Everyone is talking about digital transformation, but few organizations are implementing it in a practical way. Digital transformation requires taking a holistic view of the organization to see how process, content, and technology can be orchestrated to improve efficiency and provide a competitive edge. That holistic view of the organization is also the first step in developing an ontology, which is why digital transformation and ontology are so closely linked.
The term “ontology” originated from the discipline of philosophy, and in that context it means a description of the nature of being. That sounds a little amorphous, but there are parallels to digital transformation. In a business context, an ontology should capture the essence of the organization—what its key components are, how they are structured, and how they relate to each other.
Three kinds of relationships are typically captured in an ontology.
The simplest one is a relationship among words, such as synonyms or antonyms. This type of relationship can provide consistency across bodies of knowledge by establishing equivalencies among terms that are related, whether similar or opposite.
Another relationship is a hierarchy in which parent-child structures are established. These may be broad or very specific but these relationships are essentially taxonomies, which organize information within a domain.
Finally, associative relationships provide a way to weave different taxonomies together. For example, documenting the fact that certain products are useful to certain categories of customer is a way to capture context and relevance, making the intersection of two taxonomies more useful than either one would be alone.
To transform a collection of taxonomies into an ontology, these relationships need to be mapped systematically, which requires a long-term commitment. Over time, the ontology can be developed to become the knowledge scaffolding or framework for the organization. The ontology can be applied to many data and content sources, and becomes a valuable resource for problem solving that can be used by people who are not subject matter experts. This type of resource is particularly useful in today’s job market, in which people are moving between jobs very quickly, causing a constant brain drain. An ontology captures the context that is behind domain knowledge. If you have that context, new workers will get up to speed faster, and support for the learning curve will make it more enjoyable.
The taxonomies on which an ontology is based are focused on one dimension. They are often set up for navigation, to assist the user in finding products, for example. Initially, companies may thing that a product taxonomy is all they need, but once they start expanding their thinking on this, they start realizing all the things they need to document. In the life sciences domain model, there might be taxonomies for drugs, diseases, and chemical entities. The model should begin with a conceptual list of things that are important to the business, and then it can be broken down in many different ways. The ontology links the different domains, leveraging the information to present a meaningful picture of the vital components of an organization.
We often hear clients say, “Why can’t we just get Google and find our information that way?” The answer is, because it’s important to organize, track, and manage information resources just like you would accounting resources. You can’t just point AI tools at the data and expect them to make sense of it. Sure, some applications do not require a reference architecture (i.e., a coherent schema with metadata). For example, machine learning can train a system for image recognition to spot cancers by showing it examples, without having to provide any other information. The dilemma is, how do you then apply that knowledge? Is it headed for quality control, manufacturing, or a consumer? When the information is tagged, you can perform different operations on it, such as routing it to a relevant individual or flagging restricted information.
Getting the right information to the right people at the right time is the Holy Grail of knowledge management. It can be argued that the real problem is not information overload, but the lack of a filter, which all too often sends irrelevant information to the user. A filter can be provided by classifying and organizing information properly. When you make information more findable by tagging it, search time can be cut in half. In the case of one medical claims processing company, agents needed to find specific answers in a 300 page claim. Developing a sophisticated content model allowed the company to reduce processing time by 20%. Multiplying this number by the number of agents reflected a substantial ROI.
In another insurance company, a client initially wanted a taxonomy to assist its agents in writing up policies for a new type of business insurance being offered, but the truth was that they needed much more. They needed to componentize corporate content for questions and answers so that the agents could get a specific answer—they did not need a list of documents in which the answer was buried somewhere. The appropriate answers were developed based on a variety of use cases, with thousands of pieces of content. We developed a “helper bot” that was one of the first virtual assistants that had practical value. An essential step was to develop an information architecture and use cases. This has tremendous value, allowing the ability to re-use content in multiple contexts and multiple scenarios.
If your company is trying to tackle digital transformation, the best place to start is by looking at the biggest business problem that your organization has, and then set up a project to address that problem quickly. We live in a world where ROI is important, so showing an immediate benefit is critical. Start with an overall plan but then plan for small steps each day. If you say you have a 10 year plan, no one will buy off on it. Identify some use cases, then build a domain model and drill into a particular area.
Developing an effective information architecture is important in other respects. Internal efficiencies provide a smoother experience for employees and customers alike. You cannot have a seamless customer experience if you are trying to produce it with acts of heroics. Heroics do not scale efficiently, and they burn people out. In order to provide the efficiencies to reduce friction in enterprise process, you need to optimize upstream systems. Every piece of information used in an organization comes from an internal system. Optimization comes from categorizing and structuring information for retrieval, whether it is being retrieved by systems or humans.
Some organizations have created entire new departments for AI content, but in fact the structure needed for AI is the same one that humans need to find and access information for training or operational use. Ideally the systems will capture, codify, and present knowledge in context of anyone’s role in their job. This speeds up all information flows, which allows for faster and better decisions, getting products to market faster, and understanding customer needs. However, without a well-designed enterprise architecture, digital transformation cannot occur, because it is the foundation on which more advanced initiatives rely.
In any organization, efficiency comes with standardization, and competitive advantage comes with differentiation. It’s great to be efficient, but your company also needs to stand out. For example, if you are using the same supply chain software as everyone else, where is the competitive advantage? Ontology will give you an advantage because it is tuned to the specific needs of your customer, and will therefore differentiate your organization. If you can zero in quickly on a solution to your customer problems or can surface information as needed to the customers, your company will be far more successful than a company that does not have this ability.
Nothing about any of this is easy (or sexy) but it needs to be done if your initiatives are going to make headway. Our team of information science experts can help. Give us a shout if you'd like to talk.