Google offers a service known as ngrams that gives access to each time a word or phrase is used in a certain year in a certain amount of books from all books every written (to the extend that they have). They offer the raw data for download so what I did was make a c++ program that parses and analyzes this data to find the most common word of phrase every out of all books.
I made a c++ program that looks through a text file named data.txt and finds the most common x word phrase where x is a chosen number. In the screenshot I used the most common 3 word phrase and listed the top 20 (I can also list the least commonly used phrases) from a file that contained a bunch of wikipedia articles that I crawled using a java web crawler I wrote. This project was particularly difficult to optimize. The first version of the program was very very slow, but I soon used lots of pointers, lots of optimization techniques I thought of, and clever memory management, the final version is lightning fast compared to its predecessor.
I made a C++ program that essentially looks for patterns in text, and generates text based off what it finds. It finds common words that follow each word and then generates text picking common words that follow each previous word.
I wrote a BigNumber class in c++. This project taught me so much about polymorphism and object orientation in c++. The class I wrote allows for the addition and subtraction of 2 numbers of any size. I used this class to generate very large Fibonacci numbers.