CODEWITHSUNDEEP
androidartificial-intelligenceibm-watson
Droidme
Droidme

Reputation: 1252

How to handle an AI response in an Android app

I am trying to integrate AI with an Android app. I have a scenario where I need to make an API call and get data based on the response from watson on the entities recognized. Like "Show me blue Jeans Pants" will make an API call to get details from server.

But it should make an api call in the case of "Show me blue Jeans shirt" or Blue Jeans caps. The portal responding with high confidence score as it is recognizing 2 entities (blue, Jeans) in all the three cases. How do we handle such cases? Should it be done on the portal level or app level?

Upvotes: 0

Views: 224

Answers (1)

Mrutyunjaya Jena
Mrutyunjaya Jena

Reputation: 301

If your application is a chat-based app, then you can use Watson Conversation service. It can detect intent, multiple entities along with the confidence of each of those entities and also you can create dialogue flows which will provide you with a great user experience.

But the Watson service can't make an API call to get the actual data from the server/database. That you'll have to handle from your application itself. The ideal way to do this is to parse and analyze the response from Conversation, save the intended entities and accordingly make a database call from the app to get the actual data from server/database and then show it back to the user in your intended way.

Whereas if your application is not a chat-based one, then you can look into Watson Natural language Understanding service. NLU can give you entities, relations, keywords, sentiments and much more. The idea of integrating it with your application is still the same as Conversation just that the response structure will be a bit different.

In case you want to have your own custom model in NLU, you can look into creating your custom model using Watson Knowledge Studio and then deploying your custom model to NLU.

Upvotes: 1

Related Questions