Genesys Cloud - Main

Hello,

Today we're really excited to release an improved sentiment analysis model for English.

The updated model uses context from surrounding phrases to improve the detection rates of positive and negative sentiments. 

Interaction processed by older sentiment model:

The newer model improves the accuracy of classifying negative sentiment in the above interaction. It was able to use context to classify negative sentiment in 'I cannot wait too long', and also in a very long and complex phrase 'Although my new iPhone 14 worked really well for the first 2 weeks, it started having screen issues in the third week, and now Apple wants to charge me over 300 dollars to repair the screen even when the phone is almost brand new and covered by warranty.'

For more information, see About sentiment analysis. 

Please reach out to anik.dey@genesys.com if you want to share any feedback on the updated sentiment model.

Thanks,

STA Team

#Reporting/Analytics
#Roadmap/NewFeatures

------------------------------
Anik Dey
Genesys - Employees
Staff Product Manager
------------------------------

As mentioned in my email, there is a lot of interest about what is behind sentiment like:

• Is this a Genesys engine or 3rd-party
• What is used in the analysis or what rules are used?
• How can we affect what good or bad sentiment means?

------------------------------
Robert Wakefield-Carl
TTEC Digital, LLC fka Avtex Solutions, LLC
Contact Center Innovation Architect
https://www.Avtex.com
https://RobertWC.Blogspot.com
------------------------------

Original Message:
Sent: 02-15-2023 13:48
From: Anik Dey
Subject: Improved sentiment analysis for English

Hello,

Today we're really excited to release an improved sentiment analysis model for English.

The updated model uses context from surrounding phrases to improve the detection rates of positive and negative sentiments. 

Interaction processed by older sentiment model:

Same interaction processed by newer sentiment model:

The newer model improves the accuracy of classifying negative sentiment in the above interaction. It was able to use context to classify negative sentiment in 'I cannot wait too long', and also in a very long and complex phrase 'Although my new iPhone 14 worked really well for the first 2 weeks, it started having screen issues in the third week, and now Apple wants to charge me over 300 dollars to repair the screen even when the phone is almost brand new and covered by warranty.'

For more information, see About sentiment analysis. 

Please reach out to anik.dey@genesys.com if you want to share any feedback on the updated sentiment model.

Thanks,

STA Team

#Reporting/Analytics
#Roadmap/NewFeatures

------------------------------
Anik Dey
Genesys - Employees
Staff Product Manager
------------------------------