Semantic Web and Ontology
Zeynab Sabbaghi Bidgoli; Atefeh Sharif; Fatemeh Zandian
Abstract
IntroductionThe emergence of the web facilitated the retrieval of information. This made libraries as one of the most important centers of information considering the web for the information retrieval process. However, the fast change of the web leads to the transformation of library functions. The semantic ...
Read More
IntroductionThe emergence of the web facilitated the retrieval of information. This made libraries as one of the most important centers of information considering the web for the information retrieval process. However, the fast change of the web leads to the transformation of library functions. The semantic web is an opportunity for libraries to change their functions. Linked data as a method in the semantic web can make a major change in library functions. It can improve the discoverability, visibility, and interoperability of the resources. For example, all libraries use authority controls for organizing their information. But using authority controls in a traditional way can be challenging. Therefore, using the web can help libraries tackle these potential challenges and problems. Transforming authority data into linked data which seems an innovative and faster way for finding the resources can be a step forward for libraries and users. This paper aims to design a framework for transforming the National Library of Iran Subject Headings into linked data and publish them on the web.
Literature ReviewDesigning and proposing a framework for linking the data was the topic of some research papers. Linking the university data (Behkamal et al., 2011) linking and visualizing medicine information (Sekhavati, Farahi, & Jalali, 2011) web objects (Hosseini, 2020), table data (Mulwad et al., 2010), Industrial Data (Graube et al.,2012), and government data (Villazón-Terraza, Vilches-Blázquez, Corcho, & Gómez-Pérez, 2011; Mulwad, Finin, & Joshi, 2011) were the topics for some reviewed studies. The results of their studies indicated that in general, linked data could improve information retrieval. Implementing a linked data method in library data was discussed in some papers. Kar & Das (2020) designed a methodology for linking bibliographic information in a digital repository. Similarly, Ryan et al. (2015) examined the linking of place names in a dataset, transferring them into RDF and linking them with other similar datasets. Summers, et al (2008) provide a methodology for transferring subject headings into linked data. their results showed that transferring LCSH into SKOS affects information retrieval. The linking and publishing National Library of Iran data were also investigated by Eslami & Vaghefzadeh (2013). Fathian Dastgerdi et al (2020) tried to make a pattern for linking data in library systems. They examined the components which are needed for implementing the linked data method in library systems. Their result showed that using linked data in library systems affects the visibility of bibliographic metadata. Based on the reviewed studies, many international papers discussed publishing library linked data in theoretical and practical ways. Whereas studies done in Iran focusing on linked data mostly developed patterns and models for linking data (e.g., Fathian Dastgerrdi; 2020). Few Persian studies were done for publishing bibliographic data (e.g., Eslami & Vaghefzadeh, 2013; Sekhavati, 2011). Although there is a significant number of papers discussing linked data, the technical aspect for publishing and linking library data was rarely examined. To fill this gap, this study aims to develop a framework for publishing National Library of Iran subject headings which is unlike Fathian Dastgerdi et al., (2020) paper considers the technical tools and aspects and unlike Sekhavati’s (2011) paper examines the Persian subject headings.
MethodologyThis research is an applied study that utilizes a library method for designing a publishing framework. Linked data was implemented to ensure the possibility of publishing the research data. First, Persian subject headings which are represented in Iran MARC format were obtained in Marc XML files From the National Library of Iran. Then the method for transferring and publishing the data was applied.
Results The framework developed in this research collected National Library of Iran subject headings randomly. The selected data were first cleaned by Microsoft Excel and MarcEdit. In the next step, cleaned data were converted into RDF Using OpenRefine. The study’s project was imported to Open Refine software, linked with external datasets, and saved in a triple store. Finally, the linked subject headings were displayed through the Skosmos interface.DiscussionPublishing library data as linked data is an example of utilizing Web 3 in library systems. National libraries worldwide have tried linking their data including subject headings with other datasets. However, there remains a gap in publishing linked Persian subject headings and to the best of the authors' knowledge it seems that no paper has pointed to technical aspects of implementing Persian subject headings.
ConclusionThe current paper has transformed the Persian subject headings into a linked dataset in an RDF turtle format. Then, it visualized the linked data in the Skosmos interface. But there can be some limitations to this study. Using OpenRefine was reported successfully in this paper, but it seems that there may be a problem in data with larger sizes. In conclusion, since this framework improve the retrieval of authority data in this research, it can be used for publishing National library of Iran subject headings.
Reza Dehkhodaie; Atefeh Sharif
Abstract
Nowadays, various online resources are growing and disseminating rapidly. In order to organize these resources, attempts have been made to use automatic classification, which often uses statistical algorithms and machine learning. Recently, attention has been drawn to the use of library classifications. ...
Read More
Nowadays, various online resources are growing and disseminating rapidly. In order to organize these resources, attempts have been made to use automatic classification, which often uses statistical algorithms and machine learning. Recently, attention has been drawn to the use of library classifications. The main challenge here is that classification is an abstract, thought-provoking process, and machine techniques and artificial intelligence have not yet been able to completely replace the human mind. In this paper, we provide an overview of the importance of automatic classification, machine learning, and practical algorithms and techniques of clustering and classification like K-nearest neighbor, Bayesian models, artificial neural networks, deep learning, and hybrid classifications. Also, the steps of automatic classification of web pages and the techniques used in each step were mentioned. Achieving a clearer understanding of automatic classification will enable LIS experts to communicate with experts in the field of artificial intelligence and computers. This could pave the way for interdisciplinary research.