A natural language processing aided relational information management approach for architectural knowledge fabrication

Throughout the ages, the cumulative production of knowledge has driven civilization forward. The knowledge fabrication process is extracting and analyzing meaning from unstructured information heaps and the knowledge of architecture is not an exception to this. Accordingly, the organization of this information is one of the most vital elements for developing disciplines. Human collects and juxtaposes information within its capacity and limitations, which may inadvertently lead to biased and bounded conclusions. Therefore, no matter how detailed and extensive the content of architecture is, the fabricated knowledge will be within the boundaries of the researcher's knowledge, interest, and capacity. There is a complex body of data for the architecture discipline spread over many fields and fed from various disciplines, events, and approaches. It can be claimed that each piece of information that dwells in architecture is relationally intertwined with each other creating a network. This study aims to try a new relational information organization method for architecture that might lead to new theoretical expansions by making complex and undiscovered relational analyzes readable and manageable for researchers' perceptions. Obsidian.md was deemed appropriate for this study, which is a hypertext-based networked content tool that produces layered, relational, and expandable networks between semantic data pieces. With a potential auxiliary tool to embody and simulate the network of relationships for architectural knowledge, it is aimed to manage the chaotic information into meaningful and readable expertise to the human and reveal uncharted associations for researchers. Due to the sheer volume of data and the need for a systematic, consistent, and rapidly established structure, the use of artificial intelligence algorithms for such a tool has emerged. For this reason, the artificial intelligence field NLP (Natural Language Processing), which is used for semantic data organization, is included in the study. The scope of the thesis covers two themes: the presentation of NLP and a relational content management tool as a possible axillary tool for architectural information organization; and the investigation of fundamental principles and definitions to be used for this NLP aided content management approach for structuring and exploration of architectural knowledge. As a result of the study, a workflow was established between NLP technology tools and Obsidian.md. Within the scope of the determined subject Ankara Roman Bath, a relational knowledge network was obtained through the Obsidian.md tool and the integration of future source texts into this network graphic was optimized.