Semantic analysis of Hadith for topic classification using Latent Semantic Indexing (LSI) / Aiman Haziq Ibrahim

The aim of this project is to provide a framework utilizing Latent Semantic Indexing (LSI) to categorize topics in Hadith texts for semantic analysis. Islamic teachings place a high value on the hadith literature, which records the words and deeds of Prophet Muhammad (peace be upon him). To make it simple to access, retrieve, and comprehend pertinent information, Hadith writings must be effectively organized and categorized depending on their topics. The subjectivity, labor-intensive manual categorization, and insufficient capture of semantic links within texts are only a few of the drawbacks of the currently available approaches for Hadith topic classification. To address these challenges, LSI-based framework was proposed that leverages the latent semantic meaning in Hadith texts. LSI captures the underlying semantic relationships between words and enables more accurate topic classification. The research framework consists of six phases, including a preliminary study, requirement analysis, data finding, development, evaluation, and documentation. The data finding involves collecting and preprocessing reliable Hadith datasets. Development focuses on creating an information retrieval system using LSI. The evaluation assesses the system's performance through metrics like cosine similarity, precision, recall, and F1 Score. The experiment assessed the effectiveness of LSI by utilizing ten queries and relevant judgements, precision ranged from 5.4% to 100%, recall from 0% to 65%, yielding an average F1 Score of 19.4%. Finally, documentation encompasses writing a comprehensive report that includes background, methodology, findings, and conclusions.