<script type="application/ld+json">
{
 "@context": "https://schema.org",
 "@type": "FAQPage",
 "mainEntity": [{
   "@type": "Question",
   "name": "What is Utterance?",
   "acceptedAnswer": {
     "@type": "Answer",
     "text": "In spoken language analysis, an utterance is the smallest unit of speech. It is a continuous piece of speech beginning and ending with a clear pause. It is generally, but not always, bounded by silence."
   }
 },{
   "@type": "Question",
   "name": "How is utterance different from intents and entities?",
   "acceptedAnswer": {
     "@type": "Answer",
     "text": "An intent is the user’s intention. On the other hand, an entity modifies an intent. For example, if a user types “show me yesterday’s financial news,” the entities are “yesterday” and “financial.” Entities are also given a name, such as “dateTime” and “newsType.”"
   }
 },{
   "@type": "Question",
   "name": "How to write standard utterances for your chatbot?",
   "acceptedAnswer": {
     "@type": "Answer",
     "text": "1. Write sample dialogs.
2. Be mindful of in-skill and one-shot utterances.
3. Make Every Interaction Count."
   }
 },{
   "@type": "Question",
   "name": "How can I test utterances?",
   "acceptedAnswer": {
     "@type": "Answer",
     "text": "Sometimes it helps to get a focus group together to test the chatbot. The development and planning group have a common understanding among them, leading to the testing heading towards a happy path."
   }
 }]
}
</script>

Utterance

What is Utterance? 

In spoken language analysis, an utterance is the smallest unit of speech. It is a continuous piece of speech beginning and ending with a clear pause. It is generally, but not always, bounded by silence.

However, in NLP, an utterance has an entirely different meaning. An utterance is simply anything a user says. For example, if a user types “show me yesterday’s financial news,” this entire sentence is the utterance.

It’s the input from the user which the chatbot needs to derive intents and entities from. To train any chatbot to accurately extract intents and entities from the , it’s crucial to capture various utterances for each intent.

How is utterance different from intents and entities?

An intent is the user’s intention. For example, if a user types “show me yesterday’s financial news,” the user’s intent is to retrieve a list of financial headlines. Intents are often given a name; a verb and a noun, such as “showNews”

On the other hand, an entity modifies an intent. For example, if a user types “show me yesterday’s financial news,” the entities are “yesterday” and “financial.” Entities are also given a name, such as “dateTime” and “newsType.” 

An utterance is the whole picture; it’s anything the user says. 

How to write standard utterances for your chatbot?

1. Write sample dialogs

It may seem obvious, but writing down the kinds of conversations your users will have with your skill goes a long way in understanding how to craft your utterances. When you use real words in actual sentences, it changes how you think about a user speaking. Start with a dialog worksheet to help understand your users’ interactions.

2. Be mindful of in-skill and one-shot utterances 

It's essential to recognize that in-skill utterances are very different from one-shot utterances. This lesson is easy to forget when you’re creating your interaction model. Here’s an example:

In-skill utterances

Bot: “What can I help you with?”

User: “Tell me about the weather.”

One-shot utterances

User: “What is the weather like?”

When building a chatbot flow, it’s easy to forget about the one-shot examples and focus solely on variations of “tell me about the weather.” It’s easy to forget how different one-shot utterances can be. 

Here’s a quick list of common utterances: 

  • “Tell me about the weather”
  • “I want to know about the weather”
  • “What is the weather like”

3. Make Every Interaction Count

As you can see, there’s plenty to think about when you are writing sample utterances for your chatbot. Yes, one-shot utterances are crucial but taking the time and care to think deeply about how your user will interact with your skill will make every interaction that much better.

Variation of Utterances

For each intent, when you create utterances, try and create different variations of the possible user utterances. These sentences mean the same thing but are constructed in a variety of different ways. One way of doing this is through Natural Language Generation (NLG).

However, when doing it manually, consider the following guidelines:

  • Create utterances of different lengths; short sentences, medium, and longer sentences.
  • Change the words and also the length of phrases.
  • Vary the placement of the entity. You might want to place the entity and the start, middle, and end of the utterance. This will allow the bot to understand the context in which to expect the entity.
  • Mix the grammar up.
  • Pluralization
  • Stemming
  • Punctuation — use punctuation in some instances, not in others, and bad grammar in other cases. Anticipate the way your audience might speak.

How can I test utterances?

Sometimes it helps to get a focus group together to test the chatbot. The development and planning group have a common understanding among them, leading to the testing heading towards a happy path. But getting users of the product or even staff together as a focus group to interact with the chatbot can surface vulnerabilities not previously detected. Merely due to the absence of prior planning and design knowledge.

Review utterances

Continuous review of user utterances is of utmost importance. Dialog logs are an invaluable source of information and training data pertaining to the chatbot. These logs can be reviewed daily or weekly, and you can edit the utterance lists to improve the NLU model continuously.

When reviewing utterances, the focus should be on the 10% of errors which will have a 90% impact on the overall experience. What often happens is that we tend to get lost in the details and make adjustments that have a small impact on the overall conversation. Instead, focus on the big picture.

About Engati

Engati powers 45,000+ chatbot & live chat solutions in 50+ languages across the world.

We aim to empower you to create the best customer experiences you could imagine. 

So, are you ready to create unbelievably smooth experiences?

Check us out!

Utterance

October 14, 2020

Table of contents

Key takeawaysCollaboration platforms are essential to the new way of workingEmployees prefer engati over emailEmployees play a growing part in software purchasing decisionsThe future of work is collaborativeMethodology

What is Utterance? 

In spoken language analysis, an utterance is the smallest unit of speech. It is a continuous piece of speech beginning and ending with a clear pause. It is generally, but not always, bounded by silence.

However, in NLP, an utterance has an entirely different meaning. An utterance is simply anything a user says. For example, if a user types “show me yesterday’s financial news,” this entire sentence is the utterance.

It’s the input from the user which the chatbot needs to derive intents and entities from. To train any chatbot to accurately extract intents and entities from the , it’s crucial to capture various utterances for each intent.

How is utterance different from intents and entities?

An intent is the user’s intention. For example, if a user types “show me yesterday’s financial news,” the user’s intent is to retrieve a list of financial headlines. Intents are often given a name; a verb and a noun, such as “showNews”

On the other hand, an entity modifies an intent. For example, if a user types “show me yesterday’s financial news,” the entities are “yesterday” and “financial.” Entities are also given a name, such as “dateTime” and “newsType.” 

An utterance is the whole picture; it’s anything the user says. 

How to write standard utterances for your chatbot?

1. Write sample dialogs

It may seem obvious, but writing down the kinds of conversations your users will have with your skill goes a long way in understanding how to craft your utterances. When you use real words in actual sentences, it changes how you think about a user speaking. Start with a dialog worksheet to help understand your users’ interactions.

2. Be mindful of in-skill and one-shot utterances 

It's essential to recognize that in-skill utterances are very different from one-shot utterances. This lesson is easy to forget when you’re creating your interaction model. Here’s an example:

In-skill utterances

Bot: “What can I help you with?”

User: “Tell me about the weather.”

One-shot utterances

User: “What is the weather like?”

When building a chatbot flow, it’s easy to forget about the one-shot examples and focus solely on variations of “tell me about the weather.” It’s easy to forget how different one-shot utterances can be. 

Here’s a quick list of common utterances: 

  • “Tell me about the weather”
  • “I want to know about the weather”
  • “What is the weather like”

3. Make Every Interaction Count

As you can see, there’s plenty to think about when you are writing sample utterances for your chatbot. Yes, one-shot utterances are crucial but taking the time and care to think deeply about how your user will interact with your skill will make every interaction that much better.

Variation of Utterances

For each intent, when you create utterances, try and create different variations of the possible user utterances. These sentences mean the same thing but are constructed in a variety of different ways. One way of doing this is through Natural Language Generation (NLG).

However, when doing it manually, consider the following guidelines:

  • Create utterances of different lengths; short sentences, medium, and longer sentences.
  • Change the words and also the length of phrases.
  • Vary the placement of the entity. You might want to place the entity and the start, middle, and end of the utterance. This will allow the bot to understand the context in which to expect the entity.
  • Mix the grammar up.
  • Pluralization
  • Stemming
  • Punctuation — use punctuation in some instances, not in others, and bad grammar in other cases. Anticipate the way your audience might speak.

How can I test utterances?

Sometimes it helps to get a focus group together to test the chatbot. The development and planning group have a common understanding among them, leading to the testing heading towards a happy path. But getting users of the product or even staff together as a focus group to interact with the chatbot can surface vulnerabilities not previously detected. Merely due to the absence of prior planning and design knowledge.

Review utterances

Continuous review of user utterances is of utmost importance. Dialog logs are an invaluable source of information and training data pertaining to the chatbot. These logs can be reviewed daily or weekly, and you can edit the utterance lists to improve the NLU model continuously.

When reviewing utterances, the focus should be on the 10% of errors which will have a 90% impact on the overall experience. What often happens is that we tend to get lost in the details and make adjustments that have a small impact on the overall conversation. Instead, focus on the big picture.

Share

Continue Reading