Qur'anic words stemming

Abstract EN

Arabic words are known to have complex morphological structure.

The different structures produce various word patterns or derivatives from a root word.

This paper attempts to identify various word patterns that originate from a root word.

These word patterns are compared to the words in the 30th part of the Qur'an.

Nine stemming test cases were outlined for words in the 30th part of the Qur’an.

Analysis showed that stemming nouns and particles leads to a lower percentage error compared to stemming the 10 alphabets that can be added as affixes in a root word.

A rule based stemming engine (RSE) was also implemented and the stemming accuracy achieved was 62.5 % and the average time taken to stem 1000 word tokens was 11.7ms.

The accuracy of the results was comparable to other stemming engines such as the Khoja stemmer, Buck Walter Morphological Analyzer (BAMA), Tri-literal Root Extraction (TRE) algorithm, and Voting algorithm.