Session 11

The Essentials of Delivery Systems, Peter Hirtle, Cornell University

Image management systems are a good way to retrieve, display and navigate the many scanned images produced by digitization processes. Image management systems may also help acquire, store, index, secure, export, transform and preserve digital assets. In most cases image management systems will:

• Provide tools for searching (using descriptive metadata)
• Provide public and internal links to scanned images (using structural metadata)
• Provide the control elements needed for short and long-term access (using administrative metadata)

Image management solutions vary widely in complexity, performance and cost. The ideal solution for a given project depends upon the size of one’s database, the expected demand for images, the volatility of one’s data, the desired functionality of a given project and the availability of local technical resources. Other elements to consider include access to a controlled thesaurus, flexibility in database design, the expected life-span of one’s data and, if the lifespan is expected to be permanent, the potential for migration of data from one standard to another.

A few key concepts to keep in mind:

• No one system may meet all your needs. For example, it is common for metadata to be generated in one system and delivered in another.
• Metadata, not software, is the key. While applications will come and go, metadata is constant. The need to maintain metadata is perpetual and will allow for conformity with emerging standards, adjustment to new technical environments and addition of new functionality.
• One may get similar functionality via a number of different mechanisms. For example, the DLXS delivery system utilizes SGML; LC utilizes an inquiry engine; and Zylabs uses a full-text XML-based indexing system.

Advantages of using image management systems versus “home made” databases:

• Pre-defined data structure
• Built-in links to images
• Some are cross-platform
• Most have built-in web interfaces
• Overall, less programming expertise is required

And Disadvantages…

• Fixed, nonstandard, and/or proprietary data structures
• Limited customization possibilities
• The relative infancy of digital library software

back