|Budget Amount *help
¥2,100,000 (Direct Cost : ¥2,100,000)
Fiscal Year 1994 : ¥700,000 (Direct Cost : ¥700,000)
Fiscal Year 1993 : ¥700,000 (Direct Cost : ¥700,000)
Fiscal Year 1992 : ¥700,000 (Direct Cost : ¥700,000)
In natural language processing, it needs to build a huge size dictionary. Thus, a fast and compact date stracture is required to develope practical natural language processing syatems : a automatic sentence, or document, a generation system, an automatic proof reading system, a machine translation system, a word processing system, and so on. A trie structure is a good technique that can search the longest applicable key and all prefix keys for a given string, but it takes a huge space for a large set of keys.
This project porposed a compaction algorithm of trie structures and a fast retrieval algorithm even if a large set of keys. The approach proposed in this project is analyzed by theoretical evaluation and the evaluation is supported by a variety of sets of keys : EDR Japanese dictionaries, English dictionaries and compound words.
The idea presented can apply to dictionaries of many natural language processing systems, so a more practical evaluation should be done in the future research.