Semantic data integration for supply chain management: with a specific focus on applications in the semiconductor industry

Abstract

Supply Chain Management (SCM) is essential to monitor, control, and enhance the performance of SCs. Increasing globalization and diversity of Supply Chains (SC)s lead to complex SC structures, limited visibility among SC partners, and challenging collaboration caused by dispersed data silos. Digitalization is responsible for driving and transforming SCs of fundamental sectors such as the semiconductor industry. This is further accelerated due to the inevitable role that semiconductor products play in electronics, IoT, and security systems. Semiconductor SCM is unique as the SC operations exhibit special features, e.g., long production lead times and short product life. Hence, systematic SCM is required to establish information exchange, overcome inefficiency resulting from incompatibility, and adapt to industry-specific challenges.

The Semantic Web is designed for linking data and establishing information exchange. Semantic models provide high-level descriptions of the domain that enable interoperability. Semantic data integration consolidates the heterogeneous data into meaningful and valuable information. The main goal of this thesis is to investigate Semantic Web Technologies (SWT) for SCM with a specific focus on applications in the semiconductor industry.

As part of SCM, End-to-End SC modeling ensures visibility of SC partners and flows. Existing models are limited in the way they represent operational SC relationships beyond one-to-one structures. The scarcity of empirical data from multiple SC partners hinders the analysis of the impact of supply network partners on each other and the benchmarking of the overall SC performance. In our work, we investigate (i) how semantic models can be used to standardize and benchmark SCs. Moreover, in a volatile and unpredictable environment, SC experts require methodical and efficient approaches to integrate various data sources for informed decision-making regarding SC behavior. Thus, this work addresses (ii) how semantic data integration can help make SCs more efficient and resilient. Moreover, to secure a good position in a competitive market, semiconductor SCs strive to implement operational strategies to control demand variation, i.e., bullwhip, while maintaining sustainable relationships with customers. We examine (iii) how we can apply semantic technologies to specifically support semiconductor SCs. In this thesis, we provide semantic models that integrate, in a standardized way, SC processes, structure, and flows, ensuring both an elaborate understanding of the holistic SCs and including granular operational details. We demonstrate that these models enable the instantiation of a synthetic SC for benchmarking. We contribute with semantic data integration applications to enable interoperability and make SCs more efficient and resilient. Moreover, we leverage ontologies and KGs to implement customer-oriented bullwhip-taming strategies. We create semantic-based approaches intertwined with Artificial Intelligence (AI) algorithms to address semiconductor industry specifics and ensure operational excellence. The results prove that relying on semantic technologies contributes to achieving rigorous and systematic SCM. We deem that better standardization, simulation, benchmarking, and analysis, as elaborated in the contributions, will help master more complex SC scenarios. SCs stakeholders can increasingly understand the domain and thus are better equipped with effective control strategies to restrain disruption accelerators, such as the bullwhip effect. In essence, the proposed Sematic Web Technology-based strategies unlock the potential to increase the efficiency, resilience, and operational excellence of supply networks and the semiconductor SC in particular.