Derivational and Inflectional Stemming Approach for the Filipino Part-of-Speech Tagging
Great Allan M. Ong · International Journal for Research in Applied Science and Engineering Technology · 2018
This paper aims to present an increase in the accuracy of the existing Tagalog part-of-speech tagging (TPOST) by introducing derivational and inflectional stemming approach through hybrid and modified stemming algorithm. A collected 89 derivational and inflectional assimilatory common words and 22 basicFilipino region names and languages for the KSTEM algorithm are utilized for the initial stemming process.The assimilatory and partial duplication of assimilatory word rules were modified using the same approach.Since TPOST showed a high percentage on stemming errors and wrong feature extraction in the part-of-speech tagging, the evaluation was focused on strengthening this methods.For the entire stemming testing sets, a 17.73% decrease in under stemming index and a 0.00967 over stemming index was produced.A total of 2.32 % and 3.42% assimilation word feature success rate was produced in the combined test sets for the part-of-speech tagging.The success of hybrid stemming relies on the assimilatory word search, therefore a trained data was listed and evaluated and produced a 9.39% assimilatory word features success rate.Despite of these variations, an innate morphological study by stemming modification and KSTEM strengthening focusing on the old Filipino assimilation of word and diversified samples are recommended.