Retrieving and discovering new knowledge from documents' abstracts in scientific databases: Proposing a query-based abstractive summarization model

Abstract

Current search engines for Knowledge Retrieval (KR) and Knowledge Discovery (KD) do not effectively utilize scientifically validated documents, especially those indexed in scientific databases. Scientific databases e.g., Scopus primarily consist of document-based content and provide documents' abstract. Their Information Retrieval (IR) system only perform document searches and lack the capability to extract and discover new knowledge from documents' abstract in these databases and responding to users’ queries. The aim is to introduce a model that can efficiently perform these tasks. The statistical population for this study encompasses all scientific databases, with a particular emphasis on Scopus. To clarify the process of KR and KD as we define it, we employed a systematic review and meta-analysis framework using 33 queries. We conducted the identification, screening, eligibility, and inclusion steps following the PRISMA protocol. Next, we performed extraction, labeling, grouping, analysis, and inference. The outcome of these processes provided us with novel insights, which contribute to our exploratory knowledge. To automate these processes, we have proposed a conceptual model from query-based indirect abstractive summarization approach. The outcomes of this research offer fresh insights to database designers, administrators, and researchers, enabling the development of tools for KR and KD within these invaluable knowledge repositories. The integration of such tools into scientific databases will enhance user access to scientific knowledge to meet their informational and research needs.