Update dataset usage and to version 0.11 for Sentiment Analysis

[JRAlexander]

Revising with UCI Sentiment Dataset for greater model accuracy, and updating to 0.11.0.

[prathyusha12345]

@JRAlexander I tested the PR.working fine. But I have questions on the output.
1.The accuracy metrics is 79%. - I believe accuracy should be more than 85%.Correct me if I am wrong.
2. See the probabilities in below image. Howcome a text with probability <50% is showing prediction correctly?

I believe that probability also should be > 50% to get prediction correctly. Correct me if I am wrong.

@CESARDELATORRE Adding Cesar for reference.

[JRAlexander]

It leaves the door open for improving the model. This is the dataset we are using. It has better accuracy than using wikidetox.

[CESARDELATORRE]

Well, a rule of thumb is to have an accuracy higher than 80%. However, depending on the business/domain it might need to be a lot higher or even be acceptable to be lower.

When using small datasets it is normal in most cases that the accuracy won't be very super high, neither the probabilities.

Since we're using small datasets for short trainings, I'd suggest not to show the probability in the code or output..

[JRAlexander]

I'm going to leave it in there with an explanation about the small data

[luisquintanilla]

Looks good @JRAlexander. Added a suggestion. I'll leave the decision to implement or not up to you.

[JRAlexander]

Added your suggestion as issue here dotnet/docs#11221

[JRAlexander]

Thanks, @luisquintanilla!