A novel semantic and logical-based approach integrating RTE technique in the Arabic question–answering

Recent Developments in Arabic Conversational AI: A Literature Review

“…The answer-ranking problem does not generate the answer; instead, it ranks a set of candidate answers based on their relevance to the question. Some studies have focused on building a complete QA system [27], [28]. Others have focused on specific tasks such as QC or QSS in an attempt to improve the quality of the QA systems.…”

“…This approach aims to convert Arabic texts into semantic and logical representations to find the textual implication between the question and the passage that contains its answers [28]. Constructing a semantic and logical representation of texts helps in understanding the concepts from texts and the relations between them.…”

“…Constructing a semantic and logical representation of texts helps in understanding the concepts from texts and the relations between them. However, few studies have focused on the semantic and logic-inference-based approaches to develop Arabic QA systems [28].…”

“…W. Bakari et al [28] proposed a novel logic-based approach for Arabic QA systems based on the automatic understanding of Arabic texts in order to convert them into semantic and logical representations. They used conceptual graphs to obtain a logical representation in order to extract a precise answer based on the relationship between the question and text passage.…”

“…As discussed in the literature, some studies utilized the recognizing textual entailment (RTE) technique, which is remarkably effective in distinguishing between different expressions that express the same meaning [28], [30], [31]. However, their accuracy ranges between 70% and 73%, suggesting that they must either increase the dataset size or enhance the technique.…”