Hey Renaud,
Currently, sentiment feedback UI and API are the only two ways to adjust sentiment models if you feel the score is not accurate. There is no export ability to get the sentences and associated scores from interactions in a report, but this is something we can look into, are you able to create an idea in our ideas portal?
For improving transcripts, you can add those problematic phrases to Topics as you're doing right now. Early next year we will release a new feature called Dictionary Management, which will allow you to add these product names and also how they 'sound like' to our transcription engine, and this will significantly improve the transcription accuracy.
For betas, can you drop me an email
anik.dey@genesys.com and I can get them enabled for you.
Thank you.
------------------------------
Anik Dey
Genesys - Employees
------------------------------
Original Message:
Sent: 11-25-2022 07:44
From: Renaud Larcier
Subject: Sentiment Analysis/Speech Analytics
Hi
We recently acquired PC3 and we're starting to analyze the data reveaded by Transcript and Sentiment analysis.
- As others on this thread we're facing instances where we feel the scoring is not accurate and we're starting to experiment with Sentiment feedback to see if helping adjusting the score? Is that the only option ?
- While in interaction detail view you can see the sentences that triggered positive or negative feedback there is no export ability of those sentences. Are there any plans to expose those sentences and their scoring in a report?
- What is the process you want us to use to report on incorrect Transcripts. We have quite a few product names that we woudl like to crawl against but it requires to create a Topic for each of them with the different transscript attempts. Example Travelport - our company name - is systematically captured as travel port. So we need to build a topic Travelport - with the word Travel port, travelport and whatever weird combination transcript may reveal. Another example PCC standing for Pseudo City Code is transcripted as suicide code...?
- I will look at API explorer to see if there are ways to pull the sentiment phrases triggered in interactions and export all
We're keen to participate to Beta if still in fly
------------------------------
Renaud Larcier
Travelport
Original Message:
Sent: 10-26-2022 13:31
From: Andrew Boland
Subject: Sentiment Analysis/Speech Analytics
Hi Amber
Thanks for your message.
We actually have a Beta running at the moment, which adds Sentiment data into Topic Trends. If you'd like to be added to that Beta please reach out direct and I can get you on there, it would be great to get your feedback.
Regarding industry data, sadly this isn't data that we keep
Lastly, in the case of incorrect Sentiment. We don't have the ability to remove this Sentiment at the moment, but we do have the ability for Seniment Feedback. Essentially here you're able to go and add a specific phrase that has been picked up incorrectly and change the Sentiment e.g. change it from Negative to Positive
Hope this helps
Andrew
------------------------------
Andrew Boland
Genesys - Employees
Original Message:
Sent: 10-25-2022 08:48
From: Amber Krueger
Subject: Sentiment Analysis/Speech Analytics
We've recently begun using the Sentiment Analysis/ Speech Analytics. A few questions on this…
- Is there a canned report that pulls in Sentiment Score? I was looking through the Reports and didn't see any – but wasn't sure if I missed something.
- Does Genesys have data that can be shared on client service industry averages for sentiment score that we can use to benchmark ourselves?
- If we review a call with low sentiment to find that there's nothing wrong (i.e. the low sentiment is related to mistranslation, personal comments, "oh my god", etc., but nothing to do with our company/service/etc) is there a way to remove the tag and increase the sentiment score of the call?
#QualityManagement
#Reporting/Analytics
#Unsure/Other
------------------------------
Amber Krueger
Ultimus Fund Solutions, LLC
------------------------------