TY - GEN
T1 - Importance-performance analysis of product attributes using explainable deep neural network from online reviews
AU - Joung, Junegak
AU - Kim, Harrison M.
N1 - Funding Information:
This research was supported by the Basic Science Research Program through the National Research Foundation of Korea(NRF) funded by the Ministry of Education (2019R1I1A1A01063298)
Publisher Copyright:
© 2020 American Society of Mechanical Engineers (ASME). All rights reserved.
PY - 2020
Y1 - 2020
N2 - Importance-performance analysis (IPA) is a technique used to understand customer satisfaction and improve the quality of product attributes. This study proposes an explainable deepneural- network-based method to carry out IPA of product attributes from online reviews for product design. Previous works used shallow neural network (SNN)-based methods to estimate importance values, but it was unclear whether the SNN is an optimal neural network architecture. The estimated importance has high variability by a single neural network from a training set that is randomly selected. However, the proposed method provides importance values with a lower variance by improving the importance estimation of each product attribute in the IPA. The proposed method first identifies the product attributes and estimates their performance. Then, it infers the importance values by combining explanations of the input features from multiple optimal neural networks. A case study on smartphones is used herein to demonstrate the proposed method.
AB - Importance-performance analysis (IPA) is a technique used to understand customer satisfaction and improve the quality of product attributes. This study proposes an explainable deepneural- network-based method to carry out IPA of product attributes from online reviews for product design. Previous works used shallow neural network (SNN)-based methods to estimate importance values, but it was unclear whether the SNN is an optimal neural network architecture. The estimated importance has high variability by a single neural network from a training set that is randomly selected. However, the proposed method provides importance values with a lower variance by improving the importance estimation of each product attribute in the IPA. The proposed method first identifies the product attributes and estimates their performance. Then, it infers the importance values by combining explanations of the input features from multiple optimal neural networks. A case study on smartphones is used herein to demonstrate the proposed method.
UR - http://www.scopus.com/inward/record.url?scp=85096310957&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85096310957&partnerID=8YFLogxK
U2 - 10.1115/DETC2020-22382
DO - 10.1115/DETC2020-22382
M3 - Conference contribution
AN - SCOPUS:85096310957
T3 - Proceedings of the ASME Design Engineering Technical Conference
BT - 46th Design Automation Conference (DAC)
PB - American Society of Mechanical Engineers (ASME)
T2 - ASME 2020 International Design Engineering Technical Conferences and Computers and Information in Engineering Conference, IDETC-CIE 2020
Y2 - 17 August 2020 through 19 August 2020
ER -