Network Analytics meets Text Processing

This workflow combines two pre-processing techniques: network analytics and text processing.The goal of the text processing part is to identify the general mood of a user e.g. positive, negative or neutral based on the sentiment of its posts and comments. The goal of the network analytics part is to compute the social status e.g. leader or follower of a user in the forum community. The visualization part condenses the two result sets into a single scatterplot that visualizes the social status as well as the general mood of the users for the slashdot data set.

This is a companion discussion topic for the original entry at

Hi @Paolotamag
I have a confusion regarding the use of numeric binner to assign respective sentiment to the text. As I can see the upper bound is 27.438 and Lower bound is -14.356, I would like to learn how the category of the bin is to be decided. In your case, Negative sentiment is from -infinity to -17.0, Neutral is from -17 to 33 and Positive is from 33 to +infinity.
I would like to know, how to decide the width for each category. I have tried auto binning but the data always stick more towards neutral sentiment, all the time.
Your suggestion will be much appreciated.
Deep Gurung