Review of Text Clustering Methods and Suggested Solutions for Theme Based Clustering of the Quran

Published: 22 January 2020| Version 1 | DOI: 10.17632/kb92kdjtcz.1
Contributors:
rosalina abdul salam,
qusay bsoul,
Jaffar Atwan,
Hishomudin Ahmad

Description

In the datasets, documents of modern, unedited, and unmarked Arabic texts were utilised, which consisted of a sample of nearly 1,680 documents obtained from various online Arabic resources. The testing dataset comprised of four fields, namely: art, economics, politics, and sports articles. The other dataset collection consisted of 383,872 Arabic documents, which were primarily newswire dispatches as released by the Agency France Press (AFP) between years 1994 and 2000. Standard TREC classes and ground truth were thus established for this collection, whereby 10 classes were thus classified as part of TREC 2001. The last datasets is the Qur'an data, which converted from softcopy to database to contactes with each chapters and verses.

Files

Categories

Feature Selection, Information Classification, Arabic Language, Extraction Methods, Cluster Testing, Classifier Evaluation, Cross-Language Information Retrieval

License