TY - GEN
T1 - Automated Relation Extraction for Improved Generalizability across Different Types of Text
AU - Chen, Qiyang
AU - El-Gohary, Nora
N1 - The authors would like to thank the National Science Foundation (NSF). This material is based on work supported by the NSF under Grant No. 1937115. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the NSF.
PY - 2024
Y1 - 2024
N2 - Bridge textual reports capture technically detailed data/information about bridge conditions and maintenance actions, which offers opportunities to improve the prediction of future bridge conditions for further bridge maintenance decision-making support. To automatically analyze these reports, there is a need for relation extraction methods to extract relation information from the reports for linking recognized entities with predefined semantic categories (e.g., caused by) and representing the extracted semantic relations in a structured way. To address this need, this paper proposes a deep learning-based relation extraction model. The proposed model utilizes convolutional neural network (CNN) to encode sentence-level features, and bidirectional long short-term memory (BiLSTM) to build a relation extractor to capture the patterns of the predefined relation types. The proposed model was evaluated in extracting relations from multiple types of bridge-related textual reports for representing bridge defect information - including relations among bridge entities - in a semantically rich structured way.
AB - Bridge textual reports capture technically detailed data/information about bridge conditions and maintenance actions, which offers opportunities to improve the prediction of future bridge conditions for further bridge maintenance decision-making support. To automatically analyze these reports, there is a need for relation extraction methods to extract relation information from the reports for linking recognized entities with predefined semantic categories (e.g., caused by) and representing the extracted semantic relations in a structured way. To address this need, this paper proposes a deep learning-based relation extraction model. The proposed model utilizes convolutional neural network (CNN) to encode sentence-level features, and bidirectional long short-term memory (BiLSTM) to build a relation extractor to capture the patterns of the predefined relation types. The proposed model was evaluated in extracting relations from multiple types of bridge-related textual reports for representing bridge defect information - including relations among bridge entities - in a semantically rich structured way.
KW - Artificial intelligence
KW - Bridge conditions.
KW - Bridge inspection
KW - Deep learning
KW - Information extraction
KW - Natural language processing
KW - Relation extraction
UR - http://www.scopus.com/inward/record.url?scp=85184281912&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85184281912&partnerID=8YFLogxK
U2 - 10.1061/9780784485231.054
DO - 10.1061/9780784485231.054
M3 - Conference contribution
AN - SCOPUS:85184281912
T3 - Computing in Civil Engineering 2023: Visualization, Information Modeling, and Simulation - Selected Papers from the ASCE International Conference on Computing in Civil Engineering 2023
SP - 451
EP - 458
BT - Computing in Civil Engineering 2023
A2 - Turkan, Yelda
A2 - Louis, Joseph
A2 - Leite, Fernanda
A2 - Ergan, Semiha
PB - American Society of Civil Engineers
T2 - ASCE International Conference on Computing in Civil Engineering 2023: Visualization, Information Modeling, and Simulation, i3CE 2023
Y2 - 25 June 2023 through 28 June 2023
ER -