Extract key content from a conversation response

Question

I'm trying to understand the best approach to extract key content during a conversation. I'll use a simple travel interaction example:

I've created a conversation that supports travel activities, and after greeting the user they are asked how can I be of help?

The user responds: I have a flight to Las Vegas today and I need to make a change to my reservation.

Defining the proper Intents and Entities we conclude they want to change a reservation. I want to understand how can I identify, and extract from the response key words like "Las Vegas" and "Today"?

Is there an approach using Watson APIs or do I need to write custom code to dissect the response based on the matching Intents and Entities?

As an example I'd like to confirm to the user: I understand you have a flight to Las Vegas today and would like to make changes to that reservation. Is that right?

Appreciate any advice you can share.

Answer 1

It sounds like you need a place/destination/airport entity, for places like Las Vegas, and an entity for days. You can then use conditions in your dialog flow based on which entity value was matched, for example @place:(Las Vegas). Or you can just repeat the matched entity back to the user in the response, for example I understand you have a flight to @place.

Have a look at the Dialog reference documentation for more information on using entities.

Also, if it helps, there are some examples of day entities in the conversation-starter project entities.csv file on GitHub, and I'm tempted to add an entity for airport codes there if I get a chance.