Published in TDAN.com January 2001
Years ago I began using a definition for meta data that others have told me hits the nail on the head. That definition stated that “meta data is information that improves both business and
technical understanding of data and data-related processes.” Meta data holds the key to understanding and using data. When it is available, meta data enables end users to understand the data
and make better decisions based on this understanding. Anyone who has developed a successful business intelligence database (data warehouse or data mart) recognizes this.
Although the definition above still holds water, it needs to be expanded when the complete data to information to knowledge continuum is considered. Meta data also includes information about
knowledge and knowledge-related processes – and knowledge doesn’t always begin with data. Meta data will help a company to manage and understand its knowledge assets and will enable
that same company to leverage (or make use of) their knowledge assets.
Let us look at a simple example …
Company ABC has 10 stores in a city near you. To expand its presence and revenue stream, that same company acquires a company that has ten competing stores in the same general area. The original
ABC has a specific process for handling customer complaints that has raised customer satisfaction significantly and has increased the number of repeat customers. This process is documented in
memos, store policies, and a video that was created by customer service to be viewed and followed by individuals in all stores. This process is very valuable to Company ABC and is considered to be
one of its means of creating competitive advantage. This process is a valuable piece of knowledge.
What information is known about this knowledge? What information has to be known about the knowledge in order for it to be leveraged across the company? The information that needs to be known about
the knowledge, in this case the process of handling customer complaints, is meta data.
This article focuses at a high level on the role that meta data plays in knowledge management and how the management of this meta data will enable the company to locate and make use of it’s
knowledge. This article briefly describes three types of meta data associated with knowledge: stewardship meta data, business meta data, and artifact meta data.
Knowledge Stewardship Meta Data
In the case of the knowledge (customer complaint process) described above, someone was responsible for creating the memos, creating the policies, and creating the video. This may have been one
person, more than one person working individually, or it may have been a collaborative effort. In any case, someone recorded the knowledge. In many cases, someone was responsible for making certain
that the process was documented well. Someone was (or will be) responsible for keeping the documentation about the process up to date. Someone was responsible for approving the process and giving
permission for it to become the standard way of handling customer complaints. All of this information that focuses on the people associated with the knowledge is stewardship meta data – or
data about who is accountable for the knowledge.
When knowledge is created with it’s “next use” in mind, it becomes very important to document the people associated with the knowledge to make certain that the information is kept
accurate. Without this information, there is no way to manage accountability, and thus it becomes very difficult be certain that the knowledge assets are being managed by the correct people.
Once the knowledge stewards are assigned and documented, this information can be used in conjunction with a date in which the knowledge should be reviewed, whether that review takes place monthly,
quarterly, annually, or less often. With these two basic pieces of meta data (steward and review date), a company can begin to build a process to keep knowledge current by sending reminders to the
stewards that they are responsible for reviewing and re-approving the specific cases of enterprise knowledge.
Just like data and information (probably more so with knowledge), knowledge needs to be managed as a shared corporate asset. Shared company assets require that someone is accountable for quality,
timeliness, and delivery.
Examples of Knowledge Steward Meta Data …
- Knowledge Recorder/Creator – person or persons who originated and/or recorded the knowledge
- Knowledge Maintainer – person or persons responsible for keeping the knowledge current
- Knowledge Approver – person who must review and approve the knowledge before it is made available
- Last Review Date – date in which knowledge was last reviewed for accuracy and completeness
- Next Review Date – date knowledge needs to be next reviewed for accuracy and completeness
Knowledge Business Meta Data
A knowledge artifact is a defined piece of recorded knowledge that exists in a format that can be retrieved to be used by others. A process drawn on a napkin at lunch can become a knowledge
artifact if it can be recorded for someone else to use. Typically artifacts are something more tangible – i.e. a document, a picture or graphic, a video, an audio, a project plan, a
presentation, a template … to name several.
Artifacts are typically recorded by business units and for specific business functions. These artifacts are of certain types, and fall into different classes.
These basic pieces of information (in italics) about artifacts along with the steward meta data are the beginnings of a knowledge taxonomy that can be developed and used to classify and
record every piece of recognized knowledge in a knowledge repository (database that stores information about the classified knowledge). This taxonomy requires the development of a knowledge meta
model (see figure 1) and enables the company to name and classify the knowledge in the repository. Once the knowledge is captured in the repository, it becomes possible to search for and identify
the knowledge that exists, and retrieve the knowledge when it is needed.
Using the same example from company ABC, the process for handling customer complaints may have been initiated in the product development area. The function that the process addresses is both
technical support and customer complaints. The artifacts about how to handle customer complaints can be classified as being related to a process. Artifacts related to handling customer complaints
may exists in multiple types – i.e. policies, memos, and videos.
- Examples of Knowledge Business Meta Data …
- Business Unit – company defined organizational unit
- Business Function – the function or functions associated with one or more business units
- Business Sub-Function – functional decomposition specific to one or more business function
- Business Location – business location where knowledge can be applied, region, store, headquarters, country, …
- Artifact Type – how the artifact applies to the business function and sub-function
- Knowledge Class – the recorded format of the artifact – policy, memo, video, …
Knowledge Artifact Meta Data
Consider how difficult it is to find a document that you created on your own computer six-to-twelve months ago. To find this document, it is important to remember its name and where you stored it
on your local or network drive. Multiply that difficulty by the number of knowledge workers in your organization and the non-standard way in which documents are named. Multiply that difficulty by
the potential number of physical locations for storing that document and you may begin to get the idea of how difficult it is to manage the knowledge of an organization.
Creating knowledge artifacts with “next use” in mind is not easy especially when individuals are used to keeping knowledge (potentially documents) to themselves. Part of the knowledge
development process includes giving each artifact a name. The artifact name is an important piece of meta data about an artifact. It is important to create an artifact naming convention much like a
data naming convention that contains a context, modifiers, and class words.
A naming convention for knowledge can be built using the knowledge business meta data described in the previous section of this article. For example, artifacts may be created by a business unit and
for specific business functions. The artifact may be of a certain type and will typically exist in one or more formats. Using just this high level description of how an artifact is created, one can
derive a simple artifact naming convention that includes just this information.
For example, the customer complaint handling procedure mentioned earlier may be named as follows …
Technical Support Customer Complaint Process Memo
Technical Support Customer Complaint Process Video
As with data naming conventions, there is no definitive way of how knowledge artifacts should be named. It is important to find a naming convention that works for your business, develop it as a
standard, and enforce the standard. Using the simple naming convention above, it would be easy to find artifacts that are associated with the Development business unit, Technical Support function,
specifically the processes associated with handling Customer Complaints, that are in policy, memo, or video format.
The location in which an artifact is stored is also important. When creating a knowledge meta data repository (just like a data meta data repository), the repository itself does not store the
actual knowledge, it likely will store a pointer or series of pointers to where the knowledge artifacts are stored.
Other types of knowledge artifact meta data include the format of the artifact (Word, Excel, jpeg, mpeg, …), the proposed audience or users of the artifact, and keywords that can be manually
defined or defined by using a piece of software specifically designed to scan a textual artifact to retrieve and classify words that will be related to the artifact that can be used to search for
artifacts using a key word search.
- Examples of Knowledge Artifact Meta Data
- Artifact Name – physical name of artifact as it exists in storage
- Artifact Location – physical or virtual location of artifact
- Artifact Format – physical format of document typically tied to a piece of software
- Artifact Audience – projected users of the artifact
- Artifact Purpose – proposed use of the artifact
- Artifact Keyword – words that are associated with an artifact that relate it to a specific business area, function or purpose
Meta Data and the Impact on Searches
Getting individuals to identify knowledge artifacts, classify the artifacts using a knowledge meta data database and put them in formats that can be used by others in the organization requires a
great deal of discipline, changes to corporate culture, and a significant impact the social makeup of a company. However, once the process to harvest artifacts is in place, getting people to share
and use available knowledge requires the ability to readily locate that information through the use of portal search utilities.
Three kinds of searches that can be performed on a knowledge repository include full text searches, key word searches, and meta data searches. In the case of key word searches and meta data
searches, a company is required to assign specific information to each artifact that will be used to locate knowledge for a specific use. Several vendors has created scanners and other programs
that will scan through a document and assign key words or make an attempt at classifying artifacts according to a user defined meta data structure (meta model).
Full text searches scan documents from top to bottom looking for words or parts of words that match the user’s selection. It is difficult to apply this type of search to graphics,
audio, video, anything that is not textual in nature. This type of search returns textual artifacts that contain the selected words and ranks them by level of appropriateness according to how often
the word was found or how the words are combined in the context. Results of these searches are typically poor with more results returned (artifacts that match the criteria) then the requester
Key word searches use specific words that are associated with artifacts to identify which artifacts that will be returned as part of the search process. The application of key words
typically is applied manually at the time that an artifact is recorded in the knowledge repository or through automated functions that are developed to scan artifacts and record key words. Results
of these searches are typically good with less total results, but more focused results, then the full text search.
Meta data searches are based on a pre-defined taxonomy or classification schema (perhaps using the types of meta data defined above) that identifies specific business-defined attributes
for every artifact that is recorded. The taxonomy can be applied manually at the time that an artifact is recorded in a knowledge repository or through automated functions that are developed to
scan artifacts and record the classifications in the knowledge repository. Results of these searches are typically found to be excellent when the specific selection criteria are based on the
taxonomy and are built into the search engine.
This article has briefly highlighted three distinct (but generic) categories of meta data that can be related to the knowledge in most companies. The three categories include steward meta data (the
people associated with the knowledge), business meta data (business unit and function of the knowledge), and artifact meta data (artifact name, location, format, audience, …). The article also
briefly described three types of searches that can be performed on artifacts to enable knowledge workers to retrieve relevant knowledge through search engines.
Assigning budget and resources to meta data has been a hard sell for many companies even though most of the “experts” state that managed meta data lies at the core of business
intelligence success. Companies that have provided generic meta data to support their data warehouses or data marts have found that the ability to understand the data in the database has directly
impacted how the data warehouses are used and accepted.
Meta data also lies at the core of knowledge management. Information that is known about the knowledge that exists in a company will have the same impact on how well knowledge is managed and how
other people in the organization will identify and make use of that knowledge.