Sentiment Analysis using Support Vector Machines in Ruby
summary
Uses sparse word vectors from the imdb corpus, trains an SVM with an RBF kernel and tunes the C parameter by cross‑validation, and wires it into a small Sinatra app that stores features in redis and calls libsvm gym bindings. Deployed on Heroku with a tiny Redis instance, it reports about 95% positive 65% negative accuracy on movie reviews, showing the strengths and the bias of this approach rather than a silver‑bullet solution.
AI-written summary · about
chapters (11)
- 0:00Matt Kirk Sentiment Analysis using Support Vector Machines in Ruby
- 1:47The Challenge
- 3:31Ruby is not Complex Math
- 7:17Step 1: Collect Data
- 8:18The Features of a Chess Piece
- 10:39Features of Text
- 11:19We will use words only
- 13:28Linear Regression has a Curse
- 15:34But there's a trick (The Kernel Trick)
- 16:18Parameter C
- 23:18Conclusion
related talks
46mPractical Machine Learning and Rails
51mMachine Learning with Ruby
35mIntelligent Ruby: Getting Started with Machine Learning
01h 23mBeyond Sentiment Emotion Mining with Python and machine learning
39mSentiment Analysis, its techniques and applications
28mMachine Learning for Fun and Profit
53mA Machine Learning mood-o-meter
01h 22mWhen Rotten Tomatoes Isn’t Enough: Analyzing Twitter Movie Reviews