"Positve - Neutral - Negative Sentiment Analysis"

Legacy User

Folks,

This is a open question to all - Has anybody created a model that they would share in this forum for executing sentiment analysis on text data so that a large volume of info can be classified as positive, negative or neutral?

Many thanks!

earmijo

Download the RapidMiner-Text Add In. You can go through some of the examples that come with the add-in. I remember one of them is exactly what you want.

IngoRM

Hello,

I think newstuff is more interested in the final model than in the model building itself, right?

I am afraid that such a sentiment model is not freely available and I since the data collection and manual labeling is probably the hardest part I also doubt that anyone is going to share this for free.

By the way, we provide the prediction of sentiments as a Web Service within our new product RapidSentilyzer and you can contact us if you are interested at contact@rapid-i.com

Cheers,
Ingo

Legacy User

Thanks to all!  Ingo is right.  At this stage, I am looking for a working model.  

I have the plugin and been through the examples and have cobbled together a rudimentary model.  But, it is not clear to me that I am truly building my sentiment analysis experiment to a successful and useful end-point.  I have an Excel sheet that contains the comments that I want to classify as positive, neutral, or negative.  My goal is to do this classification within RM.  I have done a great deal of manual coding of user comments and it is very time consuming and labour intensive.  I agree that the data is definitely proprietary; however, my hope is that the framework, in the form of a RM experiment, for producing a sentiment analysis may live somewhere within the RM open source community.

Thanks!