Statistical Properties of Ordered Alphabetical Coding

摘要

The paper presents a type of text coding, called αβ-coding. The essence of αβ-coding is that letters of every word of a given text are arranged in a specific way to create a code of that word. List of codes obtained by scanning text corpora is stored in a database together with words that could be transformed into each code. Word frequencies are stored as well. Decoding is performed by transforming possibly scrambled words according to the algorithm of the coding and finding in the database the most frequent word corresponding to the resulting code. As more than one word may result in the same code, decoding is inherently ambiguous. However a study on corpora of five languages has shown that about 95% of word-tokens can be correctly decoded.