The Purpose Specific User Interface
Purpose specific user interfaces are the solution to the biggest challenge confronting structured content: creating it. Let me explain why.
We all agree that the benefits of having structured content are huge. Once you have structured content, you can translate or re-use your content or you can publish it to various online and printed channels or you can publish your online content in a screenreader friendly format. Structured content allows you to easily enforce your company style guide but also allows you to change it when your boss decides he has another favorite color. Structured content is easy to manage and process.
The structured information paradox
Creating structured content is a different story altogether. It is impossible for an automated process to add structure to content since the structure itself is information. For example, an automated process can’t tell the difference between a picture that is part of the information and a logo that is part of the style of a page.
The only reliable way to create structured information that adds value over unstructured information is to capture it at the source: directly from the information worker. However, even though your typical information worker is not interested in technology, the hierarchical nature of structured information usually dictates a technical user interface.
In practice we see four approaches to overcome this paradox. Let’s look at the example of Jane, who writes educational content for a large publisher Edupub. Edupub is planning to move to structured content in order to be able to offer online courses and personalized school books. So Jane now has to start writing structured content.
1. Jane uses a spiced up word processor
This seems a sensible idea: Jane is used to using a word processor and most modern word processors have some form of XML support. But XML does not equal structured information. The fact that a word processor can save its internal model to XML does not bring Edupub any of the benefits mentioned above. There are many efforts, usually based on templates and applying style to define structure, to restrict the word processor to produce structured content. But since word processors were never intended for this purpose, there is a limit to the amount of structure Edupub can achieve using this approach. It is impossible to completely restrict the user interface of the word processor to the bounds of Edupub’s structured content.
Since word processors are not structure-oriented, these solutions work by converting predefined styles to structure. This means that Jane can lose some of her work without prior warning, when she accidentally creates content that only looks like such a predefined style. For instance, she may be entering a multiple choice question. The answers are displayed in the style “MultipleChoiceAnswer”, which looks like bold text. When she now pastes some bold text from another application, the result may look exactly like the other answers. But the style is not “MultipleChoiceAnswer”, so the conversion will not pick it up. This loss of content may even go unnoticed completely. Edupub may attain 95% reliability, but they will never know for sure that it is reliable 100% of the time.
2. Jane uses a form to enter her content
This works fine for strictly data oriented information. But in the case of document oriented information, this approach can result in serious acceptance problems as Jane takes pride from writing professional content and she doesn’t like to see her trade reduced to filling out forms. Even complex hierarchical forms can’t capture the flowing text of the education content.
3. Jane uses a tuned down XML editor originally designed for developers
Most modern XML editors have a WYSIWYG view which hides the XML structure, so they are better geared for flowing text. But this solution has two other problems. The first is the user interface which is usually filled with technical terms like “xinclude” and “well-formedness”. These terms only confuse Jane. Luckily most editors can be configured to hide all developer features. The second problem can’t be hidden. It is the fact that these editors validate the content after it has been created. This means that Jane can step outside the predefined bounds of the structured content. This is great for developers who understand those rules, but for Jane it means that she will have to cope with technical error messages when she tries to save her work. Even worse: she will have to troubleshoot. This is exactly the technology Jane didn’t sign up for and will have a hard time fixing the problems. She will probably just close the editor and start over. A way around this is relaxing the bounds, but fewer bounds means less structure, which again means fewer benefits. In the end there may not be enough structure in the content to drive Edupub’s online courses.
4. Jane uses Xopus
Xopus is an xml editor, so its approach is close to the previous approach but with a significant difference: Xopus was designed specifically to be used by non-technical people like Jane. In the first place this means that the Xopus interface does not have developer features and uses no confusing terminology. More importantly Xopus applies the rules of the structured content before it can be created. This means that Xopus keeps Jane exactly within the bounds of the predefined structure. This is what we call pre-validation. As a result Jane can’t break out of the structure framework and she will be more inclined to use the whole interface instead of just the features she has used before. This will result in more structure and more benefits.
It’s the interface…
We see that using Xopus works great in Jane’s case, but since the Xopus interface is controlled by the structural rules dictated by Edupub’s requirements, she still has to learn those rules. That is no problem if the structural rules closely match the concepts and relations that Jane has in her job. Then Jane can understand that the structure consists of a “Course” that can contain one or more “Tests” that can contain one or more “Questions” that can contain two or more “MultipleChoiceAnswers”. She can add a Test to a Course and add Questions to that Test. That is roughly what she would do if she were using paper and pen.
But in reality Edupub has over 200 writers who all write content for the new online courses. The structural rules have to be suitable for many levels of education and subjects, each with their own methods of testing and question types. And questions have to be re-used between several tests. This means that a “Course” can contain “Tests” and different types of questions like “MultipleChoiceQuestion” and “FillTheGapQuestion” each with their own way of describing their answers. A “Test” now contains references to “Questions”. This means that Jane can no longer simply add a “Question” to a “Test”, but she will have to pick a kind of question, add that to “Course”, go back to the “Test”, add a “QuestionReference” and let that point to the “MultipleChoiceQuestion” she just created. As complexity increases, creating structured information becomes a technical process. This problem is caused by the fact that all 200 writers are using the same generic interface.
We need a purpose!
In the real world, the tools we use every day have a specific purpose. To write on a post-it, we grab a pen or pencil. To write on a whiteboard, we use a marker, but not a permanent one! In the same line of thought, there are countless types of screwdrivers and glasses. A permanent marker does quite well as generic writing tool, but in the end most tools work best when used for the purpose they were designed for.
But the Information Technology world is different. The tools most of us use on our computers are generic. On a typical day we only use a set of applications that can do a multitude of tasks. We use our word processor to write offers, contracts, documentation, reports, random thoughts, recipes, emails and many more types of documents. We have to learn how to use these generic tools to do all our specific work instead of picking the right tool for the job at hand.
The Purpose Specific User Interface
What if we can give Jane an interface with one purpose: creating Biology courses for the 8th grade? This is what we call a “Purpose Specific User Interface”. The limited scope of this interface makes it easy to help Jane edit her courses. We may know that 8th grade Biology tests only have multiple choice questions, so we can hide all other types of questions. The interface could automatically insert a question when a new test is created as there is only one kind of question. The result is an interface that is both easier to use than Jane’s old word processor and that produces exactly the structured content Edupub requires!
Creating dozens of purpose specific interfaces for 200 writers seems like a daunting task, but that is where Xopus comes in. Xopus takes care of the structured content. Xopus doesn’t allow developers customizing it to create content that violates the predefined rules. This allows them to focus on creating the interface without putting too much time in the functionality. The next version of Xopus will ship with an extremely customizable user interface which will make purpose oriented user interfaces even easier to create and maintain.
So a Purpose Specific User Interface takes a little extra development effort but is easier to use than a word processor and produces better structured content than other solutions. To learn more about Purpose Specific User Interfaces, come to see us at CMF2007 in Aarhus or keep an eye on this blog.
[This article has been edited slightly on 21-01-2008]
- Xopus Blog /
- The Purpose Specific User Interface
Write a comment