Client Login

Please contact us to set up your account with Precision Content

Enabling Content in Chatbots and AI: The Microcontent Solution

Chatbot assistant conversation, Ai Artificial Intelligence technology concept. Casual man chatting with chatbot via mobile smart phone application on virtual screenArtificial Intelligence: Powering the Future of Conversational Interfaces

Conversations about the future and potential of Artificial Intelligence (AI) have become ubiquitous. Is it the herald of job loss or the gateway to a wonderful new future?

Nihilists tell us that it will be the end of jobs as AI replaces workers en masse. Optimists profess that AI heralds a bright new future full of innovation and efficiency. In the communication community, what does the future of AI hold?

In the most immediate future, AI opens the door to increased communication possibilities via chatbots and other conversational interfaces. As we enter into this phase of automated communication, we’re faced with a number of communication challenges.

From a content perspective, how can we start implementing chatbots today? Chatbots are new, intriguing technology that has the potential to introduce efficiencies and streamline customer interactions.

It is estimated that as of 2020, chatbots powered up to 85% of customer interactions. For context, 85% of the time users go to a website or go looking for information a chatbot is pushing that content to them.

As conversational interfaces like chatbots become commonplace, the technical communications industry has been advocating new ways to publish, categorize, and organize content to make it applicable for AI use. However, the typical publishing process has, until recently, been more akin to broadcasting than to the type of communication that we need to power chatbots and facilitate voice communication.


Going Beyond The Traditional Model

The traditional model of content publishing and distribution has been to create a piece of content, post it on a website, or render it as a PDF (Portable Document Format). This model of publishing often results in the adage: “Is anybody reading this?” Without the direct statement–response kind of communication that bots offer, it’s challenging to determine the usefulness of content beyond metrics that track pageviews or scroll depth.

The opportunities of conservational interfaces rest on the usability of content through a question–answer, prompt–response format.

For example:

  • If I ask a bot, “what’s a widget?”, the bot might answer, “a widget is a type of thingamajig”.
  • If I don’t know what a thingamajig is, I would ask, “what’s a thingamajig?” and the bot would provide an answer based on the content that I’ve categorized as “thingamajig”.

As we continue to create our content, we must be planning for these use-cases. To accommodate conversational automation devices like Amazon’s Alexa or Google Home products, we must consider how we write, organize, and distribute our content.

We need to look at our content in terms of small chunks. At Precision Content, we created a Microsoft Bot Framework-based chatbot prototype using a specialized DITA (Darwin Information Typing Architecture) format and connected it to Microsoft Teams. We created this DITA format to capture small chunks of content created in a content management system. We pushed it out from that content management system into WittyParrot (a portal presentation system that focuses on small chunks of content). Inside our Microsoft Teams chat room, we can enter questions and the bot will pull from our content.

If I ask, “What is a category table?” The bot searches through our knowledge base and returns an answer. This is a good first step, but what if I wonder, “What does a category table look like?” and ask, “Show me an example of a category table”. And again, the chatbot goes out, and retrieves that information. These are small fragments of a conversation. They are individual answers to questions as though one topic is one answer.

But often, that’s not the way topics work. Our topics tend to be much broader in scope, especially when we consider concept topics and reference topics. What we have found is the standard DITA topics typically don’t fit the conversation model. The way we have traditionally created our DITA topics has typically been too broad to fit into a chat model.

If we consider that chat via a chatbot is somewhat akin to a Twitter conversation, we have character limitations to account for. Our responses need to be short, small, and fall into a sequence. When we set out to organize our content to make it accessible for conservational interfaces there are a number of considerations.

  1. We need to create DITA topics to include microcontent, rather than considering topics to be the key building block of data. Micro content is focused on one primary idea, fact, or concept.
  2. We typically have content blocks within a topic. A block has one idea and a topic can contain multiple blocks. Ideally, our blocks would be easily scannable to tell one from another.
  3. We are going to label our blocks so that we know exactly what each block is.
  4. We are going to write and format our blocks specifically so that they can be used wherever they need to be.


Getting Started

How do we start this process? We begin by determining what our content sections have to be and then create microcontent, blocks, and tight structures within the topic for specific purposes, like reader response.

When we write a blog, for example, we want people to get something out of it. We want to write it in such a way that we direct what they’re going to get. Sometimes this requires that we write the same material in different ways depending upon its use. When we acknowledge that people are coming to our content from different perspectives, we can plan to write it in multiple ways.

A common way to model DITA is using the concept of Lego blocks. In this model, the individual Lego blocks represent topics and there is a map, which has all the assembled Lego blocks to form a structure, like a little triangular wall. Our perspective at Precision Content is different because we look at a block structure that is feeding the topic. Blocks are assembled into a topic, and topics become a map.

We give our writers topic templates and ask them to create topics, but we give them some block structures within those topics to follow for specific uses. Or, they might need to add a few blocks based on their own discretion, depending upon what they’re doing. We are still authoring topics, but we are thinking about the substructure of those topics carefully. We have five blocks that we feel fit most content types. We begin from the understanding that writers must consider the intended reader response because it focuses the writers on writing what’s needed to elicit the desired action.

We have five block types: reference, task, concept, process, and principle:

1. The reference block type describes things the reader needs to know.

  • Imagine the reader is creating a new database and they need to know how much disk space is required. As content producers, we document that in a reference book. Knowing how much disk space is required is something our reader needs to know to accomplish a task, but they don’t need to retain it.

2. The task block instructs the readers how to do things.

3. The concept block explains the information that the reader needs to understand.

  • If the person creating the database needed foundational knowledge before they could create it, then the concept block type is how we tell them what a database is so they can clearly understand that this is the knowledge that they need to retain to be able to use the product.

4. The process block type demonstrates to the reader how things work.

  • Process is written from the perspective that tells the reader how things work.

5. The principle block type advises the reader about what they need to do or not, and when.


How do we get from block types to chatbots?

Chatbots operate by trying to determine user intent. Our content types speak to user intent because our types become metadata to feed into the bot process. By creating semantic blocks, and block types within our topics, we’re starting to build those conversational structures required by bots to function properly. Using these structures, our authors are guided in their efforts to produce quality content and our readers can find what they need within our content. Moreover, we can create content that is effective for conversational interfaces.

If we begin by authoring every topic with blocks of microcontent instead of looking at topics to answer a question our content can support just about any use, like AI-based chatbots and other interfaces. Microcontent also has other benefits.

  • It can render content clear for users by being scannable.
  • Titles and subtitles can be effective metadata for users. They tell the reader clearly what they are going to read.
  • Microcontent guides writers using smaller patterns of content.
  • Microcontent enables flexible future use for a variety of delivery channels.
  • A microcontent perspective can improve the quality of content through structure and organization.

At Precision Content, we believe that microcontent will make content easier to write and future-proof. Even though chatbots are a relatively new technology, our Precision Content standard predates bots. We didn’t have to restructure our content to be able to build a chatbot. Our content was prepped and ready when the technology was. Using this framework, we are able to push content out to a voice interface without having to make changes. Using a microcontent framework, you can futureproof your knowledge base and save your business time and effort as new technologies like conversational interfaces, AI-based chatbots, or other augmented reality interfaces become prevalent.


Is your content ready for what comes next?

There’s no time like the present to get started with preparing your content for the next wave of technologies so that you can move forward with intelligent and scalable solutions — Contact us to start the conversation!

About the Author

Rob HannaRob Hanna co-founded Precision Content in 2015 to pursue his goals to produce tools, training, and methods that will help organizations make their high-value content instantly available to all that need it including customers, staff, partners, and even other information systems that need to consume that content. Driving this development is the Precision Content® Writing Methods, based on the best-available research over the last 50 years into how the brain works with information. Today Rob leads his highly-skilled team of content strategists, information architects, writers, trainers, and developers to serve the needs for digital transformation for businesses across North America.

Join the Precision in Practice Newsletter

* indicates required
Sign up for the Precision in Practice Newsletter *


Recent Posts