Build a next-word frequency predictor

You are given a sequence of tokens (words) representing a corpus, e.g.:

tokens = [w0, w1, w2, ..., wK-1]

You need to build a next-word predictor that supports queries of the form:

• predict(word) -> next_word

Where next_word is the word that most frequently follows word in the corpus.

More precisely:

• For every adjacent pair (tokens[i], tokens[i+1]) , count it as one occurrence of “ tokens[i+1] follows tokens[i] ”.
• For a query word = x , return the y that maximizes count(x, y) .
• If x never appears or never has a following word (e.g., it only appears as the last token), return a sentinel such as null / empty string.
• If there is a tie for max frequency, you may return any tied word (or specify a deterministic tie-break, e.g., lexicographically smallest).

Task

Design the data structures and implement:

1. A one-time build(tokens) preprocessing step.
2. A predict(word) query.

Constraints / expectations

• The corpus is processed once (batch/offline build), then many queries may follow.
• Aim for efficient preprocessing and fast queries.