Development of Yes/No Arabic Question Answering System
This addresses the challenge of Arabic question answering, which is an incremental improvement for natural language processing applications in Arabic-speaking contexts.
The researchers developed a semantic-based open domain yes/no Arabic question answering system using paragraph retrieval with variable length, achieving improved precision through semantic similarities and query expansion.
Developing Question Answering systems has been one of the important research issues because it requires insights from a variety of disciplines,including,Artificial Intelligence,Information Retrieval, Information Extraction,Natural Language Processing, and Psychology.In this paper we realize a formal model for a lightweight semantic based open domain yes/no Arabic question answering system based on paragraph retrieval with variable length. We propose a constrained semantic representation. Using an explicit unification framework based on semantic similarities and query expansion synonyms and antonyms.This frequently improves the precision of the system. Employing the passage retrieval system achieves a better precision by retrieving more paragraphs that contain relevant answers to the question; It significantly reduces the amount of text to be processed by the system.