How to Chat about Chatbots - Part 2

[dangrady510]

This is the second post of a 2 part blog series that explores our relationship with a relatively new entity that we are all conversing with on an almost daily basis in both our consumer and work environments:

the Chatbot.

In this series, we will help demystify chatbots by sharing some of the terminology you may be hearing around the virtual water cooler when chatting with your friends and colleagues.

In the first post, we defined some of the high level  terms you may be hearing from a customer experience perspective such as conversational UIs and messaging apps.  In part 2, we will be defining some of the key aspects of the technology behind the bots like AI, Natural Language Processing and Utterances.

In order for chatbots to be successful at delivering business outcomes like improved customer and employee experience and reducing service delivery costs, they have to be "more than just a pretty (inter)face" conversing with us via our favorite messaging or business application.

If you look up "more than just a pretty face" in the Merriam-Webster online dictionary you get this definition: "being attractive but also having other good qualities, such as intelligence". 

The intelligence behind most chatbots is often referred to as Artificial Intelligence or AI. (sometimes you'll also hear Cognitive Computing).

AI is the general concept of machines acting in a way that simulates or mimics human behavior.  This is an extremely broad definition that could cover many things like visual perception, speech recognition, decision-making, and translation between languages. 

If you are specifically discussing artificial intelligence with someone as it relates to chatbots, you'll most likely hear them mention NLP, or Natural Language Processing, which is a blanket term used to describe a machines's ability to ingest what is said to it, break it down, comprehend its meaning, determine appropriate action, and respond back in a way the user will understand.  If the bot is the (inter)face, then NLP is the Brain.

The development of NLP applications is challenging because machines/computers historically have required us to interact with them in a very structured and precise way.  As you(an the person who edits my blog posts) are very well aware the human language is not precise and is filled with many complexities like slang, different dialects, social contexts, misspellings, and grammatical errors.

NLU, or Natural Language Understanding, is the subset of NLP that helps deal with all of those complexities and focuses on how to best handle unstructured inputs and convert them into a structured form that a machine can understand and act upon.  NLU is a vital component of NLP because if the machine cannot comprehend or understand the intent of the content how can it process it correctly?

When working with NLU as it pertains to chatbots there are some additional core terms or concepts it'll be helpful to get comfortable with:

• Utterance - describes anything someone is typing or saying into an interface.  An utterance is the sentence that we are attempting to understand and identify its intent.
• Intent - defines the purpose or goal of an utterance.  For example, if a user were to type "show me yesterday's email incidents", their intent would be to get a list of incidents.

For processing purposes within applications, intents  are often given a name that is made up of a verb and a noun, such as "listIncidents".  Typically when training an NLU system, you'll seed it with server utterances for training purposes and then it will continue to learn over time as users interact with the chatbot.

• Entity - a noun that adds context to the utterance and would modify the intent so the chatbot can better understand and serve the end user.

In our example above, the entities would be "yesterday" and "email".  These entities along with any contextual information we can gather about the user from the system of engagement like their location, whether they are on a mobile device or not, what assets they have access to, etc. can be used to enhance the conversation and provide a more meaningful and streamlined experience for the end user.

So now that you have an understanding of some of the entities within the sentences your colleagues have been uttering about chatbots in the office, you'll be better able to determine the intent of their conversations and engage in more meaningful conversations about the positive impact chatbots can have on the organization.

If you missed part 1 of the series you can find it here.

Learn more about the ServiceNow Virtual Agent here.

Other Blog Posts:

Natural Language Understanding makes the Virtual Agent the Road More Traveled

The Digital Ensemble That Helps Create Great Virtual Agent Experiences 

Is "Chatbot" a misnomer? Let's talk about what creates successful chat experiences