Compression by induction of hierarchical grammars

This paper describes a technique that develops models of symbol sequences in the form of small, human-readable, hierarchical grammars. The grammars are both semantically plausible and compact. The technique can induce structure from a variety of different kinds of sequence, and examples are given of models derived from English text, C source code and a file of numeric data. This paper explains the grammatical induction technique, demonstrates its application to three very different sequences, evaluates its compression performance, and concludes by briefly discussing its use as method of knowledge acquisition