Lexical scoring system of lexical chain for Quranic document retrieval

Abstract

An Information Retrieval (IR) system aims to extract information based on a query made by a user on a particular subject from an extensive collection of text. IR is a process through which information is retrieved by submitting a query by a user in the form of keywords or to match words. In the Al-Quran, verses of the same or comparable topics are scattered throughout the text in different chapters, and it is therefore difficult for users to remember the many keywords of the verses. Therefore, in such situations, retrieving information using semantically related words is useful. In well-composed documents, the semantic integrity of the text (coherence) exists between the words. Lexical cohesion is the results of chains of related words that contribute to the continuity of the lexical meaning found within the text are a direct result of text being about the same thing (i.e. topic, etc.). This indicates that using an IR system and lexical chains are a useful and appropriate method for representing documents with concepts rather than using terms in order to have successful retrieval based on semantic relations. Therefore, a new Lexical Scoring System is proposed in this study, in addition to determining the semantic relation that exists between words whereby WordNet was used as the semantic knowledge base. The proposed scoring system helped to retrieve 86.58% of the total relevant documents in the Al-Quran based on the relevance judgment, using the lexical chain approach. Based on the findings, the study concludes that, the proposed approach on representing verses using lexical chains is appropriate and suitable for a Quranic IR system.