<![CDATA[With the increasing number of electronic text documents, the process of searching and processing information has become increasingly complex, especially when these documents come from multiple sources and languages. Consequently, document summarization methods are needed to help users retrieve important information more quickly. However, existing multilingual summarization methods, such as ELSA, are limited by dataset size and the need to pre-determine themes. By integrating the Bag of Itemset representation and the Latent Dirichlet Allocatio Algorithm Modification (LDA-AM) approach, this study aims to improve the quality of multilingual document summarization. The proposed method first uses topic modeling to divide different multilingual documents into several topics. Then, for each topic, a sentence selection process is performed to generate topic-based summaries, which are then combined into a general summary. Using the ROUGE evaluation metric, experiments were conducted to compare the proposed method with baseline. Experimental results show that the proposed method performs better than ROUGE-1 with a value of 0.2623, ROUGE-2 with a value of 0.1802, and ROUGE-L with a value of 0.1231. The results indicate that in the process of summarizing multilingual documents, summary quality can be improved by combining the Bag of Itemset representation and LDA-AM.]]>
Copyrights © 2026
