From Emissions to On-Board Data
In 1996, hit with new requirements developed by the Environmental Protection Agency to deliver information on emissions in a structured format, the automotive industry scrambled to make sure they had the ability to comply. Andy Tweddle, vice-president of Tweddle Litho Company, who contracts with many of the automotive giants to develop the information for automotive manuals, was especially hard hit. He was confronted by the fact that nearly everything in an automobile is related in some way to emissions and that all their documentation needed structure. The industry began a wholesale migration into SGML. What began as a government requirement grew into a solid single-sourcing strategy that has enabled Tweddle to translate only 15 percent new information from one product release to the next.
For the information-development process to work, Tweddle had to rethink how they developed the information and find a tool to make single sourcing possible. They could have had the writers develop the information in isolation from one another, as many publications organizations do, but they decided instead to completely change the way their information-development process worked.
Tweddle needed a tool that would allow writers to create content but also differentiate information so that when the information went into production, only the right information from that model, that year, and that market made it into the deliverable. Fortunately, they already had the perfect tool on hand, XyEnterprise’s Content@, which they were using for a client.
XyEnterprise’s Content@, formerly Parlance, is a robust SGML-based, and now XML-based, content-management tool that facilitates content re-use and print publishing when used with their high-end publishing tool, XML Professional Publisher (XPP). Content@ offers robust content-management and single-sourcing capabilities at a significantly lower price than products like Documentum. Because Content@ was a requirement for another customer, Tweddle was already using it to reduce and manage translation costs by taking large, flat SGML documents and creating smaller, re-usable pieces of information. Changes made to one piece of information did not require retranslation of the entire document set.
The migration from re-using translated information to re-using English information across dimensions (model, year, and market) was obvious. Content@ allowed writers to assign metadata (using Frame+SGML or Arbortext Editor) to pieces of information, labeling them according to a certain model, year, or market. “Many cars are very similar, particularly US and foreign versions of the same car. Plus, different cars frequently use the same subsystems, like engines, radios, air conditioners, and so on. When we develop manuals, we have to make them as modular as the cars we document.”
Consequently, writers could no longer write an entire book in isolation. They now had to develop information based on automotive systems. For example, one writer is responsible for all information on heating and cooling for all three dimensions. The writer becomes the subject-matter expert in that domain and can determine easily what is common information and what is unique.
Integrated workflow software facilitates and automates much of the process and allows Tweddle to identify bottlenecks in their operation. With an easy-to-use wizard interface, Tweddle can identify where content will go and how content will be packaged once the authoring process is complete. Using a reporting system, progress is tracked and problems in the process identified.
Once information is authored, technical reviews are done collaboratively and online. PDFs are generated from the content repository in response to a workflow trigger. The PDFs contain both the common information and the product-specific information with the specific model, year, and market (the XML metadata) reproduced as a marginal note. All content is produced in context and, because the differences are noted in the margin, an editor or a subject-matter expert does not need to review and edit more than one version of the information. Also contributing to the success of this review process is that subject-matter experts within the automotive industry specialize by automotive systems. Delivering information in the way that subject-matter experts work has helped to sell the idea. But not all changes met with immediate approval from the automotive systems engineers. For example, another innovation Tweddle introduced was electronic markup of information. Instead of wading through multiple and often unreadable edits, writers and editors were able to streamline the process by having often unwilling subject-matter experts mark up information directly on the PDF. Attitudes changed toward the new review process once subject-matter experts realized the new system allowed more time for reviews, and they were no longer under the extreme time pressures they once were to complete the reviews.
Once reviews are complete and information is ready for translation, only the pieces of information that have changed since the last translation are sent to the translators. Depending on which automotive client the information is for, the new information is packed for translation. In one instance, information is sent automatically as part of the workflow system to a third-party translation company. The software creates a package, which includes the old English version, the new English version, the old translated version, and a “diff” file.
Tweddle estimates that only 15 percent of the text needs translation as a result of this process. Because Tweddle translates into twenty languages, they have seen the most significant cost savings in translation.
Deciding that their process could use further refinement, Tweddle developed other ways to make translation more efficient. Writers consider translators their first customers by making sure they are reusing as much content as possible, writing from a minimalist perspective, managing terminology, and including as many graphics as possible. Tweddle has been so successful that translation memory tools, which work by “remembering” text that has already been translated, contribute to only about 10 percent of their translation. The new content they deliver to translation is so unique that it is not a part of that “remembered” database of text!
Currently, the main mode of delivery for automotive owner guides is paper (the manual you have in the glove box of your car), but some customers require publication into Web-optimized PDFs.
Book builders, the next stage after writers, editors, and illustrators, compile the information and create an outline from the database for production. Using the XyEnterprise product XPP, they have automated much of this process. Very detailed style templates created in XPP determine much of the print and PDF output, and book builders verify only that the output is correctly laid out. Production can be a batch process with very little intervention.
The next step for Tweddle, like many in the technical-publication industry, is dynamic output.
“In just a few years, new cars will be sold with computer screens capable of displaying text right inside the car. We need to be able to make electronic versions of our manuals that can be displayed and updated as changes are made,” said Andy Tweddle in an interview with Imaging and Document Solutions.1
Tweddle has everything in place to make dynamic output a reality. Their content, which is already modular and in XML format, can easily make the transition from traditional print output to anything the future may hold for the automotive industry.
“What began as a government requirement grew into a solid single-sourcing strategy that has enabled Tweddle to translate only 15 percent new information from one product release to the next.”