An Ontology-based Semantic Web for Arabic Question Answering: The Case of E-Government Services

Thumbnail Image
Journal Title
Journal ISSN
Volume Title
The British University in Dubai (BUiD)
Since the beginning of the digital age, the amount of information has increased significantly, and the relationships among the different types of information have become more sophisticated. The fact that the number of diverse users is growing urges researchers to benefit from this information and develop techniques that analyse customers’ experiences to meet their needs. As far as e-government is concerned, the number of structured and unstructured webpages of electronic services has increased, making the repositories more complex and harder to analyse without considering semantic knowledge. Related studies have highlighted some challenges in the Arabic Semantic Web (SW) that adversely impact their results, for instance, data heterogeneity, and the differences in ontology construction approach. In this study, we present an approach for automatic extraction of an ontology-based SW constructed from Arabic webpages related to Dubai’s e-government services. Furthermore, we use the constructed ontology as the knowledge base for a question-answering (QA) task process to answer questions related to e-government services. The proposed methodology consists of two stages. The first stage is automatic ontology construction for Dubai government services. This stage is concerned with data extraction and validation from the Dubai government portal that includes the official profiles for more than 500 services. After that, the Natural Language Processing (NLP) tasks are used to process the services’ profiles and extract the ontological keywords. Next, we map the rules to link the ontology components with the extracted keywords. Lastly, the ontology is constructed using the OWL format. In the second stage, an Arabic QA approach is implemented to answer user questions relevant to e-government services. This stage comprises three steps: question analysis, information retrieval (IR), and answer validation. We conducted experimental performance evaluation for all stages in our methodology. The ontology construction stage reported high scores in terms of precision, with 87% on average, and recall, with 97% on average. Further, 414 automatic questions are tested on the QA algorithm using two methods, semantics-based and keyword-based. The accuracy results show 90% for semantics-based and 72% for keyword-based. These results confirm that the semantics-based approach significantly outperforms the keyword-based approach.
ontology-based semantic web, Arabic question answering, E-Government services, United Arab Emirates (UAE), Dubai government portal