Thai Monosyllabic Words Recognition using
Ant-Miner Algorithm
Saritchai Predawan1, Chom Kimpan1, and Chai Wutiwiwatchai2
1Faculty of Information Technology, Rangsit University, Thailand
2National Electronics and Computer Technology Center, Ministry of Science and Technology, Thailand
1Faculty of Information Technology, Rangsit University, Thailand
2National Electronics and Computer Technology Center, Ministry of Science and Technology, Thailand
Abstract: In this paper, Ant-Miner software is used to develop classification rules for Thai monosyllabic words. The hypothetical words used in this paper are composed of 65 command monosyllabic Thai words. The binary desired outputs were used during training 520 Thai words consist of 10 numerals and single-syllable, 65 words in each group were used for system evaluation. In order to improve recognition accuracy, initial consonants, vowels, final consonants and tonal level detected were conducted for speech preclassification. The parameters used in the metaheuritstic algorithms are optimized using pruning algorithm with the aim of improving the accuracy by generating minimum number of rule in order to cover more patterns. Thai monosyllabic words recognition using Ant-Miner yielded Thai monosysllabic words accuracy of recognition on test set of 88.65%, 87.69% and 91.54% for 50, 100 and 250 number of ants respectively.
Keywords: Thai monosyllabic words recognition, ant-miner algorithm, classification, Thai language.
Received March 17, 2011; accepted May 24, 2011; published online August 5, 2012