For mode 1, you have to ergod all the data in the files. So the key point to solve this problem is to save the words.
I realized that using the Dictionary function in C# is very convinient. So I tried to learn it from the internet. As a result, the "Dictionary" has done me a favor.
At the beginning of mode 2, I thought about something i learned from the "Introduction To Big Data " course.
Purify the data, choose what i need.The most frequent 2-word's word gruop must from the most frequent words.
Since i have got the frequenct of all words from mode1, maybe i can finish mode 2 by this idea.
BUt I have to ergod all the text again and againg to analyse the most frequent words to the least.This must cost me a lot of time!
The idea from mode 1 can easily be transfered to that adapt to mode 2, and the fact that the time it costs is just a lit bit more than mode 1.
So i abandoned my first thinking,chose the method of mode 1.