Chatbots have the potential to be the next big thing in customer communication. But most chatbot technologies still lack the power to easily interpret and engage in normal conversations. While companies can’t control the way people communicate, they can control how their writing teams form their content, which directly affects the technologies’ ability to interpret the content appropriately. Structured authoring bridges the gap between current chatbot technology and level of humanesque interaction customers want.
Chatbot success hinges on the ability to deliver the content that answers a user’s question. To do this, a chatbot must understand not only the user’s question, but also the user’s intent, and then deliver only the chunk of content that assists the user. Bots built on unstructured content rely on complex and opaque NLP systems to produce meaningful answers from the content. When attempting to implement a chatbot using unstructured content, many companies will find themselves investing heavily in technologies and services with uncertain results. The old adage “garbage in, garbage out” is alive and well in chatbot implementations.
Fortunately, the technology for human and machine-readable text has been around since the 1970s. eXtensible Markup Language (XML) is a semantic and structural markup language for text documents. Using XML, companies can add semantic markup to the content, making it easier for chatbots to interpret the input, even when that interpretation is being done with advanced machine learning systems.
For chatbot implementations to be successful aids for end users, companies need structured content that’s organized and well-maintained. Otherwise, the chatbots won’t be able to consume the content and generate human-like responses to questions. By improving their content’s semantic structure, companies can bridge the gap between their current chatbot capabilities and their customer’s expectations.
Implementing structured content often means a technology change, as well. There are many systems that specialize in creating and maintaining structured content, easyDITA being one example. Specialized systems for managing structured content are going to be the tools that enable your team to scale content output and control content maintenance costs.
With the right content structure and management in place, a simple chatbot can give contextual and personalized answers from the same source-content that can be used for multiple outputs, such as PDF or a knowledge-base. Single-sourcing in this manner drastically reduces the need to maintain multiple variations of content to answer the same questions in different formats.
In the end, chatbots are no different than other more established output formats that are used to support common business functions, such as learning and training, human resources, and technical documentation. Often times, all of these groups can leverage the same system and reuse each other’s content. Moreover, most single-sourced, component content management systems support multi-channel publishing, for example, to online help, learning platforms, and typical document formats like PDF. Chatbots built on systems that already support all these common business groups only adds further value to your company’s content.
Allstate’s Business Insurance ABIE is a great example of successfully using a component content management system to support a chatbot. “ABIE is a shining success of what chatbot technology can do when it’s built on structured content and a platform like easyDITA,” said Patrick Bosek, cofounder of Jorsek. “Many chatbot implementations struggle to scale because they’re built on single-purpose content that is expensive to build and maintain. Structured content provides a scalable input for chatbots to ensure long-term success.”