One technique used for predicting words in auto word suggestion software
is Kneser-Ney smoothing,
which calculates the probabilities of n-grams given an initial training body of
text. It uses absolute discounting by substracting some discount delta from
the probability's lower order to filter out less frequent n-grams.
This is a PyQt application that demonstrates the use of Kneser-Ney in the context of word suggestion. As words are typed into the text area, up to five of the most probable continuing words are suggested at the top. Only bigrams are considered, although Kneser-Ney could be extended to n-grams with a recursive Kneser-Ney formula.
For example, this was trained on the NLTK Reuters corpora.
python3 src/app.py