Compression of dictionaries via extensions to front coding

Front-coding is a technique used to reduce the redundancy in a representation of a dictionary, taking advantage of common prefixes. However, redundancy still exists in the front-coded representation; suffixes and infixes of words are not coded. The authors method attempts to remedy this deficiency by iteratively applying front-coding techniques to the suffixes. By applying a variant Huffman coding method, it is possible to represent the Huffman tree of suffixes in the form of another dictionary, to which the method can be iteratively applied. On large natural-language dictionaries the authors have achieved compression ratios as favourable as 11%.<<ETX>>