Efficient algorithm for learning simple regular expressions from noisy examples