Translative Research Assistant: A Retrieval-Augmented Generation Pipeline Refinement with Keyword Extraction Using Extended Scalable Betweenness Centrality
Chung-Hsien Chou, Chee-Hann Wu · International Journal of Semantic Computing · 2025
The objective of this research is to introduce a translation tool that addresses two critical aspects: first, the translation of research from other languages into our target language; and second, the adaptation of existing knowledge from other research to align with a researcher’s specific context. To achieve this, we propose these key approaches: summarization, keyword extraction and evaluation, which assesses the relevance of materials to a researcher’s work or identifies the need for further investigation. Our solution is the Translative Research Assistant, leveraging ChatGPT as its primary tool. To enhance the accuracy of its text generation, we advocate for a knowledge retrieval approach utilizing the Retrieval-Augmented Generation pipeline with keyword extraction using proposed Extended Scalable Betweenness Centrality. Ultimately, our aim is to promote the integration of AI across disciplines and enhance the precision of ChatGPT responses, aiding researchers in efficiently assessing the utility of new information they encounter.