Automatic Text Summarization for Hadith with Indonesian Text using Bellman-Ford Algorithm

Abstract

Automatic text summarization is one of Natural Language Processing technology to create a summary automatically by not changing the core or main idea of a summarized document. According to the agreement of the majority of Muslim scholars, Hadith is a second source of Muslim's life guideline. This study aims to extract the main meaning of Hadith document using automatic text summarization. The method used in this study is Bellman-Ford Algorithm from graph theory to extract and score the sentence based on the closeness and interconnection between sentences. The Hadith summary which resulted from Bellman-Ford algorithm is evaluated using Recall-Oriented Understudy for Gisting Evaluation - Longest Common Subsequence (ROUGE-L) metrics. Based on experiment using 20 Hadith documents with Indonesian text and ROUGE-L metrics evaluation, the summary results from sentence extraction are influenced by the type of similarity formula between sentences. The result shows that an average value of precision is 46.5%, recall value is 56% recall and f-score 49.5%. The evaluation result is not good enough because the summary result is extraction summary, while the summary evaluation database is abstraction summary. However, from human evaluation, this research contributes to understand the contents of the Hadith with a shorter text but does not eliminate the essence of the Hadith itself.