Lempel-Ziv [1977], or LZ77 is an adaptive dictionary-based compression algorithm. LZ77 uses previously seen text to build a dictionary. The dictionary consists of all the strings in a window into the previously read input stream. 

Its main data structure is a text window, divided into two parts. The first part is a large block of decoded text held in a fixed-size window and the second part is a look-ahead buffer which has characters read in from the input but not yet encoded.  

Symbols within the buffer are then compared with data in the fixed-size window. The algorithm tries to match the contents of the look-ahead buffer to a string in the fixed-size window. 

In general, dictionary-based compression replaces phrases with tokens. If the number of bits in the token is less than the number of bits in the phrase, compression will occur.

---