Overview

Socialmeter is a flexible, extensible sentiment analysis framework.

It is built to provide a baseline framework for executing sentiment analysis, while maintaining a level of customization that allows tuning to achieve more accurate classification possible.

All modules are built to be flexible to the needs of the user, and are built with this in mind. The class hierarchy is designed to allow users to drop-in their own modules that extend the baseline modules.

Usage

(Currently under construction)

Future Improvements

Concrete Tasks

• Further constriction of the SMeter class definition
  • Removal of the Module class
  • Removal and replacement of the Link class for more specific use cases
• Unit testing of feature extractors and preprocessor extractors
• Better ensemble support
• Different SMeter objects that all feed into one larger SMeter objects
• Better Facebook Graph API implementation and support
• Option for preprocessors to 'replace' the text attribute for data
• Better abstraction for column format in conjunction with the SMeter object and input modules
• Built-in powerset grid search for feature extractors
• Remove FeatureExtractorModule class and only use the FeatureExtractor objects handled by a single class.

Directional Improvements

• What do Data Scientists think of this?
• What do Programmers think of this?
• Do any other sites provide a service like the TwitterStreamAPI?

References

Here are some sources used to build this project, as well as some references you may want to use when building your own sentiment analysis project.

Academic

Twitter Sentiment Classification using Distant Supervision - Alec Go, Richa Bhayani, Lei Huang (2009) http://www-cs.stanford.edu/people/alecmgo/papers/TwitterDistantSupervision09.pdf

Twitter as a Corpus for Sentiment Analysis and Opinion Mining - Alexander Pak, Patrick Paroubek (2010) https://pdfs.semanticscholar.org/ad8a/7f620a57478ff70045f97abc7aec9687ccbd.pdf

Semantic Sentiment Analysis of Twitter - Hassan Saif, Yulan He, Harith Alani (2012) http://oro.open.ac.uk/34929/1/76490497.pdf

Sentiment-Analysis-Twitter - Ayush Pareek (2016) https://github.com/ayushoriginal/Sentiment-Analysis-Twitter

Non-Academic

GeeksforGeeks "Twitter Sentiment Analysis using Python" http://www.geeksforgeeks.org/twitter-sentiment-analysis-using-python/

Mining Twitter Data with Python (Part 6 – Sentiment Analysis Basics) - Marco Bonzanini https://marcobonzanini.com/2015/05/17/mining-twitter-data-with-python-part-6-sentiment-analysis-basics/

Published

Natural Language Processing with Python - Steven Bird, Ewan Klein, Edward Loper http://victoria.lviv.ua/html/fl5/NaturalLanguageProcessingWithPython.pdf

Datasets

W3C Wiki for Sentiment Analysis Datasets https://www.w3.org/community/sentiment/wiki/Datasets#Emoticon_Sentiment_Lexicon

Twitter Sentiment Analysis Training Corpus (Dataset) http://thinknook.com/twitter-sentiment-analysis-training-corpus-dataset-2012-09-22/

License

SocialMeter is distributed under the terms of both

• MIT License https://choosealicense.com/licenses/mit
• Apache License, Version 2.0 https://choosealicense.com/licenses/apache-2.0

at your option.