Structuring content can mean different things to different people. From editorial structure to technical structure to metadata structure, we can look at six different techniques to structure content.
Each level of structure brings its own advantages, and can be divided into two types: on-page structure and off-page structure. These levels may seem hierarchical, but are more like building blocks that, together, form the foundation of how content structure can reinforce the structures built on it.
On-page structuring techniques
On-page structure refers to editorial conventions and structures that are perceivable to the content consumer.
Technique 1: Using established editorial conventions on your copy
What is often called “editorial structure” is actually editorial convention. Those conventions can change from culture to culture; in English-based content, conventions are relatively baked in. When you squint at a document on screen, you can tell what the elements are by their placement and sizes. For example, you can identify images, headings, sub-headings, footer, page number, and bullet points – all because we understand the conventions. If someone were to leave off the paragraph that’s the preamble to a bulleted list, there will be no technical issue that prevents the file from being published. The only consequence is that readers may have to work a bit harder to understand the significance of the bullet points. Persuasive content, such as, marketing material and advertisements often break these conventions willingly, yet we still understand the messaging.
Technique 2: Following an editorial structure
Editorial structure is the next level up, and has both editorial and technical implications. Content is structured by way of multiple structural elements – or tag pairs – that machines “read”. Processing the content between the tags signals the intent or context, to help with search. Also the processing can affect the way the elements are laid out, which makes it easier for people to understand each element. For example, we can understand the elements of a recipe because we have an ingredients section and an instructions section.
Technique 3: Bringing in content-level semantics tagging
Extending the example of recipes, we can look at content-level semantics. As recipes get syndicated and shared across many sites, content-level semantics enables the automation. For example, each ingredient could be tagged with a specific structural tag that indicates they are ingredients. Another set of tags could be for quantities. This allows programmatic adjustments of quantities – doubling a recipe, for example – or to switch the quantities between metric and imperial measures. Structures are standardised and rigid to enable machine-readability, and ultimately to provide a better user experience.
Off-page structuring techniques
Off-page structures affect more than what is on a page – it could affect content at a document level or wider, even an entire body of content.
Technique 4: Creating taxonomies and ontologies
How we name things is important. So we start with a controlled vocabulary for all of our concepts. We then categorise that terminology hierarchically, and that’s the taxonomy. For example, content users could be the parent category to multiple types of users: prospects and customers.
The next level up is a thesaurus. This adds equivalency terms to a taxonomy, such as synonyms and other related terms. Extending that same example, a prospect = candidate, while a customer = client = purchaser.
Then there is an ontology, and that creates relationships between multiple taxonomies. An ontology is a formal way to represent knowledge and relationships within a domain. Seth Earley, who is a prominent taxonomist in the US, says to think of ontologies as “master data management” for AI.
Technique 5: Building a knowledge graph
Which brings us to what I think of as the most sophisticated form of classification, which is the knowledge graph. The knowledge graph is an instance of an ontology. So an ontology could be about a classification of music, for example, which establishes relationships between the various genres, artists, instruments, and so on. Whereas a knowledge graph might be used by music labels to establish relationships not in the ontology, such as which of their artists are assigned to which agents, which countries their music is licensed to play in, and so on.
Knowledge graphs act as “knowledge scaffolding” that can give a company insights into things like user preferences – think of Netflix knowing that you like documentaries, romcoms or action movies.
Technique 6: Layering in information architecture (IA)
That last structuring techniques brings us to information architecture. I sometimes use the example of a kitchen and dining room. A taxonomy is the system you use to store plates, glasses, and cutlery, whereas an information architecture is how we assemble those elements to make sense to users: place settings presented with one plate, one glass, and one set of cutlery per person.
So, let us use our example of how Netflix knows that you like documentaries or action movies. How can Netflix recommend other documentaries or action movies to you unless they have a way to assign films into categories such as documentaries or action movies? Or, how can a bookstore recommend books of the same genre (murder mysteries), or topic of interest (Roman Empire), or even in the right language (French) unless they have categories for that? This is where IA comes in. So information architecture plays an important foundational role that AI uses to give you better results, to either be offered better options, or when answering a specific query.
Structuring content for artificial intelligence
The bottom line is that companies want to use AI to unlock business value of some sort. That could be value to customers, or it could be operational efficiency. In an environment where we generate so much content that over 25% of our time each day is spent looking for information (not just any information, but for the right version of the right document for the right purpose), that’s a lot of wasted time – over a day a week.
While we don’t measure customer value in terms of the time they spend looking for information or clogging up customer support lines because they’re too frustrated to keep looking, we do measure the cost of losing a customer and needing to replace them. The cost of customer acquisition is a staggering 5 to 7 times the cost of customer retention.
There is a huge incentive to ensure that customers can be served in the most efficient and effective ways. Whether that’s a direct self-serve model, or getting the right information, and quickly, to customer support agents. Either way that’s a strong reason to return better answers through your chatbots. Using structure as a starting point lays a solid foundation for all of the advanced technologies that get layered over top.
Ready to bridge the gap between data and insight?
Let us help you take your enterprise knowledge into the AI era.
