exact  any/all
  The original knowledge-management publication
denotes premium content | Apr 26 2015 


posted 3 Aug 2005 in Volume 8 Issue 10

Finding a balance: Global and community-specific taxonomies

Making the move to hybrid taxonomies at Ernst & Young. By Rebecca McDonald, global KWeb taxonomy program manager, Ernst & Young.

Many organisations are currently grappling with making decisions about taxonomies – determining if they are needed, what they should be used for, why they are important and how to create, implement and maintain them. The members of the Ernst & Young global professional services organisation decided in 1994 that an important component of our knowledge strategy was to create, implement and promote a global taxonomy to enhance internal knowledge navigation and document-level knowledge identification. The decision has been a good one – the global taxonomy continues to play a very important role in bringing together content via Ernst & Young’s search and browse capabilities. Additionally, we are more frequently using taxonomy to enable profiling and identify focused content for portal applications. This case study will highlight some of what Ernst & Young has learnt about the global taxonomy and will provide an overview of what is being done to achieve the appropriate balance between the global taxonomy and community-specific taxonomies.

Setting the scene

Before launching into taxonomic details, it is important to understand who uses the global taxonomy and the environment in which it resides. The 100,000 people of Ernst & Young’s member firms pursue the highest levels of integrity, quality and professionalism to provide clients with a broad array of services relating to tax, audit and risk-related services and transactions in 140 countries around the globe. In order to provide the best possible work product to clients, Ernst & Young professionals need to be able to quickly isolate and access knowledge at the time and place they need it, whatever their geographic location. Because of this, it is important that knowledge is available globally.

Early on, Ernst & Young developed a standard set of databases, portals and submission processes to stimulate the creation and storage of internal and external knowledge and to enable access to this knowledge. Each tool has a different purpose – to facilitate discussions, store critical knowledge about an industry or subject, or act as a single access point or central hub of knowledge for members of particular communities of interest. Many of these tools are community-focused knowledge bases, intended to be used by a single, defined community. Most of the larger knowledge bases, often managed centrally but owned by a community, are key global knowledge bases. In total there are now over 2,000 global knowledge resources available, 15 per cent of which have a globally focused audience. A central tenet to Ernst & Young’s knowledge sharing is that ownership of content is decentralised, so each knowledge base has an identifiable owner who is responsible for its content and maintenance.

The Ernst & Young intranet, known as the KnowledgeWeb (KWeb), is a system of tools and applications, including a search engine, guided navigation and a global catalog, which facilitate access to knowledge content. From a taxonomy perspective, each global catalog record is tagged with terms from the global navigation taxonomy (GNT), or a shallow taxonomy describing the content of each of the knowledge bases in the KWeb. The taxonomy dictates where and when a knowledge base is surfaced in the browse features of the KWeb and is utilised in the guided search and taxonomy-based features of the KWeb search engine.

The global taxonomy does not stop at the knowledge base level. It continues to decompose into what is known as the global document taxonomy (GDT), which is implemented by key global knowledge bases and is applied to each document in the knowledge base. This tagging at the document level allows our search engine users to refine their searches down to the document level, which greatly improves the focus of search results. As of 2003, over 125,000 documents in ten key global knowledge bases have been tagged with the GDT. Of the other 1,990 knowledge bases in the global catalog, a large number of these have also used some form of the global taxonomy. These numbers indicate that the global taxonomy is the soft glue that is holding Ernst & Young’s knowledge content together.

Global taxonomy assessment

In 2003, after seven years in use and organisational changes within Ernst & Young, the KWeb taxonomy group responsible for the global taxonomy and its evolution began an assessment of the global taxonomy to determine if it was still adequately describing the knowledge content of Ernst & Young. The assessment focused on the effectiveness in implementation, management processes, and the place of the global taxonomy in the KWeb. The findings and recommendations are forming the current direction of the KWeb taxonomy programme.

Global taxonomy findings

Overall, the results of the assessment proved that the initial concepts, design and structure of the global taxonomy are very solid. Usage and the trend to using the taxonomy for behind-the-scenes content aggregation indicate the global taxonomy is integral to Ernst & Young’s system of knowledge organisation and retrieval. Based on these findings we should encourage new ways of using it. The assessment also identified areas for modification.

The recommendations upheld the belief that those who best understand the subject matter of a knowledge base should be responsible for applying the taxonomy, and that it is ultimately the responsibility of the knowledgebase owner to monitor whether the taxonomy is applied accurately. Additionally, the assessment pointed out the need for more education about the global taxonomy, particularly to ensure that it is better understood and accurately applied to content. The assessment also verified that not all knowledge bases using the global taxonomy categories were defining the categories in the same way, resulting in search results with different types of content associated with the same term (eg, industry of the client vs. industry of the intellectual content of a submission).

Ernst & Young’s KWeb taxonomy team has benefited from the fact that taxonomy has become institutionalised across the firm. In 2003, functional groups joined together to standardise key taxonomic reference data points globally. Many of the data reference points that this group has standardised are components of the global taxonomy. (The group validated that we had identified the right categories and has helped promote the adoption of those categories.) Since then, knowledge communities, perhaps because of the structure imparted by this organisation-wide push to use a common vocabulary, seem to be more accepting of the fact that there cannot be variations in terminology. This has significantly reduced the debate or discussions that occur when taxonomy development is underway.

The assessment also identified taxonomy maintenance as an improvement need. In 2003, the global taxonomy was associated with over 125,000 documents and 2,000 knowledge bases. With every change to the global taxonomy, large numbers of documents must be converted to keep content accessible via KWeb. Due to the huge impact on content, taxonomy modifications were often resisted because of their complexity and cost. The delay in modifying taxonomy only exacerbated the problem because when a conversion finally took place, what might have been only a few taxonomy terms usually developed into a major conversion effort.

Partially due to the inclusion of global taxonomy elements outside of the knowledge realm, Ernst & Young more recently has been able to establish change frequencies for these taxonomy categories. Changes to the taxonomies are announced twice a year. This timetable allows the KWeb taxonomy team to establish a regular review timetable for all elements in the global taxonomy. The team is able to discuss taxonomy changes proactively with the communities that are using the global taxonomy. These discussions seem to have helped knowledge-base owners understand that they need to be more disciplined in maintaining their taxonomies and ultimately their knowledge content. We now conduct taxonomy maintenance as a part of ongoing knowledge base operations, and this is resulting in fewer large-scale taxonomy conversion projects.

The last major finding that emerged from the assessment was the fact that the global taxonomy needed to be simplified with respect to both depth and coverage. Application statistics indicated that of the taxonomy terms applied to content, over 90 per cent of the terms went no deeper than three levels in the hierarchy, whereas the global taxonomy at that time went to five levels in some instances. Since the assessment, we’ve established a rule that the global taxonomy is a maximum of three levels deep: three for documents and two for knowledge navigation.

The taxonomy assessment also revealed that two of the eight original GDT categories were no longer being used by any global knowledge community. Both of these access points had been of primary importance to a service line that no longer existed within Ernst & Young. As content requirements change, it is reasonable that the global taxonomy would require change. The assessment reminded us that care must be taken to ensure that taxonomy categories in a global taxonomy are of universal importance to users. As we saw at Ernst & Young, if a taxonomy category is not relevant to those adopting the taxonomy, it will not be used.

The assessment also indicated that the inclusion of another category was limited and that there had been problems trying to standardise the taxonomy across global communities. Upon reflection, certain elements of the original global taxonomy proved to be categories in which we were unable to enforce global consistency and universal usage. As such, since 2003, Ernst & Young has not tried to globally enforce the use of these taxonomy categories. Each taxonomy category is still used and is of extreme value to particular communities, but they are no longer considered global taxonomy elements. These types of access points are now being considered community-specific taxonomy elements.

Hybrid taxonomies

Post-2003, key global communities have created hybrid taxonomies to describe their content. A hybrid taxonomy results when a particular community adopts only those global taxonomy categories that fit its needs and then supplements the taxonomy by developing community-specific categories, for example Subject/Topic. This model of balancing the global taxonomy and community-specific taxonomy categories ensures that the content of a particular community can be linked to universal content, as well as identified meaningfully within the focused audience of the community.

The move to hybrid taxonomies has placed an even greater responsibility on our communities for taxonomy discipline in development and maintenance. Many of these communities inaccurately believe that different rules of discipline apply to community-specific taxonomies. The truth is that communities have more flexibility in the terms they choose to use and how often they want to update the taxonomy, but they must adhere to the same rigor required by the global taxonomy for structure and changes. It is the responsibility of the KWeb taxonomy team to effectively communicate the dangers of poor taxonomic structure and to help teach communities how to make the right decisions about taxonomy in advance of it being applied to content. Finding out the taxonomy doesn’t work post-deployment is often costly in terms of human and technical costs, and it also can undermine the credibility of the knowledge base.

In addition, the maintenance processes associated with hybrid taxonomies are more complex because of the fundamental differences in the requirements of each of these types of taxonomy categories. We anticipate that managing the education around and maintenance of these hybrid taxonomies will be one of the most difficult issues that the KWeb taxonomy team will have to solve in the near future.

Community focused uptake of the global taxonomy

When the global taxonomy was first released, the thought was that controlled vocabulary requirements for community-specific knowledge bases were quite different from those of global knowledge bases. This belief, coupled with business decisions, meant that very little effort was made to promote the use of the global taxonomy terms in community-focused knowledge bases. In the past, the GNT had been the primary glue bringing together similar types of databases. However, the launch of a KWeb-focused portal that allows the aggregation of content from various KWeb tools and resources has led to communities identifying specific documents from knowledge bases that are of importance to them. More frequently the GDT is being used to bring together this type of knowledge. This post-coordination of content is driving a move toward the adoption of the GDT in more community-level knowledge bases. As a result, the KWeb taxonomy team must create materials that will help the owners of the community-focused knowledge bases determine when or when not to use the global taxonomy in those applications.

When communities determine they want to use the global taxonomy in their community-focused knowledge bases, they tend to customise it in such a way that it suits the community’s content. The types of customisation most commonly seen are: i) communities using only select branches or nodes of the taxonomy, and omitting non-relevant terms; ii) communities limiting their taxonomy to only the highest level terms when the taxonomy is being used for navigation or if there is not much content; and iii) some communities require more specific terms than are available in the global taxonomy, so they start their taxonomy with the most specific global taxonomy terms available.

A final trend with community usage of the global taxonomy is that of localising the taxonomy either by translating it or using aliases to improve the currency of the taxonomy to a specific user group.

There are two key issues to be resolved with this type of customisation. The first is how to notify such a diverse user base of modifications to the global taxonomy when they occur. The second is how best to educate the owners on determining how, and when, to modify taxonomy based on the impact to their content.

Where next?

Based on current trends, we envision that usage of the global taxonomy will continue to grow and that, as more focused communities start to use the taxonomy, more flexibility with respect to localisation of the taxonomy is expected. As the taxonomy is used and customised further, the KWeb taxonomy team will continue to deliver appropriate education materials and ensure that the maintenance tools and processes are in place to keep it all together.

Follow us on:

Copyright ©2015 Wilmington Publishing & Information Ltd 2010, a division of the Wilmington Group PLC. Wilmington Publishing & Information Ltd is a company registered in England & Wales with company number 03368442 GB. Registered office: 19 - 21 Christopher Street, London EC2A 2BS. VAT NO.GB 899 3725 51