Open-Source Transformer-Based Information Retrieval System for Energy Efficient Robotics Related Literature

Abstract Background and Purpose This article employs the Hugging Face keyphrase-extraction-kbir-inspec machine learning model to analyze 654 abstracts on the topic of energy efficiency in systems and control, computer science and robotics. Methods This study targeted specific arXiv categories related to energy efficiency, scraping and processing abstracts with a state-of-the-art Transformer-based Hugging Face AI model to extract keyphrases, thereby enabling the creation of related keyphrase networks and the retrieval of relevant scientific preprints. Results The results demonstrate that state-of-the-art open-source machine learning models can extract valuable information from unstructured data, revealing prominent topics in the evolving field of energy-efficiency. Conclusion: This showcases the current landscape and highlights the capability of such information systems to pinpoint both well researched and less researched areas, potentially serving as an information retrieval system or early warning system for emerging technologies that promote environmental sustainability and cost efficiency.