A compression method using link-trie structure for natural language dictionaries

A trie structure is a key search method that is frequently used in the field of natural language systems and database systems. Morita presented an efficient data structure for collocation information using trie structure. This structure stores two basic words into a trie and defines link information by a link function. This paper presents how to apply Morita's method into general key-set and compress trie. Presented method divides a key into several parts and defines link information between keys. From simulation results, it turned out that presented method is 30% smaller than Conventional method in space capacity.