QMQA: An intelligent question answering method for quality management of prefabricated buildings based on knowledge graph

Quality compliance inspection is an essential procedure to enhancing the quality of prefabricated buildings. However, the current method for extracting specific legal compliance from regulatory texts is a both time-consuming and knowledge-intensive process. Aiming at the problem of insufficient knowledge service in this field, this paper proposes an intelligent question answering method for quality management of prefabricated buildings based on knowledge graph (QMQA). This method introduces a novel neural network model that integrates BERT with an attention based-bidirectional long short term memory network (AB-BiLSTM). First, a fine-tuned BERT model is used to analyze unstructured text and assign specific semantic tags to each segment. Then, based on the position and frequency of these semantic tags within the sentence, AB-BiLSTM automatically assigns appropriate templates to domain-specific questions, extracting the relevant semantic elements. Finally, this method enables the automatic generation of Cypher, the query language for graph databases. The method is validated using the knowledge graph developed from the Jiangsu Prefabricated Structural Engineering Construction Quality Acceptance Regulations in China. Results demonstrate that the proposed method exhibits outstanding performance, with the model achieving an accuracy of 95.2%. Additionally, this paper has developed a QMQA prototype system that includes functions for quality diagnosis and inspection inquiries. This paper presents a method for rapid quality diagnosis and timely question-answering responses, addressing a critical gap in this field and contributing to the improvement of quality management in prefabricated buildings.