(5) Content and Structure

• Content is what matters most
  • content itself often has some internal structure
  • content may explicitly link to other content
• Macro-structure often is more representation than content
  • displaying the current context of the content
  • displaying related content
  • displaying some overall structure (site navigation)
• Content often is reusable across application scenarios
  • publishers have used Content Management Systems for a long time
  • adding a new publication channel to a CMS should be evolutionary
• Content may require very different support depending on the channel
  • newspapers need fine-tuned layout and good control over content size
  • Web needs good interlinking and navigation

(6) CMS Evolution

1. Web servers reading from files
2. Web servers implementing primitive content management (SSI)
3. Scripting languages implementing better management
4. Management code getting hooked up to databases
5. Better handling of client-specific behavior
6. Databases getting more diverse (RDB, XML, RDF)

(8) Serving Content from Files

(9) The Rise of the CMS

• File-based content management works well for small sites
  • simple site structure and small number of files
  • redundant parts can be manually synchronized
  • no software is required other than a Web server
• Web servers soon developed rudimentary CMS functions (SSI [http://httpd.apache.org/docs/2.2/howto/ssi.html])
  • rudimentary support is better than no support
  • managing a non-trivial setup with SSI still is a challenge
  • SSI allows includes but no backlinks and thus hides dependencies
• Content management is very similar to code management
  • simple setups require no or little tool support
  • serious projects need tools to manage dependencies and changes

(10) Serving Content from Files with SSI

(11) Files (Opaque Chunks)

• All major operating systems have file systems
• Files are typically treated as opaque chunks of data
• Applications may have special knowledge of file contents
• Advantages of files:
  • universally supported across major operating systems
  • storage and user management comes for free
  • all that is needed for a Web site is a Web server
• Disadvantages of files:
  • content access requires file system access
  • setting up parallel servers requires additional effort
  • no support for managing structure, everything handcoded

(12) File Systems are Databases

• A file system is a simple hierarchical database
  • it does not know data types and simply stores any content
  • its structure is a tree with a few extra tricks (such as symlinks)
• Many scenarios have much more structured data models
  • products, people, financial institutions all have complex data models
  • content should be stored and queried based on these models
• Databases are better optimized for storing structured content
  • better methods for structured storage and retrieval
  • better strategies for managing large datasets
  • sophisticated tools for access control, backup, and versioning

(13) Tables (Relational Model)

• Most widely used model for large collections of structured data
• Very mature products and many skilled people available
• The biggest advantage is that it is not hierarchical (no structure bias)
• Advantages of relations:
  • well-understood model and maps well to existing data
  • the non-hierarchical model allows views from different perspectives
  • highly scalable solutions available
• Disadvantages of Relations:
  • bad for sequences and variable structures (choices, repetitions, …)
  • very bad for structured documents

(14) ER Model

(15) Ordered Trees (XML)

• XML [Extensible Markup Language (XML)] has a heritage of document processing
• XML tools can be used standalone and are widely supported
• XML and HTML have a very similar foundation
• XML has two built-in directions: hierarchy and ordered children
• Advantages of XML:
  • maps well to HTML and XHTML
  • well-suited for document-oriented content
• Disadvantages of XML:
  • not good at representing non-tree data
  • databases not as mature as relational products

(16) XML Content

The term Mixed content in XML refers to elements which have text content mixed with elements [http://www.w3.org/TR/xml/#sec-mixed-content]. What these elements do depends on the elements , but the important point is that they are on the same level as the text nodes of the mixed content.

(17) Directed Graphs (RDF)

• RDF is the metamodel of the Semantic Web [Semantic Web]
• Highly granular, less rigid than tables, less ordered than trees
• Advantages of RDF:
  • any structure can be mapped to RDF triples
  • support still limited but getting better
• Disadvantages of RDF:
  • no model for document boundaries and self-contained units
  • bad for sequences
  • very bad for structured documents

(18) Choose a Matching Metamodel

• Content has some inherent metamodel properties
  • forcing that into a different metamodel is possible but unwise
• Using a metamodel which best matches a model is crucial
  • if you have large collections of rigid and highly-structured data: Tables
  • if you have structured documents with rich text: XML
  • if you have fine-granular graph-structured data: RDF
• Mapping is always possible but has severe limitations
  • things that work effortlessly in one metamodel may be awkward in another
  • there is no such thing as the one metamodel for all needs
  • RDF's claim to be the one metamodel for everything is not backed by facts

(20) Deconstructing Management

• What is managed?
• Who is managing it?
• What are the management support functions?
• Are there workflows and processes?
• Is the management integrated with other processes?
• Is it likely that processes will be followed?

(21) Managing Content with Files

(22) Integrated Management Functions

• Separating management and publishing does not work well
  • typical examples for workflows are review and release processes
  • oftentimes publishing-specific roles are required
• Integrated management takes over all tasks
  • What is managed? Database of structured content.
  • Who is managing it? Registered users based on roles.
  • What are the management support functions? Building a site around the content.
  • Are there workflows and processes? Can be based on users/roles/content.
  • Is the management integrated with other processes? APIs allow extension/integration.
  • Is it likely that processes will be followed? Easier to use than homegrown methods.

(24) System Platform

• Systems need runtime environments
• CMS are programs that are installed on some OS
  • integrated Web server or connect with Web server
  • integrated database or connect with database
  • integrated use management or connect with user management
• Typical steps for setting up a CMS
  1. setting up the runtime environment
  2. installing the CMS software
  3. initializing the CMS installation
  4. migrating existing data into the CMS installation

(25) Drupal

(26) MarkLogic

• XML-based content management
  • content is stored in XML documents in an XML database
  • non-XML content can be stored as well
  • programming is done in XQuery using extension functions
• XML-based content management works well for documents
  • XML's ordered trees are well-suited to represent documents
  • MarkMail [http://markmail.org/] provides access to well-indexed data [http://markmail.org/search/?q=erik%20wilde] of 7,227 mailing lists
• XML databases are more efficient than XML files
  • XML content is indexed and access is faster
  • building server farms is supported by the database management system

(28) Content vs. Web Pages

• Managing content is one of the prerequisites for Web publishing
• Content management can be done generically or Web-specific
• WCMS provides specific support for Web publishing tasks
• The most important part of a CMS is its metamodel