Special issue on affective design using big data

In addition to the functionality, prices, performance and quality of new products, consumers are increasingly concerned with affective aspects such as texture, outlook, colour, forms and images of new products. Nowadays, affective aspects of products without doubt play an important role in contributing to their success in the market place. As an early example, the concept of affective design was originated from Kurosu and Kashimura who developed two cash teller machines having identical functional features (Kurosu and Kashimura 1995). One machine was equipped with attractive buttons and screen, and the other was with less attractive ones. User surveys indicated that the most attractive one could help promote apparent usability (Norman 2002; Zhang and Li 2005). Using a more recent example of affective design of smartphones, surveys have shown that smartphones that were equipped with more attractive interface designs helped promote the product, although smartphones are generally developed with similar functions (Kim and Lee 2016). These two studies indicate that products with good affective design excite psychological feelings and improve consumer satisfaction in terms of emotional aspects. Therefore, considering affective aspects in engineered product designs is essential to identify and develop pleasurable features into new products that meet affective needs of customers.

Affective needs of customers are commonly collected by surveys. Potential consumers are asked to fill in questionnaires and/or participate in interviews in order to uncover their affective needs towards products. However, conducting surveys/interviews is generally expensive and time consuming and there is no guarantee that all domains of affective needs can be captured. Since only limited Kansei words and affective needs can often be addressed in surveys or interviews, important affective needs for new product development may be partially or fully overlooked. Thanks to the advanced technologies of capturing ‘big data’, 2.5 quintillion bytes of data can be captured on a daily basis through the internet such as pervasive sensor networks, social media, web pages, or blogs (IBM 2015). ‘Big data’ can be used to capture useful information for developing corporate strategies, marketing campaigns and new products. Many companies adopt affective computing to realise product differentiation strategies. Techniques involving big data can potentially be applied to affective design.

In line with the technologies of big data, affective computing has been examined over the past few years including product design (Ayas 2011; Koutsabasis and Istikopoulou 2013), fashion design (Sokolova and Fernández-Caballero 2015), web design (Koutsabasis and Istikopoulou 2013), media communication (Bergen and Ross 2013; Cao et al. 2014), computer game (Yannakakis et al. 2014), human computer interaction (Bakhtiyari, Taghavi, and Husain 2015; Park and Zhang 2015), service development (Hensher 2014; Morris and Guerra 2015) and urban landscape design. From the literature, a growing interest in mining multi-disciplinary affective data by both researchers and industry can be seen. Although several special issues/sessions have targeted research and development in affective computing (Doctor 2015; Fritsh and Markussen 2012; Pantic, Pentland, and Nijholt 2009; Schwenker, Scherer, and Morency 2015; Soleymani et al. 2015; Yannakakis et al. 2014; Zhang 2016), there seems to be no special issue/session dedicated to the integration between affective computing and product design.

Considering more recent developments in affective computing, it is vital to capture the latest use of computing technologies in affective design with big data, which is the main focus of this special issue. Specifically, this special issue has solicited original papers describing innovative technologies/techniques to conjunct affective design using big data. The Call For Papers (Chan, Wong, and Kwong 2016) was circulated in November 2016, which received 34 submissions, resulting in six selected papers for this special issue. These papers discuss a variety of novel computational approaches in using big data for affective design. Sources of big data used in these studies are online questionnaires (1 paper), online customer reviews (3 papers) and crowdsourcing (2 papers).

A machine learning framework to automatically collect product information from online shopping websites is presented in the first paper entitled ‘Dynamic mapping of design elements and affective responses: A machine learning based method for affective design’ (Wang et al. 2018b). Based on product information, online questionnaires are automatically generated in order to survey affective responses from potential consumers. The survey data is used to develop a model that correlates design elements and consumer affective responses. The framework attempts to overcome the limitations of the traditional questionnaire consumer survey, which cannot be updated automatically with respect to the changing life cycles of products and the changing affective responses. A case study of affective design of smart watches is conducted to validate the effectiveness of the proposed framework.

The second paper in this special issue is entitled ‘A multi-objective PSO approach of mining association rules for affective design based on online customer reviews’ (Jiang et al. 2018). The paper proposes a novel Particle Swarm Optimisation (PSO) based methodology that contains three mechanisms: (i) the first is a process of identifying affective dimensions from online reviews using opinion mining software; (ii) the second process addresses design attributes using morphological analysis; and (iii) the last process of applying PSO to develop association rules taking inputs from the first and second processes. A case study on affective design of compact cars is conducted. Big data is collected from online customer reviews relating to six popular compact cars and the proposed methodology is able to uncover different design alternatives that can meet the affective needs of consumers. The paper provides a useful tool that guides the process of developing a product following online reviews that can be collected over time. The use of PSO offers some useful insights in extracting hidden yet interesting correlations between design attributes and affective dimensions that may not be identified using conventional approaches.

‘Mining of affective responses and affective intentions of products from unstructured text’ (Wang et al. 2018a) is the title of the third article. The authors attempt to enhance the traditional affective design that mostly relies on manual surveys to understand consumer affective needs. Conducting manual surveys is costly, time-consuming and relatively small scale. The authors have developed a text mining framework based on the commonly used Kansei attributes and Kansei words, which are collected from online product descriptions and consumer reviews. The framework attempts to correlate affective intentions from product manufacturers and affective responses from consumers. To evaluate the effectiveness of the proposed framework, a case study is conducted based on Amazon.com data which consists of both online product descriptions and consumer reviews.

The fourth paper is entitled ‘Combining rough set theory with fuzzy cognitive pairwise rating to construct a product platform for tablet design and recommendation’ (Wang 2018). Despite using online reviews, online survey was used to collect a wide range of consumer opinions, which are involved with affective needs. A framework is proposed to transform vague user perceptions of affective features into design attributes of new products. In the framework, Kansei engineering, theory of inventive problem solving, and rough set theory are used to generate decision rules that attempt to relate consumer perceptions, affective needs and demographic variables such as age, gender, and occupation. The rules can be used for affective design. A case study of tablet design is employed to illustrate the effectiveness of the proposed framework.

The fifth paper is entitled ‘A product affective properties identification approach based on web mining in a crowdsourcing environment’ (Chang and Lee 2018). The paper proposes a computational framework for determining and identifying affective properties of a new product. Crowdsourcing platforms are embedded to capture a large number of consumer comments from different data sources. In the computational framework, web and text mining techniques, sentiment analysis and domain ontology are used to correlate design and affective properties in order to perform more effective affective design. A pilot study of iPhone 7 shows that the proposed framework is able to identify and rank influential affective properties.

‘Realising the affective potential of patents: a new model of database interpretation for user-centred design’ is the sixth paper of the special issue (Wodehouse et al. 2018). The authors have proposed a crowdsourcing framework to capture and summarise technical contents from patent database. In the framework, text mining tools are used to further capture the summarised contents that are related to affective parameters including appearance, ease of use, and semantics. The summarised contents are used to generate patent clusters that provide an alternative perspective on relevant technical data, and support user-centric engineering design. The framework generates sets of descriptive words for each patent, which are different to those generated by only using functional requirements. Based on those sets, one can utilise desirable affective qualities scouring relevant functional patent information in new product development.

We would like to thank all the authors who submitted their research outcomes in affective design using big data to this special issue, and also thank to the reviewers who contributed their time and effort to evaluate the submitted articles. We would like to thank Prof. Alex Duffy, the Editor-in-chief of the Journal of Engineering Design, who supported us to edit this special issue and guided us through the reviewing and editing processes. Particularly thanks to his effort in speeding up the decision and reviewing processes. He took a significant role to ensure that this special issue can be delivered on time. We hope that the readers will find this special issue useful for developing research and technologies on affective design using big data.