Intermediate representations to improve the semantic parsing of building regulations

Stefan Fuchs, Johannes Dimyadi, Michael Witbrock, Robert AmorAdv. Eng. Informatics 2024 · OpenReview

Abstract

Highlights•Generative neural networks allow one to formalise building regulations end-to-end.•Reversible intermediate representations simplify and accelerate semantic parsing.•Lossy intermediate representations enable hierarchical semantic parsing.•Potential for a simple and efficient human-in-the-loop formalisation approach.
BibTeX
@inproceedings{fuchs2024intermediate,
  author = {Stefan Fuchs and Johannes Dimyadi and Michael Witbrock and Robert Amor},
  title = {{Intermediate representations to improve the semantic parsing of building regulations}},
  booktitle = {Adv. Eng. Informatics},
  year = {2024},
  url = {https://openreview.net/forum?id=b3VdOcB8xu},
}

Loading related content...

Figure