Your thoughts: using Microsoft Logistic Regression Algorith for sentiment analysis?
-
Thursday, January 03, 2013 5:52 PM
Consider a dummy data-set:
Text | Sentiment
Wow That is amazing | Positive
That was worst! | Negative.
Can we build a sentiment classifier using Logistic regression algorithm?
Are there other MS Data Mining algo's that I should consider?
Your thoughts and comments are appreciated!
Paras Doshi (Blog: ParasDoshi.com | Twitter: @Paras_Doshi )
All Replies
-
Monday, January 07, 2013 10:20 PMAnswerer
You cannot build good sentiment analysis system simply by extracting different terms from the sentence and applying one of the data mining algorithms in SQL Server. You have to look at the whole sentence to understand if it is positive or negative. Examples:
The product is good
The product is not good.
The product does not look very good.
No one thinks that the product is good.
The product is not only good but amazing.
The product provides value.
The product provides little value.
Tatyana Yakushev [PredixionSoftware.com]
-
Monday, January 07, 2013 11:24 PM
Thanks for the reply.
Sorry for not being clear. Yes, I meant using the entire sentence to create sentiment classifier.
Paras Doshi (Blog: ParasDoshi.com | Twitter: @Paras_Doshi )
.png)
