1. Application of COReS to Compute Research Papers Similarity
- Author
-
Muhammad Abdul Qadir, Muhammad Afzal, and Qamar Mahmood
- Subjects
General Computer Science ,Process (engineering) ,Computer science ,content based similarity ,02 engineering and technology ,Ontology (information science) ,computer.software_genre ,Semantics ,ranking ,Similarity (network science) ,Comprehensive similarity computation ,020204 information systems ,0202 electrical engineering, electronic engineering, information engineering ,research paper similarity ,General Materials Science ,ontology ,Cluster analysis ,Measure (data warehouse) ,Information retrieval ,General Engineering ,Encyclopedia ,Ontology ,020201 artificial intelligence & image processing ,lcsh:Electrical engineering. Electronics. Nuclear engineering ,Data mining ,lcsh:TK1-9971 ,computer - Abstract
Over the decades, the immense growth has been reported in research publications due to continuous developments in science. To date, various approaches have been proposed that find similarity between research papers by applying different similarity measures collectively or individually based on the content of research papers. However, the contemporary schemes are not conceptualized enough to find related research papers in a coherent manner. This paper is aimed at finding related research papers by proposing a comprehensive and conceptualized model via building ontology named COReS: Content-based Ontology for Research Paper Similarity. The ontology is built by finding the explicit relationships (i.e., super-type sub-type, disjointedness, and overlapping) between state-of-the-art similarity techniques. This paper presents the applications of the COReS model in the form of a case study followed by an experiment. The case study uses InText citation-based and vector space-based similarity measures and relationships between these measures as defined in COReS. The experiment focuses on the computation of comprehensive similarity and other content-based similarity measures and rankings of research papers according to these measures. The obtained Spearman correlation coefficient results between ranks of research papers for different similarity measures and user study-based measure, justify the application of COReS for the computation of document similarity. The COReS is in the process of evaluation for ontological errors. In the future, COReS will be enriched to provide more knowledge to improve the process of comprehensive research paper similarity computation.
- Published
- 2017