Arabic Non-Functional Requirements Extraction Using Machine Learning

Amro, Reem; Althunibat, Ahmad; Hawashin, Bilal

doi:10.1109/icit58056.2023.10225951

Cited by 1 publication

(1 citation statement)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…KNN organizes demands into categories according to the degree to which they resemble their immediate surroundings. The purpose of SVMs is to locate the hyperplane that provides the most accurate differentiation between ambiguous and unambiguous criteria (Amro et al, 2023). In contrast to XGBoost, which uses boosting methods to produce a set of weak learners, random forest mixes a large number of decision trees in an attempt to enhance accuracy.…”

Section: Algorithmsmentioning

confidence: 99%

Detecting Ambiguities in Requirement Documents Written in Arabic Using Machine Learning Algorithms

Althunibat,

Alsawareah,

Maidin

et al. 2024

International Journal of Cloud Applications and Computing

Self Cite

View full text Add to dashboard Cite

The identification of ambiguities in Arabic requirement documents plays a crucial role in requirements engineering. This is because the quality of requirements directly impacts the overall success of software development projects. Traditionally, engineers have used manual methods to evaluate requirement quality, leading to a time-consuming and subjective process that is prone to errors. This study explores the use of machine learning algorithms to automate the assessment of requirements expressed in natural language. The study aims to compare various machine learning algorithms according to their abilities in classifying requirements written in Arabic as decision tree. The findings reveal that random forest outperformed all stemmers, achieving an accuracy of 0.95 without employing a stemmer, 0.99 with the ISRI stemmer, and 0.97 with the Arabic light stemmer. These results highlight the robustness and practicality of the random forest algorithm.

show abstract

Section: Algorithmsmentioning

confidence: 99%