From the beginning, the process of research and its publication is an ever-growing phenomenon and with the emergence of web technologies, its growth rate is overwhelming. On a rough estimate, more than thirty thousand research journals have been issuing around four million papers annually on average. Search engines, indexing services, and digital libraries have been searching for such publications over the web. Nevertheless, getting the most relevant articles against the user requests is yet a fantasy. It is mainly because the articles are not appropriately indexed based on the hierarchies of granular subject classification. To overcome this issue, researchers are striving to investigate new techniques for the classification of the research articles especially, when the complete article text is not available (a case of nonopen access articles). The proposed study aims to investigate the multilabel classification over the available metadata in the best possible way and to assess, "to what extent metadata-based features can perform in contrast to content-based approaches." In this regard, novel techniques for investigating multilabel classification have been proposed, developed, and evaluated on metadata such as the Title and Keywords of the articles. The proposed technique has been assessed for two diverse datasets, namely, from the Journal of universal computer science (J.UCS) and the benchmark dataset comprises of the articles published by the Association for computing machinery (ACM). The proposed technique yields encouraging results in contrast to the state-ofthe-art techniques in the literature.
The profusion of documents production at an exponential rate over the web has made it difficult for the scientific community to retrieve most relevant information against the query. The research community is busy in proposing innovative mechanisms to ensure the document retrieval in a flexible manner. The document classification is a core concept of information retrieval that classifies the documents into predefined categories. In scientific domain, classification of documents to predefined category (ies) is an important research problem and supports number of tasks such as information retrieval, finding experts, recommender systems, etc. In Computer Science, the Association for Computing Machinery (ACM) categorization system is commonly used for organizing research papers in the topical hierarchy defined by the ACM. Accurately assigning a research paper to a predefined category (ACM topic) is a difficult task especially when the paper belongs to multiple topics. In this paper, we exploit the reference section of a research paper to predict the topics of the paper. We have proposed a framework called Category-Based Category Identification (CBCI) for multi-label research papers classification. The proposed approach extracted references from training dataset and grouped them in a Topic-Reference (TR) pair such as TR {Topic, Reference}. The references of the focused paper are parsed and compared in the pair TR {Topic, Reference}. The approach collects the corresponding list of topics matched with the references in the said pair. We have evaluated our technique for two datasets that is Journal of Universal Computer Science (JUCS) and ACM. The proposed approach is able to predict the first node in the ACM topic (topic A to K) with 74% accuracy for both JUCS and ACM dataset for multi-label classification.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.