The construction industry aims to fully digitize the processes that are time-consuming, error-prone, and labor-intensive. One of the processes that needs digitization, is verifying the regulations and standards to ensure the validation of the construction work. The common obstacle that most of the studies address is that the regulations are not structured for machine interpretation and that sparks the exploration to develop tools and methodologies to parse the regulations effectively, enabling accurate information retrieval. This paper proposes a deep learning-based approach, which relies on the sequence labeling of key text in regulatory documents, to empower information extraction. Our proposed method commences with comprehensive text preprocessing and deep understanding to ensure effective and accurate extraction of information. Additionally, an entity list is outlined to provide a guideline for the prominent text. We employed Named Entity Recognition tagging to label the key text in German regulatory documents from the Research Society for Roads and Traffic regulatory documents. Subsequently, we leverage the pre-trained word embeddings to transform the textual dataset into numerical representations, which could be understandable by a neural network. The numerical representations work as a foundation for our model, which is based on a Bidirectional Long Short-Term Memory network. The labeled data is trained on the network, which provides an evaluation of the extracted data.