Table 1 Existing construction-related QA...

Table 1

Existing construction-related QA studies

No	References	QA targeted specific areas	QA-used models	Training-free	Knowledge scope for QA	QA performance test dataset
No	References	QA targeted specific areas	QA-used models	Training-free	Knowledge scope for QA	Question source	Number of questions
1	Chou et al. (2024)	Risk management in river dredging projects	A BERT-based deep learning model	×	Dredging risk knowledge collected by interviews	Listed by experienced dredging personnel	16
2	Kim et al. (2024)	Construction market knowledge in overseas projects	A BERT-based deep learning model	×	3 versions of a FIDIC standard contract written in English, Korean, and Indonesian	The FIDIC documents	80
3	Xue et al. (2024)	Building codes	A BERT-based deep learning model	×	2 Chapters of the IBC 2015	Manually generated for model testing	175
4	Lee et al. (2023)	Steel manufacturer equipment procurement	A machine learning model combining KG and QA	×	An equipment procurement document from a steel-making company	Generated questions based on relevant arbitration and clause settings	45
5	Tian et al. (2023)	Construction safety hazard	A BERT + BiGRU + Self-Attention-based deep learning model	×	6,325 safety hazard texts	Dedicated questions for model application	25
6	Wang and El-Gohary (2023)	Construction safety hazard	A CNN-based deep learning model	×	20 OSHA sections related to fall protection	Manually developed for model testing	671
7	Xu et al. (2023)	Coal mine construction safety	A BERT-BiLSTM-CRF-based deep learning model	×	43 sections of 80 papers from coal mine construction safety management standard specifications	Example questions used to validate the semantic query and entity information modules	Unspecified
8	Sun et al. (2020)	Construction document information transmission mining	A TF-IDF-based machine learning model	×	A monthly construction report containing 1734 words	Posed by three construction managers	5
9	Zhong et al. (2020)	Construction procedural constraint	A BiLSTM- + CRF-based deep learning model	×	14 types of national standards of CACQ in China	Sentences labeled by experts	400
10	Rajpurkar et al. (2016)	Multiple domains including building regulation domain	A logistic regression-based machine learning model	×	536 Wikipedia articles	Contributed by 5 civil engineers	Unspecified

No	References	QA targeted specific areas	QA-used models	Training-free	Knowledge scope for QA	QA performance test dataset
No	References	QA targeted specific areas	QA-used models	Training-free	Knowledge scope for QA	Question source	Number of questions
1	Chou et al. (2024)	Risk management in river dredging projects	A BERT-based deep learning model	×	Dredging risk knowledge collected by interviews	Listed by experienced dredging personnel	16
2	Kim et al. (2024)	Construction market knowledge in overseas projects	A BERT-based deep learning model	×	3 versions of a FIDIC standard contract written in English, Korean, and Indonesian	The FIDIC documents	80
3	Xue et al. (2024)	Building codes	A BERT-based deep learning model	×	2 Chapters of the IBC 2015	Manually generated for model testing	175
4	Lee et al. (2023)	Steel manufacturer equipment procurement	A machine learning model combining KG and QA	×	An equipment procurement document from a steel-making company	Generated questions based on relevant arbitration and clause settings	45
5	Tian et al. (2023)	Construction safety hazard	A BERT + BiGRU + Self-Attention-based deep learning model	×	6,325 safety hazard texts	Dedicated questions for model application	25
6	Wang and El-Gohary (2023)	Construction safety hazard	A CNN-based deep learning model	×	20 OSHA sections related to fall protection	Manually developed for model testing	671
7	Xu et al. (2023)	Coal mine construction safety	A BERT-BiLSTM-CRF-based deep learning model	×	43 sections of 80 papers from coal mine construction safety management standard specifications	Example questions used to validate the semantic query and entity information modules	Unspecified
8	Sun et al. (2020)	Construction document information transmission mining	A TF-IDF-based machine learning model	×	A monthly construction report containing 1734 words	Posed by three construction managers	5
9	Zhong et al. (2020)	Construction procedural constraint	A BiLSTM- + CRF-based deep learning model	×	14 types of national standards of CACQ in China	Sentences labeled by experts	400
10	Rajpurkar et al. (2016)	Multiple domains including building regulation domain	A logistic regression-based machine learning model	×	536 Wikipedia articles	Contributed by 5 civil engineers	Unspecified

Note(s): BERT: Bidirectional Encoder Representations from Transformers; BiGRU: Bidirectional Gated Recurrent Unit; BIM: Building Information Modeling; BiLSTM: Bidirectional Long Short-Term Memory; CACQ: Code for Acceptance of Construction Quality; CRF: Conditional Random Fields; FIDIC: International Federation of Consulting Engineers; IBC: International Building Code; IE: Information Extraction; KG: Knowledge Graph; NHC: National Hurricane Center; NLG: Natural Language Generation; NLP: Natural Language Process; NLU: Natural Language Understanding; OSHA: Occupational Safety and Health Organization; TF-IDF: Term Frequency-Inverse Document Frequency

Source(s): Authors’ own work

[ViewLarge]