/Sentinet

Sentiment Analysis using neural networks

Primary LanguagePython

SENTINET

Sentinet is a sentiment analysis command line app using Neural Networks to track sentiment.

This code is being developed by Todd Ervin at the University of North Carolina Asheville. This application is in development. Please be aware of this as you look at the code.

ABSTRACT

Todd Ervin and Charley Sheaffer (Mentor), Computer Science

Sentiment analysis is the attempt to extract subjective experience from text using Natural Language Processing, Computational Linguistics and Text Analytics. Artificial Neural Networks (ANN) are modeled after Biological Neural Networks and mimic the information processing capabilities of the brain. ANN are ideal for detecting and finding hidden patterns in a large data set. Twitter generates massive amounts of natural language text data in real time and is ideal for sentiment analysis. This research will use an ANN to try and determine if a given tweet is “Happy” or “Sad”. The research will have two phases. The first phase will train the ANN with emoticon tagged tweets and the AFINN word list (an affective tagged word list) to establish possible sentiment. The data set will be generated from the Twitter streaming API and will be bundled into “data snapshots” consisting of 20-25,000 tweets forming a larger dataset of 500,000 tweets. The second phase will use the trained ANN to detect sentiment in individual tweets from trending hashtags in real time and scoring the tweets for emotional content. The accuracy will be verified by sampling the phase two data set and scoring the tweets by humans to establish the accuracy of the ANN.

REMINDER: This app is under construction and exists in a constant state of development