An interesting solution to troll comments would be auto-detection via Machine Learning. For a class final some friends and I wrote an algorithm (neural networks

An interesting solution to troll comments would be auto-detection via Machine Learning. For a class final some friends and I wrote an algorithm (neural networks and SVM's) to detect intentionally deceptive comments on hotel-review sites. It worked with about 82% accuracy. I'd imagine something very similar could be done with troll comments.