Rhys Williams, Wuhan Coordinator, Faculty of Business, RMIT University [HREF1], GPO Box 2476V, Melbourne 3001. rhys.williams@rmit.edu.au
Troy Boulton, Business Consultant (RMIT Web System), Information Technology Services, RMIT University [HREF1], GPO Box 2476V, Melbourne 3001. troy.boulton@rmit.edu.au
Izabella Bartosiewicz, Web Manager, RMIT University Library [HREF2], GPO Box 2476V, Melbourne 3001. iza.bartosiewicz@rmit.edu.au
RMIT successfully deployed a new corporate web/content management system approximately 14 months ago. A key feature of the system is a widely distributed publishing system, which a large number of staff are required to use in the publication of core university corporate content.
This paper discusses how RMIT went about designing and implementing this fundamental change in approach to web publishing, how the system works, issues encountered, and how a process of continuous assessment and improvement is being embedded, based on a large usability analysis undertaken at the conclusion of the first iteration of the change process.
The Web Refurbishment Project, conducted during the period May 2000 - December 2002, moved RMIT from a situation of having hundreds of web servers using a variety of manual web publishing practices, with limited controls over standards and QA, to having all core corporate content in a centralised accessibility compliant, searchable, web publishing system, using dynamic publishing techniques with distributed publishing, workflow and QA processes in place.
The milestones of the project were:
1998 Website Review:
A review and risk analysis of the existing website.
1999 RMIT Web Strategy:
A business and technical strategy developed and signed off to meet the recommendation
of the review, and designed to take the RMIT web environment in a sustainable
way into the 21st century.
The following table summarises the then old and proposed systems:
| LEGACY SITE |
NEW SYSTEM | |
| Hosted on multiple servers | University high-capacity servers | |
| Static sites and content | Dynamic content delivery, better meeting audience needs | |
| Ad-hoc, uncontrolled publishing and updating | Distributed, authorised development and publishing | |
| Manual linking and indexing | Automated, "intelligent" linking and indexing | |
| Duplicated development efforts, inconsistent data on site | Coordinated and consistent, validated content across site | |
| Fragmented "look & feel" and navigation | Consistent look & feel and navigation still enabling faculty and group individuality | |
| Uneven access to development and serving technology | Simplified, centrally managed development | |
| Cost ineffective | Improved ROI- From $55/webpage to $15/webpage |
2000 Prototype System:
Centralised prototype build and acceptance, covering core required functionality,
and 3 key representative content areas - one Faculty (Constructed Environment),
one Resources area (People Services) and the top level of the RMIT website.
2001 (May - November):
Production System finalisation, integration
2001 (November) - 2003:
Go Live and University-wide Rollout
The following vision and objectives were the foundation of the project:
Vision
Objectives
At the core of our website lies a content management database called TeraText™ [HREF3] (previously called SIM), developed at RMIT and used as a commercial product . The TeraText™ database:
The publishing system was derived from the system built by the Australian Standards [HREF4]. It uses MS Word® as an authoring environment and a web browser for publishing and cataloguing of documents and objects.
Functionality to support particular content management, publishing processes and features was identified through the requirements analysis and prototype phase, and implemented as part of the custom RMIT Web Publishing System sitting on top of the TeraText™ platform. To this day, functionality is being assessed, revised, added and tweaked as needs emerge and are prioritised.
For overall RMIT site architecture, a number of default paths or sections was determined, specific to each of the publishing groups: Department, School, Centre, Faculty, Resource and Editorial.
In analysing how to build or migrate a site, each group receives a template (as a spreadsheet) with the base paths on top of which the entire, group-specific information architecture (IA) can be built. New, 'permanent' paths can be added to the base path to fulfil the information architecture requirements while additional sub-paths can be added 'on the fly' at any time during the publishing process.
Once the IA is finalised (but not necessarily 'set in concrete'), the migration
area is set up and publishing responsibilities assigned to group members so
that the content migration can proceed. At this stage the content uploaded and
approved is visible to the owner group only.
Certain type of content gets 'automatically' placed in a group and a path generated.
Examples are:
Once the migration process is completed and usability tests conducted, the entire site can be 'switched on' in a matter of minutes.
Look and feel (branding graphics, colours, and potentially even different page layouts) are applied to publishing groups, and implemented automatically for all content belonging to a particular publishing group.
Before the authoring of documents can take place, a custom-designed web publishing template needs to be downloaded and installed in order to activate the publishing menus in MS Word®.
Next the starter documents for the different types of content need to be downloaded. Starter documents are Word documents which can be accessed by all staff; these include: Generic, News Prototype, Services Prototype, Staff Profile, Gallery, FAQ, Homepage, Course Guide and Form. Each starter document contains a metadata table specific to the type of content it is designed to display and indicates a default layout for the content (where applicable) to encourage consistency.
Figure 1: A starter template used for News documents
A combination of custom-built menus and macros, and a selection of commands available in Word® are used for authoring of web documents.
Figure 2: Custom-built menus
Adding links and embedding multimedia objects is managed via clues. Clues can point to document/object titles, to Internet addresses and can also be used to embed a search query, which will locate and list all documents matching the query once the specified document is uploaded to the server. Only the reference to the media object is added to the document, not the actual object.
Figure 3: Clue-based links in Word® document
Both hyperlinks and multimedia links are reconciled upon uploading of the document, through a metadata table. The system prevents linking to published documents or objects unless they have been approved. The advantage of this restriction is that internal links can only be resolved if the document is live, thus preventing this type of broken links from occurring. If the document is not available, a 'defer' option is available that will allow to complete the editing process. Unreconciled links are indicated in red on the page and are difficult to miss.
Figure 4: Reconciling links to internal document using page title (RClueTitle)
Display styles are managed primarily through a global style sheet, which applies to all documents published to the Web, regardless of their type. The style sheet cannot be customized by individual publishers, however can be changed and applied per group, giving opportunity for individualisation and quick mass change (although RMIT has opted to currently apply the same general style across the site.)
This multi-function environment allows for publishing of text documents and media objects (images, movies, pdf, sound, etc.). It controls file uploading/downloading and moving files through the approval cycle.
Once the file is uploaded, the user is presented with a metadata table. The metadata table stores information about the document and controls a wide range of functions. Some of these include:
The options contained in the metadata table are unique to the individual document type. For example, in the Homepage document the page title can be hidden from public view making the design of front pages more flexible, while in the News document, each news item could be associated with an image.
All compulsory fields must be filled out before the file can be successfully uploaded.
Figure 5: Metadata table for a generic document
Media objects get uploaded to the media asset management system (MAMS). This is a central place for storing all kinds of multimedia and learning objects, including images, sound, video, pdf and all other digital media RMIT owns, or has a license to use. The same procedure applies for uploading media objects, as is the case with text documents. Each object needs to be catalogued (creator, copyright owner, intended use, retrieval information, rules of use, keywords, abstract, etc.) before it can be uploaded and approved. It is a time consuming process, however the benefits of cataloguing the individual objects are many:
Another component of the publishing environment is the status table which allows users to move documents between the different statuses. It displays only after a document has been successfully uploaded.
Figure 6: Status table
The publishing system recognises four main types of users:
Creators – can upload documents into the draft status and move
them to the proposed status.
Editors – can edit documents in proposed status and move them to
draft or to pending.
Approvers – can edit documents in the pending status, move them
back into draft or approve them for live access.
Super-user – can edit and move documents regardless of the status
they are in.
Quality assurance was to be enforced through a strict approval process requiring at least two people to be involved in creating and approving each document. In this scenario, creators would not be allowed to approve their own documents. This functionality was switched on for a short period of time and caused a lot of frustration and delays during the content migration. It became clear that the 'business' and 'technical' approval processes would need to be reconsidered and aligned with the 'real world'.
| Since the go-live: | |
| 500 | content areas (publishing groups) |
| 400+ | trained staff |
| 150+ | regular creators, editors, approvers |
| 217,250+ | live items of content published |
| 360+ | course guides authors |
| 18,284+ | course guides published |
| 250,000 | page hits per day (average) |
| 382,000 | peak page hits (in a single day) |
| 4 | load-balanced public browse servers |
| 12,361 | staff user accounts |
| 77,858 | student user accounts |
| 6 | support calls (average) daily |
| 6 | training courses (introductory to advanced, plus specialised such as Writing for the web and course guide editing) |
In the second half of 2002, RMIT undertook a usability study of the new website encompassing end-user acceptance, search functionality, quality assurance processes, W3C technical compliance testing and competitor benchmarking. An important component of this study was the evaluation of the web publishing system. It formed the first step leading toward major improvements to the beta release of the system, scheduled for 2003.
A heuristic evaluation was chosen as a method for conducting the evaluation. It allowed us to involve local, expert users of the system and adopt a number of existing and well-documented usability principles to guide the evaluators through the process. [HREF5], [HREF6]
The outcome of the evaluation was a list of concerns - they were documented, ranked according to severity and prioritised accordingly. The range of problems identified varied from system bugs to issues relating to complex interfaces and multiple steps required for moving documents through the publishing cycle. Although the custom-built web publishing system was found to be usable and technically robust for the initial phase of prototype build and migration the evaluation highlighted the fact that the system would require a range of refinements and improvements before it could be truly flexible and easy to use by anyone in RMIT community.
Some key areas of improvement included:
1) Simplifying MS Word® -based authoring and editing of content
MS Word® required a number of customisations in order to expand its functionality as a web authoring environment. Some users (new and occasional users in particular) found it confusing which native Word functions they could use and when they had to use the custom-built functions specific to the web publishing system. Future developments will need to be directed towards utilising the existing capabilities of MS Word® more effectively rather than adding to its already complex interface.
2) Adding flexibility for diverse users
The intended simplicity of the web publishing system is likely to overcome
the skill barriers of people who are inexperienced at web publishing, however
it should not come at a cost of restricting the experienced web publishers.
The evaluators suggested exploring the alternatives to MS Word® for content
authoring and editing. Supporting a range of authoring applications has many
benefits:
3) Uncluttering of publishing interfaces
A major disadvantage of constructing the publishing system in an 'organic' manner is the unavoidable increase in complexity of interfaces as the new functionality is being added and modified. Redesigning the interfaces that clearly present the options and guide users through the publishing steps will be undertaken with input from users across the entire spectrum of web publishing expertise.
4) Aligning real life business approval processes with technical functionality
Key publishing business rules, especially approval processes, require simplification, and better alignment with 'real world' business practices. In addition, greater rigor and discipline needs to be applied to the enforcement of RMIT publishing standards and style guide.
If you can get it right you win one convert; if you get it wrong you lose a potential convert who talks to on average four mates…
Communicate clearly and frequently - have targeted communication strategies for different audiences.
OO methodologies are not just important to programmers. In a complex project, like building a corporate web system, issues surrounding content, visual design, human-computer interaction and technical development abound. Develop strategies to ensure disparate issues such as visual and technical design are separated and not dependent on each other, but still integrated into a cohesive, appropriate outcome.
Constructing the publishing system and developing a website prototype by engaging representative groups from the University as 'pioneers' requires careful planning, flexibility to make changes on the fly, diplomacy and excellent management skills - from all parties involved! Make sure everyone agrees to what you are trying to achieve.
When embarking on an IT project, many stakeholders come in with the expectation that defining business and functional requirements, then sending the developers off to implement the vision will lead to an positive outcome. Business Processes are the fundamental building blocks of any system, and until these are understood and aligned with desired changes, no amount of technology, developers or graphic designers will build a better system.
To support essential shifts in business process, it is imperative that rock-solid support is gained from key stakeholders, and that strong governance and support from senior management is in place to drive these changes.
Enterprise portal alignment in the RMIT landscape was new and not fully understood. Stakeholders were quick to see benefits, but were often unclear on their roles, the work required and the time it would take to reach an optimal solution.
The RMIT web project ended up with a technical solution that broadly met initial high-level requirements, however took a year longer than envisaged at initiation, and identified a raft of new additional issues along the way. Key management issues which have arisen include:
IT is a complex beast. The web mixes this with complex people. Clear business and technical analysis needs to occur before a developer's keyboard sees one key-press. 'Gotchas' include:
Change management is an art, as well as a science. Issues have included:
The way around this is to not only communicate out, but to ensure the stakeholders are heard and their needs integrated into the analysis, planning and implementation of any change they might be exposed to.
Critical determinants in medium term system improvement will be the effectiveness of systems integration at RMIT, and delivery of critical usability improvements identified in the recent usability study.
The key value of the web system is as the authorised repository for, and publisher of, most core corporate and teaching and learning content. The value will grow through more comprehensive data and content integration with other systems, especially:
As noted before, one of the key objectives of the project was to fairly radically change the business processes involved in publishing to the RMIT website. Large universities are financially squeezed organisations, which typically cannot afford to dedicate the resources private sector organizations do to their web and content management systems. However, unlike many corporates, the core business of a university is knowledge creation and transfer, which has rapidly moved into the online domain. In addition, there are far more content creators and owners in a university who are widely distributed through the organization from very senior to relatively junior levels.
Consequently a key project objective has been to develop a simple, robust, scalable publishing system, requiring of staff very few technical skills, and is easy to administer and provide flexible training around.
This can be technically described as a move from multiple linked web systems, each with its own technical gate keeper called a web master, to one system with a unified, simple publishing system based on MS Word®, requiring very simple technical skills.
Thus the skills now required in a new website manager, as opposed to a traditional web master, are online content and information management skills. Content creators and editors can publish more directly to the web, however require new skills in writing for an online medium, correctly cataloguing their content, and managing their content to fit into a pre-determined architecture.
The take up of the publishing system has been very encouraging. However, the key challenge into the future will be improving the quality and searchability of content through cost effective supports to enhance the new skills of content creators and managers. In this regard, improved publishing interfaces and intuitiveness, are critical.
RMIT is currently undergoing significant changes in structure, governance and strategic focus. Maintaining such a system through this period of change will require renewed commitment and energy from its stakeholders, and a clear determination to learn from the lessons dealt in the last 3 years. The true test of this initiative will be to see what the University is doing with its website in 3 years time.
Kannegieter, T., Sefton, P. (2000). "Content Management for all of us: The Standards Australia Experience" in Proceedings of Ausweb2K, pp. 488-489. Available online [HREF4]
Nielsen, Jakob. (1999). Ten Usability Heuristics. Available online [HREF5]
Tognazzini, Bruce. (1998). First Principles. Available online [HREF6]