Cyril on Business Intelligence

We appear to have consensus that effective management of tacit knowledge is a key part of any successful corporate BI environment. Henry Minzberg is the father of this movement with his seminal article in HBR July-August 1975 and the splendid quote “The strategic database of an enterprise is in the minds of its managers, not in the memory of its computers”.

This post is intended to start a discussion thread on how we can maximize the value of tacit corporate knowledge – the opinions, ideas, explanations, suggestions that are known, but not shareable. The data warehouse, schema, metadata, etc. are all unconscious of these data items, yet they are critical to successful business.

If we don’t manage this resource, we are condemned to:

Solve problems many times over

Make errors in interpreting hard data – because we don’t have the expert’s explanation

Be unaware of early warning signs that people know about, but the data doesn’t yet show

Corporate performance management is impractical if the reported information is not accompanied by the explanations of subject experts, and assessments of implications.  You’re not measuring corporate performance, you’re just keeping score without knowing who is winning. The hard without the soft is pretty much useless.

Service Oriented Architecture is just so much spaghetti, unless the sharing of what is known about the business is not only possible, but empowered.

A first step in managing the milieu of tacit knowledge is to realize that we can neither share, nor search, nor retrieve soft information that is not classified. It’s like having an iPod with all the music dumped in one sector, or navigating a city without street signs.

Classification is the heart of the task, and classification requires that you have a vocabulary. Here is one area of corporate IT where Nazi principles are required. Everyone must use the same “label” to describe the same concept – synonyms can be poisonous, one person’s synonym is another’s swear word.

Vocabularies, often called taxonomies or category listings, are becoming an important part of information technology as used by people and corporations to manage the huge pool of documents, email, news and business intelligence that flows throughout the community.

Vocabularies are valuable components of a knowledge management support system, providing the categorization model that facilitates efficient searching, browsing and alerting. Without an adequate vocabulary, the users are forced to build their own personal rules for document searches, and agent specification. Often, this yields poor retrieval, or people just pass over the opportunity to become better informed, because it is too onerous.

A vocabulary for BI tacit knowledge is a set of preferred topics or keywords that facilitates categorization of documents in an enterprise.

It is equally applicable to business or government and will usually be specific to the particular organization.

The vocabulary typically will comprise business entities (e.g. customers, competitors, government programs, products), business issues (quality, problems, strategies, employee matters, plans), technologies (telecommunications, optics, computing) and competitive intelligence (industry news, patents, reports, email), useful to the corporation.

For BI applications, the vocabulary ought to have both concrete concept related topics, e.g. customer names, places, projects, technologies, and soft concepts e.g. strategic partners, problem projects, important issues, etc.

Included in the vocabulary construction often is a set of rules used to automatically associate specific topics with relevant documents – the auto-classification process.

This is why you need to consider building an enterprise wide vocabulary:

Tacit information oriented BI systems facilitate collaboration among colleagues.

Vocabularies used for classification improve document access via searching, alerting, browsing.

It is almost impossible to share, browse, search or filter a large amount of information unless it is categorized.

Automatic categorization of a large, existing and evolving collection of documents in many databases is often required, and vocabulary based classification is proven as effective for this.

Pure text searching is not adequate in a corporate context, it throws up too many hits and the user needs to know what to search for.

Browsing the collection using topics from a vocabulary as the directory is efficient and satisfying to the user.

Alerting to new material requires individual specification or personalization of the type of documents required, and the subject category, chosen from a vocabulary, greatly facilitates this.

The growing enthusiasm for metadata as a standard document description process, using XML as the storage format stimulates the need for, and the value of, vocabularies for BI

In subsequent posts, I will describes how vocabularies may be used in Business Intelligence, what they contain, how they are best constructed and issues with scalability. These principles are based on my several years experience with the Grapevine knowledge sharing system.