• suffix-array
• acb
• prediction
• compression
• context
• content

Context-based compression methods are the most powerful approaches to squeeze arbitrary textual data. They offer a good predictive model for the subsequent data based on the already seen one, without assuming any probability distribution for the input source.
In this thesis we analyze the adaptive ACB method (Buyanovsky, 94) which is mostly unexplored in the literature, although preliminary results showed compression ratios comparable (or even superior) to the best known data compression utilities.

The novel feature of ACB consists of deploying both the previous context and the subsequent content to find a succinct encoding for the latter one. We perform a large set of experiments to study the experimental behavior of ACB and to compare it with known compressors, thus devising variations of the basic ACB-scheme that result promising for future developments.