Re:VaderSharp – Sentiment analysis using Vader with C#
by com2kid on Friday June 09, 2017 @03:04PM
(#14523600)
This is crucial. There are a LOT of companies selling social media sentiment analysis tooling, mostly wrapped up in web apps and large monthly subscription fees.
When I've seen teams do manual classification to test out the accuracy of their tools, it has turned out that the accuracy is generally all around horrible, most tools heavily bias towards "neutral" which really means "not sure". Not too surprising, tweets doesn't provide much context.
Re:VaderSharp – Sentiment analysis using Vader with C#
by erikb on Friday June 09, 2017 @06:50PM
(#14525166)
The title seems to have changed from something like "now it is rewritten from Python to C#". I now suppose all the downvotes of my comment happened after the retitling.
Re:VaderSharp – Sentiment analysis using Vader with C#
by erikb on Saturday June 10, 2017 @03:42AM
(#14526517)
It's like investing energy to switch from "My app runs on Kubernetes" to "My app runs on Windows XP" and being proud about it for unexplainable reasons.
Re:VaderSharp – Sentiment analysis using Vader with C#
by emodendroket on Saturday June 10, 2017 @08:27AM
(#14527092)
It's actually not like that at all since C# is a lively, rapidly evolving platform and also probably friendlier to a large codebase than Python (and almost certainly more performant).
When I've seen teams do manual classification to test out the accuracy of their tools, it has turned out that the accuracy is generally all around horrible, most tools heavily bias towards "neutral" which really means "not sure". Not too surprising, tweets doesn't provide much context.
Also all the capitalization logic has meaning only in languages that use a bicameral script.