Temu Kembali Informasi Dokumen XML dengan Pembobotan per Konteks

Abstract

Various research in the field of information retrieval using XML documents have been conducted.However, those systems didnot consider the structure or tags of the XML documents. To improve the performance of the information retrieval system, we develop an information retrieval system using context weighting. Each weight which is implemented on each context is obtained from the result of correlation between the similarity of each context of the document to the queries and the similarity of the whole document to the queries using linear regression method. The information retrieval system that has been developed using the context weighting is evaluated by using 30 queries. It is found that this experiment decreases the average Mean Average Precision (MAP) value by 1.15% if compared with the information retrieval system without context.