On November 24 at 2:00 p.m., Benedict Lohse will defend his master's thesis on “Comparison of Selected Generative AI Approaches for Indexing Heterogeneous Document Collections” in Room 320 (Konrad Zuse House). Benedict is a master's student in the Business Informatics program and was supervised by Felix Holz and Kurt Sandkuhl (both from the Chair of Business Informatics).
Abstract
The indexing of unstructured, heterogeneous document collections represents a central challenge for knowledge management in organizations. In this master's thesis, various architectures for structuring heterogeneous document collections using generative artificial intelligence will be developed and presented. The focus is on two different approaches, which are developed as prototypes. These are then compared with each other in an experiment based on defined criteria. Each prototype represents a value chain (toolchain) through which content (e.g., the document collection) is indexed using generative AI. While the basic “WiSiMV” system relies on traditional strategies, the two implemented architectures use future-oriented methods in the field of generative artificial intelligence to create representations of document content that can be retrieved in combination with semantic similarity.