Penerapan ECS Stemmer untuk Modifikasi Nazief & Adriani Berbahasa Jawa

Abstract

Stemming Javanese affix words using Nazief & Adriani modifications still has problems that cannot be solved such as overstemming, understemming, and unchange. Then there needs to be improvements to improve the performance of Nazief & Adriani modifications. This study aims to improve the performance of Nazief & Adriani modifications using the Enahnced Confix Stripping (ECS) modification method. The results of this study indicate that Enhanced Confix Stripping can improve performance that previously had an accuracy of only 78.2% to 97.9% with an error rate of 2.1%. And fixing errors that originally numbered 98 to 9 errors. But Enhanced Confix Stripping still has problems with the words "ngetan, kumanggah, kumarut, kumasis, kumareg, kumadul, kumaras, katawakake, and pangenan". The next research is expected to be able to solve this problem.

Stemming Bahasa Jawa Menggunakan Damerau Levenshtein Distance (Dld)

“…Metode Nazief dan Adriani dimodifikasi sesuai aturan morfologi bahasa agar dapat digunakan untuk Stemming teks dalam bahasa Jawa [15]. ECS stemmer juga pernah digunakan untuk menghasilkan kata dasar Jawa [2], [16]. Metode-metode ini berbasis bentuk kebahasaan yang sudah umum digunakan.…”

Sundanese Stemming using Syllable Pattern

“…Therefore, this research is driven by the lack of corpus data availability and development activities in software tools used for various applications, especially those related to the Sundanese language. However, recent studies that discuss stemming techniques for regional languages have begun such as Javanese [3] [4], Madura [5], Balinese [6] [7], and also Sundanese [8][9][10].…”

“…The stemming technique in previous research was developed in morphology form including derivation and inflectional affixes removal for regional language [3][4][5][6][7][8][9][10], Indonesian [11] [12], and English [13]. Several stemming methods used in previous research are rule-based [3][10] [13], the combination of rule-based with corpus or dictionary [4][5][6][7][8][9][10][11][12], also rule-based with n-gram [7]. The rule-based technique adopted from Porter's algorithm focuses on the process to remove the suffix to the right of the root word [13].…”

“…This method is widely applied to stemming development for Indonesian and later adapted to stemming for regional languages. Another technique was developed which applied to several regional languages stemming using confix stripping [11] and enhanced [4] [5] for confix stripping (ECS) method. Although the ECS technique has good stemming results for Indonesian, but not for regional languages.…”

“…In addition, the difference of morphological patterns in regional languages has many differences compared with Indonesian or English language such as derived words constructed by the combination of prefixes, infixes, suffixes, or confixes [9] [10]. So, the application of the rule-based method becomes ineffective for some cases especially if it depends on the completeness of the dictionary database [4][5][6][7][8][9][10][11][12] because there are even some words that have no equivalent in Indonesian [15]. To overcome this problem, the research uses a different rule-based approach by utilizing syllable patterns that form derived words in the Sundanese language.…”