High Speed Korean Morphological Analysis based on Adjacency Condition Check

Kwangseob Shim

School of Computer Science and Engineering
Sungshin Women's University
Seoul 136-742, KOREA

Jaehyung Yang

School of Computer and Media Engineering
Kangnam University
Kyungki-do, KOREA

appeared in: Journal of the Korea Information Science Society, 31(1) : 89-99, 2004. (in Korean)

Abstract

This paper proposes a morphological analysis method that enables morphological analysis by checking conditions between two adjacent morphemes. These conditions are fed from a dictionary. This method eliminates a code conversion module and the application of transformational rules for candidate generation. The method claims that very high speed morphological analysis is attainable through simple bit operations for adjacency condition check. MACH, an implementation of the proposed method, is a supersonic Korean morphological analyzer which is able to analyze a document of 1 GB in 5 minutes on a PC with 1.13 GHz Pentium III CPU. The analysis accuracy of MACH is 99.2 %.

Back to the Publications.