|
If the least frequent "slot" in your search (the "slot" whose combined frequency is the least in the corpus) has a frequency of 40,000,000 or more,
then we use pre-calculated "n-grams" tables rather than searching the corpus itself. These n-grams tables contain the top 10,000,000 strings in the corpus for each of
the two word strings (2-grams), 3-grams, 4-grams, and 5-grams. But because it's just the top 10,000,000 n-grams, it may not
contain a full listing of matching strings. |