Session: CIE-24-02 - AMS-CAPPD-SEIKEM: Artificial Intelligence and Machine Learning in Design and Manufacturing

Paper Number: 90688

90688 - Knowledge Extraction Method to Support Domain Integrated Design Methodology 

Nowadays, bio-inspiration has enhanced the creation of sustainable and innovative solutions to modern engineering problems. As an excellent source for multifunctional and optimized designs, nature could inspire mechanical engineers to develop innovative ideas. However, it is very challenging to extract desired design knowledge from primarily text-based databases and describe nature and biology systems. The main objective of this present study is to build a Natural Language Processing (NLP) and text mining model to extract useful information from the desired research papers and databases to create the output to support domain integrated design methodology. The proposed system is called a research paper classification system. First, various design functionalities were summarized based on the available resources from the asknature database, which contains over 1600 examples of biological systems that are extraordinary designs. Second, the main information extracted from the database and papers were manually labelled according to their respective functionalities. Finally, a multi-label classification model was implemented on the built dataset after the label step was finished. The study shows that the proposed research paper classification system can categorize bio-inspired papers efficiently and effectively. Results, discussion, conclusion and future recommendations are presented. This research may guide the further study of knowledge extraction related to bio-inspired design.

Presenting Author: Siyuan Sun McGill University

Presenting Author Biography: Siyuan Sun is a Master of Engineering thesis student at McGill University. She has been working on research about bridging texted-based knowledge with bio-inspired design by using text mining and Natural Language Processing techniques during her graduate study. She has also implemented and evaluated different machine learning models for the keyword extraction system and optimized the NLP and text mining techniques for future researchers to extract information from the database faster and more efficiently.

Authors:

Siyuan Sun McGill University
Pavan Tejaswi Velivela McGill University
Yong Zeng Concordia University
Yaoyao Fiona Zhao McGill University