Word Prediction

So far, we have discussed smoothing a probability mass using Simple Good-Turing. After computing smooth mass values, we compute the following using them. Our word prediction is associated with the $P(w_1\cdots w_i)$ that maximizes the following equation.

$$P(w_i\vert w_1\cdots w_{i-1}) = \frac{P(w_1\cdots w_{i})}{P(w_1\cdots w_{i-1})}$$ (2)