Visual editing of semi-structured documents on the web

Visual editing of semi-structured documents on the web
Outputting the same data with different designs is already a reality. So why not add one more design for the editing interface.

Abstract: Exploring patterns and approaches

XML editing for non-trained (novice) users has been a surprisingly hard problem to tackle. Through the twenty years of the existence of XML, no definite way has emerged. This is even more surprising as most editorial workflows are based on XML derivates. Allowing a wide range of people to edit the content in these systems should be seen as one of their fundamental tasks.

As most editorial systems are web-based today, editing should ideally happen directly in the browser. Some products and initiatives have launched in recent years to provide more friendly XML editing tools in the browser (editors). Our research explores the benefits and weaknesses of the different approaches employed by these editors.

For this, novice friendly has been defined to have two main properties: The edit- ing interface should first of all communicate the content model to the user. This means the user should understand the meaning of elements and their structure just by looking at and using the editor. Secondly, the editor should ensure the integrity of the content with regard to the predefined model. This usually means that the produced XML-Data is valid.

Novice friendly also means that the user should not be expected to manipulate the underlying XML directly. The editor should employ some kind of abstraction to communicate meaning, structure and relationships of elements in a way that is more familiar to untrained users.

Two basic paradigms for data-entry could be determined: The form-based approach for highly structured data and the document-centered approach for semi-structured data.

All of the editors are based on one of these two paradigms. They add new behaviors on top of the basic paradigm to compensate for its weaknesses. Styling forms to create the impression of a document or embedding sample content into empty elements are just two examples for this.

The research has also shown, that the comprehensible representation of a content model is always a serious design challenge. It is therefore essential that the presentation of elements is easily customizable.

Finally, we have found areas, like mixed content treatment, where unresolved issues still exist and which could be the topic of future research.


+ Aucun commentaire