Over a million developers have joined DZone.
{{announcement.body}}
{{announcement.title}}

The 4 Key Elements of Conversation

DZone's Guide to

The 4 Key Elements of Conversation

We need to include the four key elements of human conversation in our conversational design process in order to naturally engage users in interacting with systems.

· Big Data Zone
Free Resource

Learn best practices according to DataOps. Download the free O'Reilly eBook on building a modern Big Data platform.

"Conversational interfaces" is a term often used to define systems that hold human-like conversations. Examples include chatbots, voice user interfaces, and embodied conversational agents. The user interacts with these using text input, voice recognition, and gestures. These types of interfaces are often unable to provide a good conversational experience.

In order to create a better experience, we need to be aware of the following four key elements of human-to-human conversations and how they relate to conversational interfaces.

1. Turn-Taking

A turn is one interaction between the user and the system. A conversation is made of at least two turns.

Hi Poncho on Facebook Messenger

Poncho is a chatbot that provides weather forecasts. Despite having a conversational interface that is better than most, Poncho lets the users down by not providing good turn-taking. As you can see in the above screenshot, in the second turn the Poncho has to interrupt the conversation because it is not able to understand the user’s prompt. That is not a good conversation. In this example, the conversation only has one turn. In a real human-to-human conversation that’s similar to meeting a friend and having the following dialog:

Me: “Hi Jane, How are you?”
Jane: “Hi, I’m good thank you!”
Jane walks away...

Now that's an awkward conversation — people rarely have conversations that last a single turn.

2. Context

If the system can handle more than one turn, it should remember what occurred in previous conversational turns (context).

Google Voice Search does a good job of this. Asking Google Voice, “Who was Oscar Wilde?” gives the user the following result:

Google Voice Search interface - First turn

Google Voice Search interface - Second turn

In the second turn, the user responds with, “Can you tell me more?” A human-like answer where the name of the author has been omitted, the system understands that they are referring to Oscar Wilde and it shows relevant results.

3. Reference

The system should understand follow-up questions that include pronouns in order to feel more conversational.

Mitzuku on Facebook Messenger

Mitzuku is a chatbot that won the Loebner Prize in 2013 and 2016. Users can have very long conversations with her, and its site also provides examples of users’ most recent chatlogs. In the screenshot above, you can see that the system understands pronouns so that it can refer back to something that was previously mentioned.

4. Variety

The system should ask open questions only if it is able to handle different ways of saying the same thing.

Hipmunk is a chatbot that promises to help you to plan travel easily and quickly. The two screenshots below show that the system doesn’t support the fourth key element of conversation. The system misinterprets my answer the first time that I enter the airport because it accepts city names and airport codes only.

There is a number of ways that the user could reply to open questions, and spending time on researching them will have a positive impact on the user experience.

Hipmunk on Facebook Messenger

Hipmunk on Facebook Messenger

Conclusion

Researchers have proven that humans tend to anthropomorphize machines in a natural way. That’s why investigating the structure and process of social interaction between humans can enable better conversational interfaces. We need to include the four key elements of human-to-human conversation in our conversational design process in order to naturally engage users in interacting with systems.

This article was first published on the Codurance blog.

Find the perfect platform for a scalable self-service model to manage Big Data workloads in the Cloud. Download the free O'Reilly eBook to learn more.

Topics:
big data ,chatbot ,conversational interfaces ,artificial intelligence

Opinions expressed by DZone contributors are their own.

{{ parent.title || parent.header.title}}

{{ parent.tldr }}

{{ parent.urlSource.name }}