Hi Lee,
The ability for users to provide their own words / phrases that indicate negative / positive /neutral sentiment is a roadmap item.
Currently, the Sentiment Analysis capability within Genesys Cloud has been trained from a data set of millions of external/general samples and then fine tuned from actual contact center interactions.
Note that the word "brilliant" on its own may not represent positive sentiment. The same is true about "anger" being negative. The model uses it's neutral network to determine if the words / phrase actually represents a negative / positive sentiment in contact center conversations. It may be that the system sees it as positive, but not positive enough to tag it.
If you are interested in getting more analysis on this, you can provide your calls through Customer Care who can pass them on to our Data Scientists to give you more insight on why certain phrasing has not resulted in a positive / negative marking. This may also help in tuning our global model.
Hope that helps.
Thanks,
------------------------------
Rakesh Tailor
Director, Product Management - Workforce Engagement
Genesys Cloud
------------------------------
Original Message:
Sent: 10-29-2020 09:36
From: Lee Hetherington
Subject: Define customer sentiment
Good afternoon, I'm looking for some assistance with sentiment testing. Are we able to define words, phrases, that we deem positive or is this already predefined for us?
I have a basic text script which I read out during the call and words such as "brilliant" are not being defined as positive. I haven't been able to trigger a neutral test so far and sometimes words like "anger" will flag as negative but other times aren't flagged at all.
Any feedback welcome.
Thank you,
Lee.
#Reporting/Analytics
------------------------------
Lee Hetherington
British Telecommunications PLC
------------------------------