Data compression using surrounding contexts

In this paper, we present a simple way to estimate probabilities of symbols occurring in a given sequence by considering the context where they appear. Every statistical adaptive compression method can use those estimation to encode a given sequence. The main goal is to find convenient estimation models which assign to every symbol in the sequence a probability as large as possible and still avoid decoding ambiguities. Beside the pragmatic purpose of data compression to make data storage and transmission more efficient, there is another interesting aspect of data compression. An efficient method to represent data is an intrinsic efficient method of information analysis and prediction.