Intelligent question answering method for construction safety hazard knowledge based on deep semantic mining

Abstract

Timely safety hazard management can reduce the probability of safety accidents at construction sites. However, the formulation of safety hazard management measures is a time-consuming and labor-intensive process. This paper describes a safety hazard knowledge question answering method to automatically generate safety hazard management measures. The method builds a deep learning network fusing Bidirectional Encoder Representation from Transformer (BERT), Bidirectional Gated Recurrent Unit (BiGRU), and Self-attention mechanism to extract text semantic features. Taking the text semantic feature extraction mechanism as a subnet, an answer selection model based on a Siamese neural network is built to implement the deep matching of safety hazard questions and management measures. Experimental results from hydraulic engineering construction demonstrate that the proposed model outperforms the existing answer selection model. Meanwhile, a question answering system based on the proposed model is developed to address safety hazard management problems, which verifies the reliability and applicability of the model.