Article Preview
Top1. Introduction
In recent years, several exciting technologies empowered the web and made it richer and dynamic. Soon after its introduction by Berners-Lee, linked data (Berners-Lee, 2006) has grown to be a major technology and led to the creation of the Web of data (Bizer, Heath, & Berners-Lee, 2009). Data providers are continuously adopting a set of best practices (Atemezing & Villazón-Terrazas, 2014; Berners-Lee, 2006; Schmachtenberg, Bizer, & Paulheim, 2014) to publish their data on the web. For instance, the Linking Open Data (LOD) project (Abele, McCrae, Buitelaar, Jentzsch, & Cyganiak, 2017) included in its core 1163 published datasets along with statistical analysis about them (Abele et al., 2017; Schmachtenberg et al., 2014). These datasets are both cross-domain, such as DBpedia (Christian Bizer et al., 2009; Lehmann et al., 2015), and domain-specific such as LinkedGeoData (Stadler, Lehmann, Höffner, & Auer, 2012). Most of these datasets are available in English; thus, neither non-English speakers nor applications can benefit from them.
According to Cisco, the number of internet users in the Middle East and Africa is expected to grow by 6-folds between 2016 and 2021 with a compound rate of 44% annually (Cisco, 2017). Therefore, the number of Arabic users and Arabic resources on the web is and going to be enormous. Only a few Arabic datasets are available as linked data, yet. In addition, Ktob and Li (2017) identified the challenges pertinent to the creation of the Arabic Knowledge Graph (AKG) in which the scarcity of Arabic linked datasets was a significant factor. Besides the proprietary factor of data, the inexistence of tools designed with Arabic support in scope affects significantly the number of Arabic datasets published. Therefore, Marefa, meaning “knowledge”, is a tool developed to automatically extract and ease the publication of Arabic catalogs’ data as Resource Description Framework (RDF) datasets. It, to the best of our knowledge, is the first tool implemented to accomplish the goal mentioned above. Marefa, built with a flexible extraction framework, is user-friendly and easy to use.
The main contributions of this work are:
- 1.
Marefa1 is the first framework developed to convert Arabic catalogs’ data to linked data;
- 2.
Compared to DBpedia and Makhtota+ (Al-Rajebah & Al-Khalifa, 2012), our solution has more coverage (cf. section 2);
- 3.
A 3-phase rigorous extraction process based on a simple and extensible framework (simplify the task for users and can be extended to other languages and dialects after some tweaking, cf. section 4);
- 4.
A set of statistical measures produced along with the RDF dataset to further clarify the results (cf. section 5).
The remainder of the paper is organized as follows: Section 2 examines some of the related works. Section 3 discusses the different frameworks, Application Programming Interface (API), and technologies used by Marefa. The extraction framework and its constituent modules (including algorithms) are detailed in Section 4. Thereafter, Section 5 elaborates on the experiment conducted and gives some analysis of the obtained results. Finally, a conclusion along with future work ideas are included in Section 6.
TopOver the previous decades, there were few similar works to Marefa that aim to improve the presence of Arabic datasets on the web of data. Although they share the same goal, Marefa is the first tool that targets Arabic catalogs to help publishers embrace the Web of data. In this section, the authors examined some of these works and their contributions to the Arabic linked data.