect more analysis functions that can take personal content fi les as input and produce nearly human-level abstractions of those fi les as output. We call such functions harvesters and discuss them further in Chapter 5. 4.5 How does Metadata Benefi t the User? From the perspective of a mobile device user, the main advantage of metadata is that it makes personal content easier to manage and thus 24 the device easier to use. The metadata facilitates a paradigm shift from managing content instead of fi les and it has emerged because of many reasons (recall the discussion about fi le management versus content management in section 3.1). First, the number of fi les has skyrocketed and the fi les are often distributed in several devices. Second, because the content is created and edited by many people, mere fi le attributes (such as fi lename, date, size, and access rights) cannot describe the subtle differences between the content objects without understanding the information that the objects contain. For example, try telling the differences between the successive two images with names like IMG_2745.JPG and IMG_2746.JPG, without looking at the contents. There are several uses of metadata. For instance, it is an important enabler for searches, especially when searching for media objects that do not contain any text. Furthermore, it can automate various tasks related to organizing, summarizing, and viewing of the content. Next we introduce some of those operations that profi t of metadata. Our treatment of these operations builds on the classifi cation by El-Sherbini (2000). 24 A “paradigm shift” refers to a fundamental change in thinking. We believe that focusing on content instead of fi les is a fundamental change indeed.