How does the study use semantic analysis of historical texts to reconstruct the urban space of Luoyang in the Northern Wei Dynasty?
The study applies semantic analysis to historical texts about Northern Wei Luoyang using HanLP for part-of-speech tagging and syntactic parsing. This produces semantic trees and abstract meaning representations that identify buildings and spatial relations such as which temple lies north or south of another. By cross-referencing texts like A Record of Buddhist Monasteries in Luoyang, the researchers verified the relative positions of 48 temples to map the road network, water system, and temple distribution. Semantic analysis at the block scale then used quantitative architectural descriptions to infer building sizes, typologies, courtyard layouts, and street layouts.
The research pipeline begins by parsing ancient sentences with the HanLP model, which performs part-of-speech tagging and syntactic analysis to construct semantic trees and extract entities and their interaction relationships. For example, analysis of a sentence describing relative temple locations correctly identified nouns such as "Guo Zi", "Zong Zheng Si", and "Tai Miao", recognized directional terms like "Nan(south)", and revealed that Zong Zheng Si stands north of institutions including Tai Miao and Hu Jun Fu. Because the model could missegment proper nouns or misread terms, the study expanded the historical text corpus, supplemented proper nouns, and built a specialized dictionary to improve recognition of urban spatial descriptions. At the urban scale, the researchers cross-referenced multiple texts to verify the relative positions of 48 temples and mapped them onto a basic road network, water system, and temple distribution, later refining the reconstructed space with modern GIS satellite data. At the block scale, semantic analysis of descriptive phrases such as "the pagoda has four sides, each with three doors and six windows" and "twelve gates and twenty-four doors" provided modular information for estimating building dimensions and types, which were matched to typological archetypes such as linear, L-shaped, C-shaped, or rectangular forms. Descriptions of courtyards, gates, and streets similarly enabled deductions about the layout of blocks and the relationships between courtyards and streets, collectively producing a reconstructed spatial landscape of Northern Wei Luoyang.
Key points
- Historical texts are parsed with HanLP for part-of-speech tagging, syntactic analysis, and semantic tree construction.
- Abstract Meaning Representation analysis extracts building entities, directional terms, and spatial relationships between buildings.
- A specialized dictionary and supplementary proper nouns improve accuracy when parsing ancient urban place names.
- Urban-scale semantic analysis cross-references texts to verify positions of 48 temples and map roads, water systems, and temple distribution.
- Block-scale analysis reads quantitative descriptions to estimate building sizes, types, courtyard layouts, and street arrangements.
- Modern GIS satellite data refines the reconstructed urban space as archaeological details become available.
AI for Architecture: Proceedings of the 5th International Conference on Computational Design (CCD 2024)
Jiayan Fu
Springer Nature Singapore Pte Ltd.