Ignores capitalization and punctuation
Markov chains are an interesting combination of graph theory and probability. If we look at every pair of letters which appears in a sentence, we can count the number of occurences of each pair and work out some statistics. Here we have a circular graph with edges representing the "tranisition probability". If the letter "e" follows "h" 80% of the time (in a given sentence) while "i" follows "h" only 5% of the time, the next time we see an "h", we are going to bet on the next letter being an "e" over an "i". This graph shows you a real time update of these types of statistics. The grid on the side shows the same information encoded as a matrix. The darker the square, the more often that pair appeared. The matrix represents statistics on letters on the left followed by letters on top.