Dumbach P (2024)
Publication Language: English
Publication Type: Thesis
Publication year: 2024
Publisher: OPEN FAU
City/Town: Erlangen
URI: https://open.fau.de/handle/openfau/31160
Artificial Intelligence (AI) and related fields, based on decades of research, have seen a rise in recent years and are expected to have a significant impact on large parts of industry and society. Besides pure technical advancement, there are further social and economic factors, that influence the adoption of AI and need to be considered. There is an interest in detecting trends and analysing their impact from an industrial perspective, which concerns applications of AI but also from a societal and political perspective. In this cumulative dissertation, I demonstrate the existence of research gaps not only in the individual research areas of Web Mining (WM), trend detection, and AI adoption but specifically in the combination and parallel consideration of these areas within industrial and public sectors or organizational entities such as Small andMedium-sized Enterprises (SMEs). These identified gaps led to the following research objectives addressed in the contributed publications.
The first objective is to investigate German industrial and political data, from five business magazines and two central political protocols of the German federal parliament, in trend detection and to investigate whether and how industry trends are reflected in political data sources. In the journal publication [P1] I led an effort to address this objective by proposing a WM approach that was utilized for the creation of a novel data set with business and political data sources comprising 1.07 million documents from 1998 to 2020. Using 246 identified AI-related buzzwords the AI trend could be visualized and its development further examined over more than two decades based on the relative occurrences and importance of the words. Within this study, the adoption of AI trends in German business and politics was investigated. The results showed the reflection of industrial trends in political data sources and revealed a faster adoption in business compared to politics. In the last years of the period under review, the analysis showed a notable increase in political discourse regarding the area of AI. In addition, specific and interpretable topics in business and political data sources could be identified using topic clustering.
The second objective is to use the emerging digital medium of podcasts for trend detection and to propose a novel data-driven approach in WM based on podcasts. This objective was addressed in the journal article [P2] by examining the AI trend development in healthcare based on podcast data. For this purpose, I proposed aWMapproach together with my co-authors, which was used to create a novel data set of more than 3,400 episodes of 29 healthcare podcasts from 2015 to 2021. Various glossaries and hype cycles were considered for the identification and extraction of 102 AI-related buzzwords. Again, these buzzwords were used to successfully detect an AI trend and to analyze its development. In addition, interpretable topics in healthcare could be identified and examined over time based on the data source of podcasts. The detection of 14 topic clusters led to the visualization of the most dominant topics. The study outcome illustrated the transferability of the approach for trend detection besides AI and showed opportunities for future research using podcasts as a research medium.
Finally, the third objective is to examine the AI adoption in industrial organizations based on SMEs in the healthcare sector and to conduct a cross-national comparison of China and Germany. The article [P3] addressed this objective by investigating the development and adoption of AI from the company’s perspective. I conducted together with my co-authors a multiple-case study collecting primary data from 14 SMEs equally distributed from both countries and gave insights regarding the perceived advantages and challenges of AI adoption among healthcare SMEs. Furthermore, the views in both countries on the expected future development of AI were presented and the results showed organizational requirements for AI implementation in healthcare.
This cumulative dissertation presents methods to identify, collect and exploit data from new or less-considered sources in an industrial and political context. DifferentWMapproaches including the pipeline are shown to address the challenges of large (public) databases as sources and enable other researchers to recreate the novel data sets as well as generate further insights beyond the described application domains and sectors. In addition, this doctoral thesis contributes to empirical research with a cross-national comparison of the use of AI in the German and Chinese healthcare industries, with a particular focus on the organizational group of SMEs, which has received less attention in the literature to date.
APA:
Dumbach, P. (2024). Adoption of Artificial Intelligence in Industry and Politics: An Analysis Based on Web Mining and Case Studies across Diverse Data Sources (Dissertation).
MLA:
Dumbach, Philipp. Adoption of Artificial Intelligence in Industry and Politics: An Analysis Based on Web Mining and Case Studies across Diverse Data Sources. Dissertation, Erlangen: OPEN FAU, 2024.
BibTeX: Download