Digital Assistant and Mobile

Get Involved. Join the Conversation.


    Vikhil Khobragade
    Prediction and Recommendations in ODA
    Topic posted July 29, 2019 by Vikhil KhobragadeBlue Ribbon: 750+ Points, tagged Analytics, Bots, Conversational AI, Messaging, Notification, OMCE 
    44 Views, 4 Comments
    Prediction and Recommendations in ODA
    Implementation for prediction and recommendations


    I'm looking for sample implementation for prediction and recommendations in Digital Assistant.

    e.g. For banking use case, when the existing customer log into the bot based on his customer profile or previous interaction to the bot.

    i would like to recommend him/her certain banking product i.e car loan or home loan.

    Can you please guide me how this can be implemented in ODA.






    • Grant Ronald

      Good question - however this kind of requirement for a "recommendation engine" really sits outside of ODA - using I guess something like Oracle Analytic cloud might have some knowledge - after all, the recommendation of e.g. a banking package, is being made based on data not necessarily related to ODA.


      • Vikhil Khobragade

        Is there any API for who predict the user recommendation. after all when user interact with the bot means bot must learn automatically because we train the intent and entities using ML model in the ODA.

        When I'm train the model using ML is the bot learn something or what?

    • Grant Ronald

      The bot is "learning" that, for example "What is my balance" or "how much do  I have in savings" can be classified as the CheckBalance intent.  It has no knowledge whether your credit and spending history is such that you should be offered a gold credit card - when you find the right recommendation engine then it easy enough to call that from ODA, but its not the bots job to know what is the best loan for you.


    • Frank Nimphius


      best to look at Oracle Digital Assistant as a smart user interface that can call backend systems within a user conversation. In your case, you need to call a backend system through a customer component to get the user recommendation you are looking for. One option I am aware of (and there may be others) at Oracle for such a system is Oracle Adaptive intelligence.

      Note though that Oracle Digital Assistant is not restricted to work with Oracle systems. Any REST API can be accessed through Oracle Digital Assistant and custom components.