How does SGML translate into so many formats?

Conventional documentation programs (word processors, desktop publishers, multimedia) each have a different way of storing information, which means they cannot easily exchange the same documents. SGML can translate into so many display formats because, in its raw format, it is not concerned with the way in which the contents of a document are displayed, but rather with the types of information a document contains.
The information in an SGML document is interspersed with a special type of meta-information, known as tags, which describe the information's structure and content. These descriptions may be used by a converter or a browser (such as SoftQuad's Panorama Publishing Suite) to define the display of the various types of information contained in the document.
A single SGML file may contain any kind of information - text, graphics, video, sound, even the contents of a database.
An SGML converter can be designed to isolate any number of individual information components of in SGML document, and produce a separate new file for each one, catered to fit the appropriate publishing platform. An SGML browser can be configured to immediately display each different kind of information it finds in an SGML document, whether that means playing a video, showing an image, or printing a hardcopy.

How Does SGML Work?

SGML is a simple, text-based annotation (or tagging) standard used to describe the contents of a text or multimedia document. An SGML document consists of content (information) and annotation (mark-up).
One part of a typical SGML document may contain basic text, another part may contain a specific kind of text, and yet another part may refer to a video file, and there are annotations (""). The annotations describe what the information is, not what it should look like. The content can be any form of electronic information.
An SGML file, then, is a collection of information with tags that describe that information.

Follow the Rules

While the information in an SGML document may be described with SGML annotations (mark-up), it may only be described according to a specific set of possible tags. Every SGML document must be accompanied by a kind of legend or key which explains the mark-up used in that document.
Each individual legend, or set of predefined SGML mark-up, is known as a Document Type Definition (DTD) or a Rules File.
Different DTDs may be created for different types of documents, but each individual document must follow only its own DTD. An SGML DTD outlines all the possible mark-up for particular class of documents and specifies where the different mark-up can go; An SGML document is made up of electronic information described by some or all of the types of mark-up defined in the corresponding DTD.