Explainable Artificial Intelligence in Prediction Models: A Bibliometric Analysis

Explainable Artificial Intelligence in Prediction Models: A Bibliometric Analysis

As predictive models are applied in various fields, the demands for both good performance and clear explanations of how such results were obtained. This study performs a bibliometric analysis on studies applying Explainable Artificial Intelligence (XAI) to prediction models. Articles from 2013 to April 30, 2024, are extracted from the Web of Science database using the dimensions of explainability, AI, forecasting, and algorithms. A total of 243 articles are selected and analyzed for number of publications, number of citations, journal, country of origin, keywords, and topic trends. The analysis reveals that the number of publications increase in 2019, with the number of citations rapidly increasing from 2020. China was identified as the most productive country, and the journal IEEE Access had the highest number of articles. By comparing keywords and topic trends before 2021 and after 2022 using authors' keywords, more detailed XAI-related keywords occurring after 2022 are identified. Additionally, the range of topics is found to become more diverse, revealing classifications of specific XAI technologies. This study provides insights into the current status of research on prediction models and XAI and suggests future research directions.