A semantic analysis-driven customer requirements mining method for product conceptual design

Abstract

Precise customer requirements acquisition is the primary stage of product conceptual design, which plays a decisive role in product quality and innovation. However, existing customer requirements mining approaches pay attention to the offline or online customer comment feedback and there has been little quantitative analysis of customer requirements in the analogical reasoning environment. Latent and innovative customer requirements can be expressed by analogical inspiration distinctly. In response, this paper proposes a semantic analysis-driven customer requirements mining method for product conceptual design based on deep transfer learning and improved latent Dirichlet allocation (ILDA). Initially, an analogy-inspired verbal protocol analysis experiment is implemented to obtain detailed customer requirements descriptions of elevator. Then, full connection layers and a softmax layer are added to the output-end of Chinese bidirectional encoder representations from Transformers (BERT) pre-training language model. The above deep transfer model is utilized to realize the customer requirements classification among functional domain, behavioral domain and structural domain in the customer requirement descriptions of elevator by fine-tuning training. Moreover, the ILDA is adopted to mine the functional customer requirements that can represent customer intention maximally. Finally, an effective accuracy of customer requirements classification is acquired by using the BERT deep transfer model. Meanwhile, five kinds of customer requirements of elevator and corresponding keywords as well as their weight coefficients in the topic-word distribution are extracted. This work can provide a novel research perspective on customer requirements mining for product conceptual design through natural language processing.

Mapping artificial intelligence-based methods to engineering design stages: a focused literature review

“…Second, we also observed that four machine learning methods had been used in customer reviews, ratings, and design requirements analyses (Suryadi and Kim 2019; Hein et al 2021; Lin and Kim 2021; Saidani et al 2021). Further, two deep learning methods were used along with NLP for design-requirement-extraction through semantic analysis (Wu et al 2022) and automatic extraction of functional requirements (Akay et al 2021); two network theory methods for requirements-identification (Sen and Summers 2013) and customer segmentation (Park and Kim 2021); and one probabilistic method that used Latent Dirichlet allocation (LDA) to identify and generate latent topic for managing design requirements (Chen et al 2021).…”

“…In the problem definition stage, as customer reviews or requirements data are often available in a textual or natural language format, most research works used NLP techniques to process or analyze them. NLP techniques (Chowdhury 2003; Hirschberg and Manning 2015) have proven to be effective in encoding and processing large-scale textual data to perform operations such as text mining (Tucker and Kim 2011; Zhang et al 2021), semantic analysis (Wu et al 2022), sentiment analysis (Suryadi and Kim 2019), topic extraction (Ayoub et al 2019), and text summarization (Hou et al 2019). These analyses can be beneficial for understanding the key customer and design requirements and can, therefore, assist human designers translating such requirements to apt design parameters such as form, function, and aesthetics.…”

Identification of Demand Type Differences and Their Impact on Consumer Behavior: A Case Study Based on Smart Wearable Product Design

“…Wu, X.Y. used TF-IDF to extract customer demands during the product concept stage, and used it as a driving factor for design [32]. Liu, Q.Y.…”

Contextual Meaning-Based Approach to Fine-Grained Online Product Review Analysis for Product Design

“…In [12], the term frequency-inverse document frequency (TF-IDF) method was used to extract prioritized words related to product features. Second, other studies employed linguistic structure analysis-based methods [11], [17], [18], [26]. These methods emphasize the syntactic and grammatical relationships within the review text, considering the structural patterns to be indicative of the underlying meanings or sentiment.…”

“…In [11], part of speech (POS) tagging was performed on review sentences to extract nouns as opinion targets. In [17] and [26], a dependency parser was employed to extract significant terms and phrases related to product features from reviews. In [18], linguistic pattern mining was used to extract both the product function and the specific environmental context related to it.…”