GRADED RELEVANCE FEEDBACK IMPLEMENTATION ON INFORMATION RETRIEVAL

• Main Author: - NIM: 13512031 , JONATHAN
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/28306
• Institution: Institut Teknologi Bandung

Currently information searching can be done with the help of technology. The technology that can search information is the information retrieval system. The information retrieval system is expected to provide relevant documents with high precision. This means that as much as possible all documents taken must be relevant documents. Information retrieval systems with relevance feedback have been developed, but many of them only use binary relevance feedback. To improve the NIAP of the system, retrieval system will be made to use graded relevance feedback. <br /> <br /> <br /> <br /> Graded relevance feedback develops the ideas that already used on relevance feedback. To apply this method, it is necessary to change the assessment of relevance judgment that already exists in the document collection. In order for this change in assessment to improve system performance, the assessment will be carried out by more than one person. Development is carried out to accommodate the need for assessment of documents other than relevant documents and relevant documents. <br /> <br /> <br /> <br /> Testing on the application is done to ensure an increase in the NIAP of the system. Based on the test results, the application has produced a higher NIAP value but the categories used are limited to 3 categories. <br /> <br /> <br /> <br /> Based on the experimental results, the implementation of the algorithm shows that addition of categories other than relevant and irrelevant documents can add to the NIAP value of the information retrieval system. Algorithms can still be developed with the addition of new categories in addition to relevant documents, documents that are partly relevant and not relevant documents