To achieve these cost savings, there are several steps. First, to reduce the total amount of information that you send to a translation house, you would break up your information into smaller modules. Often, the size of information modules that you choose for reusing information works well for reusing translations too.

Second, to simplify the translation process, you may wish to simplify the terminology and sentence structure that your authors use. Several companies provide software to enforce simplified English, such as Smart Communications (www.smartny.com/).

Third, because you will be managing many components in multiple languages, you may want to implement a content management system from one of our repository partners (www.arbortext.com/partners/index.asp).

Finally, you will want to obtain translation software from a company such as Trados or work with a company that specializes in translating XML such as OmniLingua.

Automated publishing software applies a stylesheet to your content to produce a Web, print, PDF, or other version of your information. Multiple technologies are required to support multiple types of media, although a few vendors such as Arbortext bring these together under a single architecture.

To produce documents automatically, your requirements may extend beyond the formatting capabilities of the technology you're using. For example, you may want to generate content automatically, either by deriving it from your document to produce, say, a list of figures, or by extracting it from a database, such as pulling part descriptions based on part numbers.

Delivering information in multiple languages adds another dimension of complexity. You need to be sure not only that the publishing software can support the composition of all of the languages you need, but that it also provides other capabilities in support of those languages such as hyphenation and index sorting.

To provide different information based on the audience, we could spend an entire article - and recently did (www.arbortext.com/resources/xpn_oct_03.htm) - discussing the ways to approach building a system to provide customized information on demand. The main point of the article was that you could either use a "master document" approach or an "assembly" approach to creating customized views of your information.

Since the master document approach is easier to implement, let's consider it in more detail. From a master document, you can publish subsets of that document for different audiences. For example, if you make products that come with a variety of options, you could tailor each customer's owner's manual to match the specific combination of options that the customer selected.

To do this, you first add information to your document about the audience for each part of that document. You can customize information all the way down to a word in a sentence or a cell in a table. Then, when you publish the document, you select the audience and the publishing software uses your selection plus the additional audience information you entered in the document to decide which information to omit from the document for publishing.

The value of such services increases with the amount of information you're managing, the number of people who contribute or edit your information, the geographical dispersion of your people, and the complexity of your processes. About two-thirds of Arbortext's customers have implemented a CMS along with their implementation of Arbortext.

Considerable effort is involved in setting up a CMS, such as assigning users and permissions, setting up directories, designing workflows, and implementing other automation. You may also want to customize the metadata to suit your specific needs.

One of the main issues to consider is whether your organization or your vendors are responsible for updating your integration as software versions change.