Title: Integrating clustering method with graph-based ranking for Hausa text multi-document summarisation

Authors: Abdulkadir Abubakar Bichi; Ruhaidah Samsudin; Rohayanti Hassan

Addresses: Faculty of Computing, Yusuf Maitama Sule University, Kano, Nigeria ' School of Computing, Universiti Teknologi Malaysia, Malaysia ' School of Computing, Universiti Teknologi Malaysia, Malaysia

Abstract: Automatic text summarisation is one of the promising solutions to tackle the ever-growing amounts of textual data. It produces shorter version of the original document with less bytes but same information as the original document. Despite the advancement in automatic summarisation research, researches involving the development of summary extraction method for Hausa text are still at early stage. This study proposes an extractive multi-document summarisation method for Hausa text by adapting the existing HauRank, proposed earlier for Hausa single-document summary extraction. The HauRank ranking algorithm is modified by replacing the document level adjacency matrix W with the cluster level adjacency matrix WCi. The method minimises the redundancies associated with multi-document summarisation by first clustering the documents sentences according to the document's subthemes. And the modified ranking algorithm local HauRank is applies on individual clusters to determine the most salient sentence in each cluster. The results of Rouge-L simulation metrics show that the proposed method outperforms both LexRank and MEAD methods with respective 0.0133 and 0.0142.

Keywords: multi-document summarisation; Hausa summarisation; graph-based method; extractive summarisation; document clustering.

DOI: 10.1504/IJSSS.2024.140475

International Journal of Society Systems Science, 2024 Vol.15 No.1, pp.59 - 72

Received: 28 Apr 2023
Accepted: 06 Apr 2024
Published online: 19 Aug 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article