Metadata Models for the Web

Metadata has not been invented recently. Data that man has ever exchanged with his peers for the first time had the metadata associated with it. Metadata is the answer to the question asked, “Where did you get this information?” Whether observed or untold, it has always existed side by side with data. Due to its nature, and since it also gives us information about the other data its referring, people either categorized metadata as data, or completely ignored it.

As we entered the information age by the end of the century, vast amount of information generated resulted equivalently important and vast metadata to emerge, which enforced researchers to separate metadata from data, and develop new specifications and technologies to manage it.

The Web is an offspring of the information age. Its extensive usage created world-wide libraries in forms of Web pages and digital multimedia content. Any individual or institution that conducts research on the Web and the Web content must either adopt technologies to produce, manipulate, or contain the information available to them, or develop technologies that will integrate or communicate legacy systems with the new ones. Consequently, organizations and working groups have developed specifications and technologies to model and manage metadata for the Web content using the same means information is represented on the Web. This results the unavoidable fact that metadata in some cases may locate along with the data its referring.

In the following sections, we describe several metadata models for the Web. They have different characteristics as well as different uses. First, we introduce the most common method of populating, or describing data with, metadata: attribute-value pairs. Attribute-value pairs are also the simplest among the others. The Dublin Core Metadata Set is a widely used method to populate metadata with attribute-value pairs for the Web content.

Second, we introduce RDF for metadata statements that associate metadata with the Web content externally and with certain commonly accepted semantics. That means that content developers can assign IDs to their content, populate these IDs with metadata, and build relationships with other content, whose semantics are well defined and accepted world-wide.

Later, we describe the Semantic Web efforts. The Semantic Web is a web of Web resources where metadata specification languages based on RDF describe its nodes, edges, and actions to perform on them. Such developments allow people to automate software processes to make intelligent searches on the Web.

All the methods mentioned above are general purpose metadata modeling specifications that help developers to model their own metadata about their Web content in accordance with open standards. However, institutions or interest groups can create their own open or proprietary metadata models as well. Many organizations choose to: either adopt the modeling methods, or develop their own models, or combine both.

    Attribute-Value Pairs and The Dublin Core
    Metadata Statements and RDF
    Schema-based Structured Models
    Summary