
		<paper>
			<loc>https://jjcit.org/paper/273</loc>
			<title>LIGHT-WEIGHT, SEMI CONTEXT-FREE, RULE-BASED ARABIC TEXT CLASSIFIER FOR POS TAGGING</title>
			<doi>10.5455/jjcit.71-1749752027</doi>
			<authors>Bilal Alqduah,Mohanad Alhasanat,Abdullah Alhasanat,Hatem Alqudah</authors>
			<keywords>Part-of-speech (POS) tagging,Arabic rule-based classifiers,Natural languages,Context-free grammars</keywords>
			<views>2186</views>
			<downloads>912</downloads>
			<received_date>16-Jun.-2025</received_date>
			<revised_date>  12-Jul.-2025 and 12-Aug.-2025</revised_date>
			<accepted_date>  13-Sep.-2025</accepted_date>
			<abstract>In this research, we address the challenges associated with part-of-speech (POS) tagging and morphological 
classification of Arabic text where word structure is the subject of study.. Our focus is on Classical Arabic (CA) 
and Modern Standard Arabic (MSA), where the text is typically vocalized and includes diacritics on most letters. 
Our proposed classification method does not require a lexicon, stemming processes, or artificial intelligence 
techniques. The goal is to minimize the resources needed for classifying Arabic text. This method is based on the 
principle that each verb in the Arabic language adheres to a specific pattern, we refere to as (wazn   نزو or  tafʿīl 
ليعفت), that can be utilized to identify a word. The classification process is governed by a finite state machine, 
which is translated into regular expressions. Each verb tense is represented by a set of regular expressions (REs). 
The order in which these regular expressions are processed is crucial for the accuracy of the results. Whenever a 
match is found, the word is marked to prevent further matches. The proposed method is lightweight and functions 
as a best-effort classifier, assigning the closest match as a tag. In terms of performance, the proposed classifier's 
execution time is linear and does not require high processing capabilities.</abstract>
		</paper>


