Hypermedia Version Support for the Online Design Journal


Sandy Kydd, Alan Dyke, and David Jenkins

Department of Computing & Information Systems
University of Paisley
High Street, Paisley PA1 2BE, Scotland
Email: {sandy.kydd, alan.dyke, dgj}@cs.paisley.ac.uk

1. Introduction

This position paper introduces the requirements for version support in a hypermedia system which seeks to support the work of designers in the early stages of design. The Online Design Journal (ODJ) is a concept which is being developed at the University of Paisley, and will be a used in two funded projects - Safety Critical Integrated Design Support (SCIDS), funded under the DTI-SERC Safety Critical Systems programme, and Designers Using Cooperative Knowledge (DUCK), funded under the DTI-SERC CSCW programme. The ODJ uses the metaphor of a designer's journal or daybook to gain access to sketch-based information in the early stages of a design. The ODJ makes use of hypermedia technology to link designs and design concepts in the journal, providing a powerful tool for cross-referencing design information. The belief is that by capturing design history, and providing online access to current and past design information, designers will be able to reuse design ideas, and check on the progress of current design work by other members of a design team. The creation of the design history is also considered to be the creation of the hypermedia network which supports it in the ODJ, so issues of design reuse, alternative designs, and the preservation of historical design information are paramount for the ODJ.

There are a number of issues of version support which need to be addressed in the design of the ODJ. The reuse of part or all of past designs implies that the reused material is common to more than one design, but the issue of whether the common design information should be copied to the new design or the new design regarded as a new version of the original design needs to be investigated. Similarly, the effect of having concurrent alternative designs should be covered. This can become complicated when there are many versions of partial design - the tree structure involved in normal version control is complicated by the possible versioning of hypermedia links. Finally, there are cases where design information on supposedly finished pages needs to be amended. These alterations could involve copying of the previous page, or there could be support for an amended version of the original page.

2. Design History Capture

One of the main problems for design researchers is that of capturing data about the history of a design, especially in the early stages of design. Designers tend not to start using computer tools to record their design information until the crucial initial decisions have been made. These are usually done by using the traditional paper and pencil method of annotated sketches, which are recorded in design journals. In strict environments, these journals can be used for purposes ranging from auditing to formal recording of design modifications. However, the fact that the information is not entered into computer records of the design causes great problems when checking on the history of previous designs, and trying to provide traceability of design intentions from the completed computer generated design drawing back to the early design decisions in the annotated sketches. The design history in the computer is incomplete as long as the early design information is kept in a paper-based form.

An aim of many design researchers is to provide designers with a tool which will allow the capture of information from all stages of design, so that they can study the way that designers work. Such tools have been built [1, 5], but many of them impose some specific design method on users, preventing the design process from being captured in its natural form. Designers using these tools will end up designing on paper using their own method, and then translating the design information into the tool's method for input into the computer. In this way valuable design information will be lost. What is needed is a tool which will capture design information in a method-free manner, that is, it is not based on assumptions about how the designers work [4].

3. The Online Design Journal

The Online Design Journal (ODJ) is a concept for the use of a shared design journal which will be kept in electronic form. The ODJ is specifically designed to capture information in the early stages of a design, which is why it uses the metaphor of the familiar design journal. Designs are sketched and annotated on conceptual pages, and the annotation labels used to implicitly link designs in the journal through the use of hypermedia techniques. The ODJ does not impose any particular design method on the designer - it just captures and stores the information which the designer inputs. As well as storing the information which the designer creates on a page, the ODJ also stores the actions which the designer took to create that information, including the order of creation of sketches and labels, and any changes, deletions, etc. - in effect it stores the design history [3].

The information on pages in the virtual design journal is linked in two main ways. The first method uses hypermedia links as the headers for a page, to give the page context in the virtual journals it may be included in (there may be several virtual journals in the ODJ database, and a page may be in more than one journal). Hypermedia links are also used to link pages which have references to the same labels. Labels are text annotations which the user makes on a page. The hypermedia links are implicitly created by the ODJ to link annotations with the same label text within the currently defined scope of the page.

The fundamental metaphor of a design journal gives the users the impression of a linear structure to the pages in the ODJ, as opposed to the hierarchical and network structures which are seen in many other hypermedia systems. This should initially make ODJ journals more navigable from the user perspective, but it may prove detrimental for the grouping of similar concepts and design elements unless pages can be explicitly shifted around in the virtual journal. The linear structure may also pose problems in issues of design traceability, design reuse, and alternative designs. The effect of versioning for the ODJ also needs to be considered - since the aim is to capture design history, the importance of version control for maintaining and manipulating the design history captured in the hypermedia network is fundamental [2]. Version support in the hypermedia system of the ODJ may provide solutions to these problems, but the consequences of where versioning is used need to be considered.

4. Version Support for Design Reuse

One of the main aims of providing an online design journal is to provide easier access to past designs in order to promote design reuse. Design reuse implies the reuse of part or all of a previous design in a new design. Within the metaphor of the design journal, this would involve copying the old pages onto new pages to be used by the new design, with whatever modifications were necessary to incorporate it. However, a simple copying of information would lose any traceability in linking to the original design. There should therefore be some way of linking the new design to the old. However, the copying process is wasteful, and does not take advantage of the power of hypermedia. It also does not provide a specific way of showing that the new design is a reuse of the old design, thus hindering design traceability. A better solution would be to have some form of versioning where the new design was considered as a new version of the old design, which is what it really is if the design is being reused. The difficulty here is to incorporate the versioning mechanisms for the design pages, which are essentially a tree structure, into the more linear book metaphor of the design journal.

Design reuse means using part of a previous design in a new design. If pages aren't going to be changed, they could be explicitly copied onto new pages (and links to the original pages added) to retain the linear book metaphor. However, this might hinder the design traceability. Alternatively the pages could just be referenced via links, but this would abandon the linear book metaphor in favour of a more confusing network structure. A third option which should be considered would be to use versioning in some way to get the best of both worlds - retaining the linear book metaphor, but creating an alternative design version for the new design.

5. Version Support for Design Alteration

Since one of the main aims of the ODJ is to provide design history capture, finished pages in the design journal are considered to be unalterable. It has not yet been determined at what point a page is finished, although it has been suggested this should be when a designer creates a new page in the journal. In this case, the journal metaphor breaks down, as in paper-based journals there is nothing to stop the designer from returning to a previous page and making alterations to a design. If the ODJ prevented this, data would have to be copied from the old page to a new page, and alterations made on this page, creating a link back to the original page. This keeps pages in a linear sequence, but the point of the linear sequence is lost in that we need to do a hypertext jump to get to other pages of relevance (the original unaltered version).

For the purposes of design history capture, it might not be necessary to prevent a designer from doing this, but to record that this was what had happened. In this case, the design history would show a new version of the page, rather than preventing its alteration. How this versioning would be accomplished if the design changes involved the editing of hypertext labels is still to be decided.

6. Version Support for Alternative Designs

Another area where version support may be required is that of alternative designs. There may be situations where a designer may wish to depict alternative design paths branching from a single design page. This is similar to the reuse of designs, but not the same, in that all of the alternatives may not necessarily be used in a final design scenario. The linear book model could cope with this, in that it would be possible to have the alternatives, but without any way of grouping them in a hierarchical manner. The labelling facilities could be used to reference alternative designs from a common starting page, but this would require a work around on the part of the user. They would have to explicitly create a starting page for the design alternative, and use labels to reference these. If they did not know initially that alternative designs would be produced, the starting page for the alternatives would have to be inserted after the first design had been created.

Use of versioning would allow alternative designs to be started from any page, with the only explicit action being the notification that a new version was being created.

7. Conclusions

The Online Design Journal is a promising concept for the support of design team work in the early stages of design. Its use of hypermedia techniques means that it can provide more design concept linking and traceability than traditional paper-based journals, while still retaining the familiar metaphor. However, its need to capture design history raises many issues on how it should support the reuse and alteration of existing designs. One promising solution to this problem is to support the existence of design versions within the hypermedia network. This has implications for the way the ODJ will be built and used, and there are still a number of unanswered questions on what the effects of version support will be.

8. References