... And Why Shouldn't I Call It That?

Let's start with the broadest, most ridiculous term first: online indexing. Everybody has a different understanding of what this means, so I am going to make a suggestion right off: Don't use this term.

If you're using a computer, you're indexing with computer assistance. Not only is this a broad concept, but it's even broader than our original online indexing! But when people use the term online, sometimes they aren't talking about the Internet at all, and so "computer-assisted" seems more precise than "online."

One specific example of paperless indexing is embedded indexing. This form of indexing uses software to insert indexing information (known as "tags" or "markers") into the documents that are being indexed. The characteristic that defines embedded indexing is that the indexer does not need to be concerned with locators (page numbers, for example) at any time. This is one great advantage to indexers: it diminishes the workload by removing the need to type or verify page numbers. This simplifies and shortens the overall production process, because with embedded indexes, the indexer can complete the job before pagination is finalized -- something that by necessity waits until the very end of the process. Thus many embedded software applications will automatically determine the page numbers before going to a printer.

Another advantage to embedded indexing is reusability. Since the embedded index tags are combined with the text, that text can be reused (for example, in new editions or in related texts), and the index tags will get reapplied automatically. In this way, duplicated information is indexed for only the first time.

The major disadvantage to embedded indexing is that it is relatively complicated and/or time-consuming to edit the index tags. Editing an index involves opening the text files, finding and changing the index tags, and then regenerating the index to include the edited index tags.

If you are publishing in a medium other than paper, however, embedded indexing is mandatory. In fact, even if you are taking a document originally published on paper and converting it for computer use (such as CD-ROM reference books), embedded indexing must be implemented. This is because there are no true page numbers in computer documents. Instead of page number locators, computer documents use anchors, which are used in place of locators in the index. Embedded index tags can be easily converted into these anchors. (Anchors are required for any type of link within the document. For example, anchors are used to link lines in a table of contents to document or chapter titles. Similarly, an in-text cross reference requires an anchor at the reference's destination.)

I think that Web indexing is interesting because the final index is optional.

If you do have an actual index, the index is usually a Web document that contains several ordered links to other Web documents. Search engines, for example, generate online annotated indexes: the user types in certain keywords, the search engine software inspects its database for occurences of these words, and a list of Web addresses with other data is presented to the user. This list is an index. (For example, there is such a list at the bottom of this document.)

Often, however, web indexing (or indexing the Web) involves creating links among documents to aid in navigation. This document, for example, has links within the text that can direct the reader from one paragraph to another in a nonsequential manner. Thus by a general definition, this document is indexed. I think this is faulty reasoning, but this is because people have different ideas of what an index is. If an index is simply a navigational tool, then this page is self-indexed. If an index is a self-contained navigational tool, then all the links should be in the same place, as described in the above paragraph.

In most cases, however, indexing the Web simply involves the creation of hyperlinks within documents that are accessible on the Web.

The term online help has two implementations. The first, which is simply known as "online help" and nothing else, is recognizable because there is usually some sort of manual that can be read in full in some sort of order. There is also often an alphabetized or other ordered list of concepts and words. The second, context-sensitive help, has no suchreadable "book," but instead is comprised of independent bits and pieces of information that become available to the user when appropriate. When discussing "online help," it is crucial to understand which implementation is being used. I strongly recommend that the terms "context-insensitive" (or, if you're a marketing person, "context-independent") and "context-sensitive" be used exclusively. By the way, nowadays only one implementation is used in an application; having both is both redundant and wasteful, and writing both takes twice as long (the language used for each implementation is different).

Context-independent online help works under the same principles as indexing, but rarely is the help material written by an indexer. Context-independent help utilizes help "screens." These screens, presented to the user like a book, contain several "pages" of information, available in some order. You can find information on what you are looking for by browsing a topic index -- a list of important concepts, ideas, and titles -- or by following a subject or task hierarchy. Content-independent help usually is accessible by selecting a Help (or "?") button or pulling down a Help menu. How the material is organized is up to the authors and/or designers.

This is a whole other indexing field, and it has almost nothing to do with anything else in this document. Computer-generated indexes are indexes that do not involve the human decision-making process (or at least, not until the very end, when the index is reviewed). This is because the software creates the index based on certain algorithms designed by programmers. The software inspects the text, make certain determinations regarding the importance of text elements (based on placement within sentences, format, the number of times the term is repeated, and so on), and writes the index from scratch. The subsequent list is known as a concordance and should not be confused with an index. The single advantage that a concordance has over an index is the speed with which one can be created. Where an index might take a few weeks, a concordance can be generated in minutes.

But -- and this is a big one -- although index-generation technology is improving, concordances are astoundingly inferior to human indexing. In fact, using index generation software is considered downright bad practice for almost all projects, and many indexers shun the thought of computer-generated indexes to begin with. This is because computers have yet to demonstrate an understanding of language, a prerequisite for good indexing. Although concordances can identify words and terminology, they fail to understand concepts. They also are unable to distinguish between levels of importance, and thus tend to index trivial mentions as valuable and fail to differentiate between general ideas and specific examples.

Index