Genesys Cloud - Main

 View Only

Sign Up

  Thread closed by the administrator, not accepting new replies.

  • 1.  Knowledge Article - Accuracy of phrasings and its confidence score tuning?

    Posted 12-19-2024 05:27
    Edited by Yaw Chin 12-19-2024 05:28
    No replies, thread closed.

    Hi All,

    Working on a simple FAQ digital bot flow with Knowledge Article.

    I did some test with the knowledge articles but find the "phrasings" are not accurate.

    For example on the article screen shot below I have setup with 3 different phrasings. But when I ask a different phrase e.g. "can my cat pay for the delivery" it returns a confidence of 71% accuracy, which is not correct. 

    Is there anyway I can fine tune this? 

    Hope to get some advice. Thanks a lot.


    #DigitalChannels

    ------------------------------
    Yaw
    ITAPPS
    ------------------------------



  • 2.  RE: Knowledge Article - Accuracy of phrasings and its confidence score tuning?

    Posted 12-19-2024 11:38
    No replies, thread closed.

    Hi Yaw,

    I think the only way to give feedback on the knowledge articles surfaced is: Send Knowledge Feedback action

    But maybe someone else can clarify this



    ------------------------------
    Sam Jillard
    Online Community Manager/Moderator
    Genesys - Employees
    ------------------------------

    Original Message


  • 3.  RE: Knowledge Article - Accuracy of phrasings and its confidence score tuning?

    Posted 12-19-2024 21:21
    No replies, thread closed.

    Hi Sam,

    Thanks for reply. If I understand it correctly the "Send Knowledge Feedback action" is more on a Architect level actioned on the to control bot behavior, https://help.mypurecloud.com/articles/control-knowledge-behavior-in-your-bot-flow/

    But what I am trying to do is not creating too much logic at Architect level. I like to make it simple for business users to easily configure the accuracy from Knowledge Article level and publish it. Hope to get more ideas.

    Regard,

    Yaw Han.

     



    ------------------------------
    Yaw
    ITAPPS
    ------------------------------

    Original Message


  • 4.  RE: Knowledge Article - Accuracy of phrasings and its confidence score tuning?
    Best Answer

    Posted 12-19-2024 23:38
    No replies, thread closed.

    As you discovered, there is not a litany of ways to make your knowledge work better.  The suggestion I would have is that if that particular phrase is really important to you, use an intent with a slot fulfilment.  So for your example, you would have an utterance like "Can my [personslot] pay for the delivery?"  Then in the slot, list a bunch of person types like "mother, brother, son, wife, guy down the hall, etc" and then from that Intent, Play the knowledge article by choosing it in the exit path for that Intent.  



    ------------------------------
    Robert Wakefield-Carl
    ttec Digital
    Sr. Director - Innovation Architects
    Robert.WC@ttecdigital.com
    https://www.ttecDigital.com
    https://RobertWC.Blogspot.com
    ------------------------------

    Original Message


  • 5.  RE: Knowledge Article - Accuracy of phrasings and its confidence score tuning?

    Posted 01-08-2025 06:12
    No replies, thread closed.

    Hi Robert,

    Thanks for your suggestion. Seems like there is no way to optimize the "knowledge article" phrasing and need to do the heavy lifting at architect level. Which again I find it not practical for business users to manage it at architect level.

    Honestly I am not sure IF this is the correct "Article phrase behaviour" for e.g the output still show the article (with 73% confidence) even the phrase is not configured. 

    Not sure if anyone has experience in the Knowledge Article phrases best practise and how to control its behaviour?



    ------------------------------
    Yaw
    ITAPPS
    ------------------------------

    Original Message


Related Content