Content-Based Document Retrieval Using Graph-Based Approach

Abstract

The purposes of the research were to design method for content-based document retrieval using graph-based approach and to test the effectiveness of the content-based document retrieval using graph-based approach. Experimental data was electronic books which database design subject.

The results revealed that the proposed method consists of four major steps; 1) text processing, 2) indexing, 3) similarity measure and 4) document ranking. The experiment showed that using Jaccard’s Coefficient with degree of document tree to measure similarity of documents can rank the query result more efficient than only using Jaccard’s Coefficient. Furthermore, this proposed method was good average precision (0.75), average recall (1.00) and average F-measure (0.846).  This results of the study provided the proposed method could be used in content-based document retrieval effectively.