Natural Language Processing and Topic Modeling for Exploring Trends in Human-Robot Interaction

The interdisciplinary field of research involves collaborating and integrating knowledge, methods, and perspectives from multiple disciplines. One example of an interdisciplinary field is human-robot interaction (HRI), which integrates perspectives from robotics, AI, psychology, sociology, human-computer interaction, cognitive science, and other fields. Comprehending the complexity of HRI requires semantic analysis of scientific publications to advance scientific knowledge, and foster collaboration & innovation.
Literature mapping tools such as KATI, help researchers extract relevant data from large volumes of scientific publications. KATI offers bibliometric analysis of scientific publications for data-driven foresight. In order to improve the KATI tool’s capabilities, an exploratory study is conducted to investigate the evolution of the interdisciplinary field of HRI between 1999 and 2020.
Scholarly literature is used to analyze the evolution of HRI using the methodology of Wang et al. and topic modeling approaches. Topic modeling identifies hidden semantic patterns, enabling researchers to obtain a deeper understanding of their field of research. Two state-of-the-art topic modeling techniques such as LDA and BERTopic are employed to extract subfields of HRI. By identifying semantically related topics in HRI empowers researchers to formulate novel hypotheses regarding future directions. The evolution process examines different sub-fields of HRI, which are essential for advancing innovation and exploring potential applications. A comparative analysis of two topic modeling techniques is conducted enabling researchers to advance the field of topic modeling, inform best practices, and empower decision-makers to effectively leverage topic modeling techniques for knowledge discovery and insight generation.