Graph-based ambiguity handling in the chain of narrators for hadith information retrieval / Nursyahidah Alias

Information Retrieval (IR) is essential in retrieving information automatically and quickly. Therefore, a wide range domain area applies, and testing the IR models with a standard dataset is provided. Besides, much attention is given to hadith data sets in retrieving information automatically. Hadith is divided into a sanad and matan. These parts are the essential factors in hadith. Sanad refers to a chain of narrators. A degree of authenticity in hadith can be determined by the chain of narrators which are: sahih (sound), hasan (good), and da’if (weak). Hadith IR focuses primarily on the matan part, which retrieves hadith based on a topic that is more appropriate with the existing traditional techniques in IR, which are terms and frequency. However, current research in the hadith IR working with the sanad part is in indexing individual narrator names in the chain of narrators. The indexing could be used in retrieving hadith documents. However, the document cannot be retrieved if a user needs to retrieve the hadith according to the authenticity determined by the chain of narrators, not the individual narrator. Therefore, this thesis concerns hadith document retrieval based on a chain of narrators. Three main studies were conducted, each with a different and specific aim. The first was a study conducted by a literature review of traditional indexing and retrieval techniques in IR and hadith IR, the chain of narrators, an ambiguous element in the chain of narrators, and existing NLP tools for Malay language and hadith IR. The first study's findings led us to draw a conceptual model of hadith IR that handles ambiguous elements in the chain of narrators based on a graph. The second study aims to design a graph-based indexer and rule-based hadith IR. The findings in the second study resulted in a test collection, indexer, and retrieval design based on 1000 hadith from the Shahih Bukhari Book validated by hadith experts.