Given a massive, continuous stream of text data, design an algorithm to efficiently track the top K most frequent tokens in real time.
