ARTICLE | Migrating to DITA – best practices for authors to consider before converting legacy content
When first making the move to DITA, there are some very important best practices that authors should consider before converting their legacy content.
When doing any sort of conversion from one format to another, the riding principle is always GIGO (garbage in, garbage out). If your legacy content is not particularly good, then converting it to DITA will only create indifferent content in a so-so DITA markup. The result: failure.
So what authoring best practices should you consider before converting your legacy content?
1. Topic-based writing
Your content needs to be able to split nicely into topics. The DITA model uses tasks, concepts, and references, but there are variations like super tasks, glossary, and scenarios that could be useful to keep in mind as well. However, if your content is still in chapter-like narrative form, the resulting conversion will be problematic—and your ability to leverage the advantages of DITA (like re-use) will be negatively affected. Worse still, if you have content that contains a mix of concept, task, and/or reference all in the same “chunk”, then it needs to be re-written.
2. Minimalism
Applying minimalism to content is vital. You could do it post-conversion, of course, but you can save a lot of time by doing it pre-conversion. Minimalism has three important facets to it:
- Write goal-based content: Writing goal-based content means more than just planning your content around tasks (although that, too, is important). It means identifying what your users will be doing with the product and writing those tasks. This is a departure from the feature-based content we often see. It’s the difference between writing the task “Using the Fine Tune feature”, which focuses on the product, and the task “Enhancing your Audio”, which focuses on the user. They might cover similar steps, but the focus of the task is on what the user needs to do, not what the product does. When you start identifying really good user goals, you end up writing about many different features, stringing them together so the user gets the information they need to complete that goal. If you have a lot of tasks that have “if-then” types of choices, you are probably mixing different goals into one task. You can often break those out into separate and more meaningful stand-alone tasks. Once you have your goals written, then it’s time to include the conceptual and reference information to support those goals.
- Provide only the information users need to perform the task inside the task and write nothing extra: This means writing one way to perform a task, not all ways. It means providing troubleshooting information in line with the steps where it can be useful. It means providing step results when they are informative rather than obvious. It also means not documenting the “cancel” button.
- Get rid of the chaff: This includes those rather unnecessary lead-in sentences like “This chapter introduces you to …” or “this table contains information about…”. When your topic and table titles are clear and your content is well written, you can completely remove those extra words (that users, ahem, skip over anyhow).
3. Navigation
Authors are frequently in love with cross references, especially inline cross references (example: “For more information about x, see link”). Users, however, are not so in love with them. One user recently referred to it as “falling into a spinning circle”. By the time they have followed the link from the original topic (which led them to a web of five more topics), they not only can’t find their way back to the original task, they can’t even remember what the original task was. I call it “the spaghetti mess of links”.
Links between topics should be kept to an absolute minimum and should only be inline in rare circumstances. If you need to link two topics together, the best place for that is at the side or at the bottom of the topic and often DITA will do that automatically for you. What are valid reasons for linking topics? When the user will never be able to guess that those two topics are related, or that the two topics are not “siblings” to each other in the hierarchy of content, or when you want to introduce a sequence between topics.
There are other valid reasons to include links, but the goal here is to keep linking to a minimum so that users will find the links useful. As a result, they will also pay more attention to links when you provide them. Another goal is to minimize dependencies between topics. You should be putting all links in relationship tables, which are linking mechanisms that live in the DITA maps instead of inside the topics. By removing the links from inside the topics and putting them at a higher level, inside the map, you leave the topics dependency-free to be re-used wherever necessary.
4. Structural monstrosities
Let’s face it: we sometimes do odd things to our content in order to organize it as best we can. The result is sometimes something too complex to convert well to DITA. One example is a table within a table. Another monstrosity is having procedures in tables for some understandable formatting and layout advantages. Clean up your monstrosities and make them pretty enough to go to the Ball. DITA will give you other ways to organize your content without that sort of complexity.
Links between topics should be kept to an absolute minimum and should only be inline in rare circumstances. If you need to link two topics together, the best place for that is at the side or at the bottom of the topic and often DITA will do that automatically for you. What are valid reasons for linking topics? When the user will never be able to guess that those two topics are related, or that the two topics are not “siblings” to each other in the hierarchy of content, or when you want to introduce a sequence between topics.
There are other valid reasons to include links, but the goal here is to keep linking to a minimum so that users will find the links useful. As a result, they will also pay more attention to links when you provide them. Another goal is to minimize dependencies between topics. You should be putting all links in relationship tables, which are linking mechanisms that live in the DITA maps instead of inside the topics. By removing the links from inside the topics and putting them at a higher level, inside the map, you leave the topics dependency-free to be re-used wherever necessary.
5. Structure without differentiating meaning
I have yet to see legacy content that doesn’t include formatting for something like bold, italics, or underline. They are each often applied for very different reasons though. For example, you might italicize a phrase to indicate that it’s a book name, or because it’s a first occurring term, or for emphasis.
In DITA, if you apply the <i> element to all of these different types of content, then you won’t be able to leverage the markup properly. The <cite> element is used to indicate a book or external reference name. A term might be put in the <term> element. Emphasis, on the other hand, is simply not an acceptable reason to change font weight anymore. Once you are using the right elements for the right sort of emphasis, you’ll be able to leverage the formatting for those items separately and do cool things like link your <term> elements to actual glossary descriptions so users have inline rollovers with definitions.
So get familiar with your DITA inline elements like <uicontrol>, <menucascade>, <cite>, <term>, <ph>, <dl>, and get them to work for you.
A good conversion method will be smart enough to make those distinctions for you or help you make them, but that’s a key piece of conversion functionality you should look out for.
6. Short descriptions
If you haven’t yet encountered the DITA short description, count yourself lucky. It is the only element that legacy content often doesn’t have. DITA best practices say that each topic (every one!) should have a short description. If you’re thinking, “Big deal, I’ll just put my first sentence as the short description everywhere”, then let me stop you right there. A short description is the single hardest piece of content to write in the DITA model.
A good short description is succinct (less than two lines for sure, better to make it one). It accurately describes the topic without repeating the title, and gives users just the right information that, when they see the title+short description combination, allows them to decide whether that topic is worth navigating to or not.
In all online outputs, the title+short description is visible every time your content is in a hierarchy (parent-child). That’s why it looks odd when some topics have short descriptions and others don’t. You should either use them everywhere or use them nowhere.
The short description is a powerful interpretive tool that lets you bridge what is often required technical jargon in the title with the terms a user might be more familiar with. It often leads to the “Oh, that’s what you mean” moment for users. Wield it wisely.
Good Results and Bad Results
The first example is an “as is” conversion.
This second example is the same content, but with ‘best practices’ applied.
Conclusion: If you follow these best practices before, during, or after your conversion, your content will become versatile, usable, and streamlined. Excellence in, excellence out is what we should all strive for.
ARTICLE | Getting great DITA conversion results
Although conversion is only a small part of your DITA adoption project, it’s the part that causes even smart people to break out into hives. How does one go from a regular, flat document to a series of DITA topics and maps with the correct markup? How do you do that without losing important information? What’s the best way to do it? What tools do you use? How do you start?
Learn the basics
My first recommendation is that, no matter which conversion method you end up choosing, get some basic DITA training first. You should understand what a topic is, the difference between topic types, how a DITA map connects content together, and the basics of attributes.
Even if you’re not doing the conversion yourself, you should have enough knowledge to recognize good results from bad. For example, you should know that if all your topics are the wrong types, the conversion needs to be re-done. One of the warning signs is if you have procedural information in a concept or generic topic with an ordered list (<ol>)—that should be a task topic, which has step and step-related elements you can leverage.
Visualize the results
There’s no way to get around it: your current content, now likely stored as chapters, books, and documents will become 100s or 1000s of topics, combined together using maps. The sheer number of objects that result from a conversion project often surprises people.
It’s important to visualize the end results of an entire document set being converted—you will have 1000s or 10,000s of files, plus graphics. However, you also need to visualize what an individual “topic” should be. Is it a chapter? Half a chapter? A few lines?
I tell my clients that it should be a “digestable” amount of content—enough for users to get their business goal completed (learn about X, perform Y, look up information on Z) but not so long that it is too big a bite and gives them heartburn. The length really does depend on the business goal but on average, if you were to print out a topic to PDF, it would be about ¾ of a page long. Of course, there are topics that are going to be shorter and longer but this average at least gives you an idea of what a “topic” might be.
If your conversion is giving you longer and fewer topics, then you have a problem with chunking. You need to either re-write into discrete topics, each with an appropriate title, or re-do your conversion to chunk at the appropriate heading level.
Clean up the content
Converting content that is minimal and matches the DITA architecture already can save you lots of time in post-conversion clean up. Apply minimalism. Remove extra words. Remove sentences that repeat titles or captions. Streamline everything. If you have 40 conditions in your FrameMaker book, it’s time to do a purge.
In general, a clean conversion means that your content maps nicely to the DITA elements without abusing those elements. This means your content already matches the DITA architecture.
This mapping is most evident in tasks, which have a very specific set of elements. For example, if you have a step command followed immediately by a result, it will not convert cleanly.
- Log in as an administrator. The administrator interface displays your dashboard, updated every 30 seconds.
If you convert this as is, you will get the markup:
<step><cmd> Log in as an administrator. The administrator interface displays your dashboard, updated every 30 seconds.</cmd></step>
Simply by adding a line break directly following the command, you can get a clean DITA conversion:
- Log in as an administrator.
The administrator interface displays your dashboard, updated every 30 seconds.
<step><cmd> Log in as an administrator.</cmd>
<stepresult>The administrator interface displays your dashboard, updated every 30 seconds.</stepresult></step>
If you think that’s not a big or important change, then think again. Having the correct markup around the appropriate piece of content is what can distinguish DITA adoptions that succeed and those that fail—fail to provide the agility and power that is possible by having content in XML.
Having the result in a <stepresult> tag means that you can programmatically hide all step results for mobile output, for example. Or you can format that non-essential information differently. It also means that you can possibly re-use this step or topic in another place, even if the step result is different in that other context.
The cleaner, more minimal, and more task-based your source content is, the easier the conversion will be. As an added bonus, you will also end up with better quality DITA content.
Develop a Content Strategy
You should develop some sort of content strategy prior to beginning the bulk of your conversion.
Among other things, a content strategy defines the elements and attributes you will use. This helps inform your conversion.
For an example not at all at random, consider the use of the short description. It’s a powerful little element—but it’s also the one piece of content that is rarely pre-existing prior to a move to DITA. Some companies decide not to use short description elements while others decide to always use them (it’s a bit like a yes or no question). It’s a powerful little element, but mostly valuable in HTML output. If you don’t have an end-to-end strategy in place, you won’t know whether you should add a short description to every topic as part of your conversion.
I can tell you that adding and putting content in an element in every single topic after conversion is a painfully time-consuming job. It’s much faster to do this work programmatically during conversion, even if you have to go back and fill in some content later.
A content strategy helps you define what you need so you can have the DITA building blocks you need in place as a result of your conversion.
Pick a conversion method
You have some options when it comes down to actually performing a conversion from unstructured (or other-structured) content to XML.
- In house: Usually using FrameMaker conversion tables, this is a way for you to completely control your conversion. You won’t benefit from time-saving scripts or built-in best practices that other options can provide you with. Errors and omissions can have a serious impact on your project budget and timelines, not to mention quality. The person doing the conversion ought to have a very good working knowledge of DITA architecture, but they can learn FrameMaker conversion tables on the fly.
- Consultant: A consultant works with you to identify the strengths and weaknesses of your content first so your conversion is of the highest quality. Consultants will also help you implement your content strategy and apply best practices. You won’t need expert knowledge of DITA but you will still have input and control over the results. Consultants can often meet very tight deadlines, when no one on the team has the time to convert large amounts of content.
- Conversion specialist company: These are companies that make a business out of converting content. They can convert custom XML to DITA and convert huge amounts of content in a short time, and have a powerful engine that can be customized to whatever you need. Although budget and pace are usually out of your hands, they get good results even from really difficult conversion projects.
- Stilo’s Migrate: In a class on its own, Stilo’s Migrate self-service format lets you leverage the time-saving tools and functionality (like converting images to SVG on the fly) of experts while still having full control over the pace, cost, and details of your conversion. It’s flexible enough to adapt to what you need and powerful enough to make the process fast and reliable. Remember that implementing best practices is still up to you, so the people using Migrate ought to have a very good knowledge of DITA architecture to ensure quality results. Alternatively, you can bounce your Migrate conversion results off a consultant to make sure you’re heading in the right direction.
The method you choose is going to depend on your timelines, budget, in-house expertise, volume of content, and comfort level. You can also mix and match them, getting help with your tough content and doing the easy ones yourself.
Perform a trial
Whichever conversion method you select, make sure you do a trial run with real content. Like taking a car for a test drive before you buy it, a trial run helps you make modifications to your conversion early on, raise content strategy questions you may not have considered, validate the quality of your pre-converted content, and validate metrics, budget, and timelines. If you need to modify your budget or choose another conversion method, this is your big chance.
There are very few drawbacks to performing a trial conversion—it does add some time to your schedule though, so factor that in.
The best type of content for a trial is the content you’re most confident with but which also shows some complexity in your content (conditions, text insets, index markers, etc.).
Hint: Stilo offers a free trial conversion that you should really take advantage of.
WEBINAR | A Tale of Two Formats; Creating content with Word and DITA XML simultaneously
Presented by Catherine Long & Rich Perry | Varian Medical Systems
You decide to move to publishing with DITA XML. It all appears wonderful. You purchase a CMS, select an editor, create your information model, and share the plan with your authors. This is great! You begin to analyze documents, get a style sheet designed, prepare and schedule author training, but then problems and issues of resistance start adding up. There are so many different document types (60) and so many global authors (100) who are not tech writers. You begin to realize that unless the company can stop needing to provide service and installation for 3 to 6 months, you are not going to be able to flip the switch and move to DITA all at once.
What can you do?
Make the move at your own pace, and publish documents with Word and DITA simultaneously. Doing so provides many benefits, such as bringing the different document types into the project a few at a time and training authors with their own content already in the system.
Catherine Long and Rich Perry show you how this dual publishing helps you move through the process of moving to DITA so that the experience is easier, and perhaps cheaper, for everyone involved.
Meet the presenters
Catherine Long
Catherine Long has been with Varian Medical Systems for five years. She was brought in to assist the service documentation department with authoring standards and the publishing process, as well as to lead the move to DITA XML. Her challenge is to design a system architecture and provide training for 100 SMEs who write documents as one part of their busy schedules. Her relaxation is to immerse herself in the worlds of Shakespeare, Wodehouse, Wilde, and others.
Rich Perry
Rich Perry manages a publication team responsible for processing technical servicing content at Varian Medical Systems. Over the course of his career work life, he has held various technical positions supporting military and medical devices. These experiences as an end-user of servicing procedures lit a fire in him that led to his avocation as a technical trainer, curriculum developer, and product support specialist. Rich’s team is in the process of transitioning from an unstructured Word authoring environment to DITA. While he and his team have not fully implemented DITA, Rich is ready to share his experiences as the struggle continues!
WEBINAR SLIDES | Migration to DITA – The Atmel Story
Presented by Morten Haaker, Senior Project Manager | Atmel Corporation
Are you thinking about moving to DITA? In addition to learning DITA, tool and vendor selection, creation of stylesheets, change management and other challenges, you may be faced with having to convert thousands of pages of detailed technical information from an unstructured format to DITA.
Learning from the experience of others can make a big difference to your project success and the Atmel story may just help you to better determine your migration strategy.
Join us for this Stilo DITA Knowledge Series webinar as we hear from Atmel Corporation’s Morten Haaker who will show how Atmel, a global semiconductor company, approached the issue of content conversion as an integral part of their DITA implementation.
Morten is a Senior Project Manager at Atmel Corporation responsible for the technical documentation project. Located in Norway, Morten has been with Atmel for more than 12 years. He holds a Master of Computer Science from the Norwegian University of Science and Technology and heads up the Nordic SDL LiveContent User Group.
Important: Due to an unforeseen technical problem, we were unable to record this webinar.
Meet the Presenter
Morten Haaker
Morten is a Senior Project Manager at Atmel Corporation responsible for the technical documentation project. Located in Norway, Morten has been with Atmel for more than 12 years. He holds a Master of Computer Science from the Norwegian University of Science and Technology and heads up the Nordic SDL LiveContent User Group.