Cross-Domain Ontology Keyword Detection for Paper Recommendation

In recent years, the rapid increase in paper publication has made it challenging for researchers to find interesting and relevant previous studies, especially across multiple disciplines. Conventional keyword search often fails to capture the complexity of cross-domain research. In computer science field, existing tools, such as Computer Science Ontology (CSO) Classifier, are limited in detecting relationships between concepts from different fields. This study proposes a cross-domain ontology approach to address this issue by combining the CSO with WordNet. As a case study, we chose Computer Science and Biology fields since there are a lot of papers on computer science and biology, such as bio-informatics. The cross-domain ontology is used to gain extended terms of the extracted papers’ keywords. These keywords and their extended terms are utilized in paper recommendations. The experimental results show that the proposed approach could improve keyword detection and increase the number of recommended papers in bioinformatics and Computer Science by 30.7% compared to conventional methods.