A Topic Map (ISO/IEC Standard 13250:2000) is an SGML or XML document used for navigation - by people or by machine - within an information set. A single information set can include any number of types of source data: for example, HTML files, PDF files, and databases could all be part of one information set. A Topic Map (TM) can be either embedded within a document, or may be, itself, a separate document. A Topic Map provides a particular view of topics within the information set. There could be, potentially, an infinite number of different TMs for any information set: for example, an information set dealing with the plays of Shakespeare could have a TM for navigation within the perspective of Elizabethan culture, or the perspective of themes throughout the plays. Each TM explicitly declares topics and provides links to relevant information.
The TAO of Topic Maps
Literally, Tao means the way; the term is expanded to mean the underlying unifying principle of the universe. In the case of TM, the unifying principle underlies the informational universe of the Web. The TAO of TMs stands for topics, associations, and occurrences. A topic is a subject: in terms of Shakespeare, for example, Avon, Romeo and Juliet, and Anne Hathaway would all be topics. Topics are broken down into types: "Avon" would be defined as a town, and "Romeo and Juliet" as a play, for example. The word topic refers to both the topic link and the topic itself. Generally, topics have three types of characteristics: names, which explicitly identify topics; occurrences, which are topic-relevant information resources; and roles in associations, which are details of how topics are related.
TMs are being heralded as an important information management tool for the development of the Semantic Web, because they can enable very precise data access - they are sometimes called "the Global Positioning System (GPS) of the Web" - within a vast, complex, and intricately interconnected global information repository. One way of looking at the way a TM works is to think of it as similar to a book's index, and to compare seeking specific information within an indexed and an unindexed book. Although the information available on the Web now is linked, it is not linked in a systematic, standardized, and autonomously intelligent way (which explains why Web searches can often turn up so many irrelevant documents). Eventually, Topic Maps may be part of a comprehensive, user-friendly index to the body of human knowledge.