Cross-Lingual Semantic Similarity in Pieces of Al-Quran Verses Translation Using Word Alignment and Semantic Vector Approach

Abstract

Al-Quran is a holy book of Muslims, that serves as a guide of life. Therefore, it is very important to understand the meaning of the Qur'an verse. The understanding is not only in the Qur'an in a single translation language but also in different translation languages. This research proposed the solutions to these problems with the analysis and implementation of Cross-Lingual Seman- tic Similarity on the translation piece of Al-Qur'an verse using Word alignment and Semantic Vector approach. The experimental result shows that the correla- tion value using combination between Word Alignment and Semantic Vector method is 0,62007 while the correlation value using Word Alignment method is 0,63231, and Semantic Vector method is 0,34841. This value was obtained by comparative value in the form of gold standard, namely the semantic similarity on the pair of translation pieces on the verses of Al-Quran in Indonesian and English based on human assessment manually. The correlation value is quite low due to the translation of Indonesian Al-Quran verses by using google trans- late feature before the semantic similarity calculation process produces many results that do not match the sentence of origin that is translated. Therefore the semantic value is much different from the gold standard.