With the rapid development of artificial intelligence, companies have established exclusive chatbots to conduct conversational commerce and strengthen the emotional connection between customers and their brands. Therefore, shaping the chatbot personality to match the brand image is often the focus of chatbot design. This study investigates how users perceive chatbot personality via nonverbal communication (avatar, emoji, sticker) and explores the effect of chatbot personality on user satisfaction in a crowdfunding website. First, six chatbot conditions were selected using the Taguchi method. Then, an online survey based on the brand personality theory was conducted to investigate the impact of nonverbal communication on a chatbot's perceived personality. Based on the results, two chatbots with distinct personality traits, sincere and insincere, were selected as the conditions for the second experiment. Finally, participants interacted with one of the crowdfunding chatbots and received recommendations within FB messenger. One hundred fifty valid questionnaires and the click rate of participants during the experiment were collected to measure participants' satisfaction. The results showed that participants felt more satisfied with the "sincere" chatbot than the "insincere" chatbot. Furthermore, the personality of chatbots also affected participants' judgment on the message quality and the continuous intention to use chatbot services.

This research aims to develop an information-sharing system between customers and employees and try to put it into practical use. We named it "Caffe System (Customer Affection System)", a system for sharing customer service notes and customer feedback to improve customer satisfaction and loyalty and facilitate the sharing of customer information among employees.

As a non-mainstream service, "Message to customers" has been used by some stores. For example, Starbucks coffee is provided with a particular message service to impress customers: providing the drink with a message or illustration. But we have no way to know its specific impact on customer satisfaction and loyalty.

In this study, we launched the "cup message card", considering that we have been conducting demonstration experiments in a dine-in coffee shop. One of the focuses of this study was to investigate the satisfaction difference between "cup message card" and "conversation with clerks", which had already become one of the most mainstream service means.

Purpose of our research is to classify tea image at each shooting time.The flavor and quality of tea leaves change greatly depending on the time of picking.The more Umami ingredient “theanine”, the higher the quality.The more Astringent ingredient “catechin”, the lower the quality.Therefore, it is important to pick the tea at the timing when the theanine content is maximized.

In an actual farm, the timing of picking is decided at the leaf opening stage, not at the component analysis.The leaf opening period consists of 8 stages, from the germination period until the 7th leafopening period.As a rule of thumb, the highest quality is achieved at the 5th leaf opening stage.

Skilled technique is required to determine the leaf stage.We shoot a tea plantation with a large drone and analyze the shot image by deep learning.Analyzing each leaf opening stage leads to quality prediction.And I've been thinking about how to predict quality.

In order to determine the leaf opening stage, we tried to classify the images using a well-known method, Luminance histogram,Spectrum analysis by FFT and AKAZE.

However, none of the conventional methods were able to classify well.The cause is that it is difficult to extract the features due to the fuzzy of the tea image.

Therefore, we devised a classification method using deep learning.

We use DCGAN, SAE, LSTM and CWD.The originality of this research lies in the fact that these analyzes are not performed individually but continuously.

Today’s markets often offer different product solutions to the same customer need. In particular, new products offer new features requiring new customer behavior. This is typically driven by technology push, environmental concerns, or legal requirements. Due to inconvenience users do not easily change their behavior and thereby reject novel products unless they offer an obvious delight in the eyes of the user. Product developers need to take this in account when designing products. Tools such as UX, Kansei Engineering and Kano model are amongst those supporting product practitioners. This is exemplified by the affective evaluation of computer mice over three generations. It can be seen that in particular women have different affective needs in comparison to men related to auditive properties of mice. Hence it is important to adjust the user panel composition in accordance with the expected effect in order to avoid biases due to lacking diversity.

The study proposes a method for the assessment of e-shop product pages (PPs) through individual assessments of the touchpoints (TPs) customers interact with while using a PP (e.g., product title, main image, image gallery, Buy Box, rating/reviews number, product description, specifications, cross sales/recommended products, video, and product warnings) rather than an assessment of the product page as a whole. A specific Kansei list is created, customer preferences are collected, and touchpoints can be classified as positive, neutral, or negative (pain points). The assessment method was applied to the assessment of PPs of craft beer e-retailers. Once pain points were identified, two distinct types of TPs among them were individually analyzed (one visual, the Image gallery, and one textual, the Product description.) In both cases, a set of PPs were employed as stimuli and design elements that were deemed preferable by the customers were identified through Partial Least Squares analysis. The results provided valuable recommendations to web designers, content creators, and retailers for improving their PP content by taking into consideration the customers' emotional states and preferences.

There is an increasing demand for “natural” products by consumers, businesses, scientists and product developers. Trends suggest that the term natural may be colloquially understood to be a plant-based material or ingredient. This study investigates whether this trend could apply to polymers by declaring the derivation of the polymer as a plant or as petroleum. Because polymer materials do provide environmentally positive attributes for products in relation to other materials, such as lightweighting, durability, and lower fabrication energy requirements, it may be helpful to understand the influence of a polymer’s derivation on the perception of a polymer product’s sustainability. The goal of this study is to assess peoples’ relative perceptions of the sustainabilities of polymer drinking cups when the base materials from which the polymers were derived are exposed. A set of six injection-molded drinking cups was given to research subjects to analyze. Each cup is made of a different polymer. The polymers have derivations including petroleum, corn, sugar and trees. Participants evaluated the cups on six qualitative design strategies for sustainability, including natural-ness. This paper compares the perceived sustainable attributes of the cups, and which attributes were most strongly influenced by revealing the derivation of the polymers.

It is said that the relationship between “novelty” and hedonic response is expressed as an inverse U-shape. The latest studies about perception emphasize “novelty” as a factor of emotion and quantify “novelty” by assessing the difference in amount of information using Kullback-Leibler (KL) divergence. In this study, we proposed a novelty index of closed surfaces using KL divergence focusing on their curvatures. To calculate novelty index, we firstly calculated Gaussian curvature of each vertex in the shape. Then, we defined occurrence probability distribution which represents probability that a vertex has a certain curvature. The KL divergence expresses the difference between the occurrence probability distributions of the standard shape and the target shape. To confirm the effectiveness of the proposed index, we conducted the cognitive experiment using the shape samples of an automobile generated by particle swarm optimization method. The coefficient of determination between the proposed index and sensory evaluation values of “difference” were very high which support the applicability of the index. Furthermore, the consideration of location information increased the correlation with sensory evaluation. This suggests the possibility to evaluate an industrial design requirement quantitatively and contributes to develop the automatic shape generation in product design.

In recent years, deep learning has attracted much attention and various techniques have been proposed. GAN (Generative adversarial networks) is one such method. When images are used as the training set, this technique learns to generate new images that are indistinguishable from the images of the training set. Using this method, new face images of non-existent people can be generated from the face images of real people, or input images can be converted into the images with the style of a particular painter such as Monet.

In this study, using this capability of GAN, a method of generating a new product design from the images of customer's favorite products is proposed. The product images that customers evaluated as preferable in the kansei evaluation are used as the training set of GAN. Since the GAN generates images that are indistinguishable from the images in the training set, the new images generated are more likely to be preferred by customers.

In the case study, the proposed method was applied to the chair design. Subjects evaluated their preferences for images of various types of chairs collected from the Internet, and new chair images were generated based on the chare images that subjects evaluated as preferable. The generated chair images were first evaluated in terms of image quality, i.e., whether they looked like chairs or not, and how innovative they were compared to existing chairs, and then subjects evaluated their preferences.

Mouthwash prevents bad breath and sterilizes bacteria that cause dental caries and periodontal disease, so it provides important means for improving oral hygiene in our daily life. Along with active ingredients such as bactericides, mouthwash contains a variety of aromas to mask the bitterness of base ingredients and increase palatability. However, given that numerous aroma studies have demonstrated various physiological efficacies of aromas on autonomic/central nervous systems, it is no wonder that aroma mouthwash has some sort of physiological impact when it is used. In this study, we investigated the effect of aroma in mouthwash on peripheral and cardiac autonomic nervous system activity. The experiment was carried out in a within-subject design wherein 20 healthy women under 5 conditions, which are 4 types of mouthwashes: peppermint, bergamot + peppermint, orange + peppermint, and lavender + peppermint, and water as a control. Participants performed a 20-minute calculation task as a cognitive stressor, and then rinsing by a mouthwash or water. We evaluated the recovery period from acute stress response for 20 minutes after the task. As a result, it was observed that a mouthwash with citrus flavor had a relaxing effect in terms of subjective scores and recovering from physiological stress response. The results may illustrate a potential benefit of using aroma mouthwash.

In recent years, cross-modal effect in which perceptions interact with each other has been drawing attention. In the case of cross-modal effect between vision and taste, the effect of the angularity of shapes on taste has been widely studied while there has been little research on the other features of shapes. Previous research have shown that the emotional valence arisen from visual perception causes cross-modal effect between vision and taste. Therefore, this study focuses on the complexity of shapes as a visual stimulus to influence emotional valence and aims to confirm the cross-modal effect induced by its sensation. First, based on previous research, the hypotheses about the effects of the complexity of shapes on taste were made. Second, by using particle swarm optimization algorithm, closed curve shapes were generated based on curvature entropy, a quantitative index of the complexity of shapes, which indicates the randomness of curvature transition. Third, cup holders, which had these closed curve shapes on their sides, were created by using a 3D printer. Finally, by comparing the tastes of orange juice in these cup holders, the effect of the complexity of shapes on the perception of sweetness, sourness and intensity was confirmed. The results suggest that the complexity of shapes controlled by curvature entropy weakens the perception of sweetness whereas it enhances that of sourness and intensity. This finding can be used for reducing sugar intake in bottle packaging.

An objective assessment method for schizophrenia is highly needed. For this purpose, this study examined the emotional characteristics of facial stiffness, one of the indices for estimating patients’ flat affect states using average faces. First, we conducted an experiment to extract psychological evaluations of facial stiffness in 14 Japanese schizophrenia. 12 medical experts rated the facial stiffness of the patients in 127 videos using a rating program. Meanwhile, their gaze data were collected. In addition, 11 average faces of 14 patients and other one of 14 healthy subjects were created in order to extract the physical characteristics of facial stiffness. The results of the emotional analysis showed that the average faces of the healthy subjects and patients were different proportions of 8 emotions. The faces of healthy subjects were "calm" with a confidence level of 99%. On the contrary, the patient's face contained 15% to 65% of emotions other than "calm," such as "anger" and "sad. In images in which the emotion of "anger" was easily identified, the gazing points tended to be concentrated around the eyes. On the other hand, images in which the gazing points were widely distributed may have been difficult to identify because they were not typical emotional expressions. Therefore, it is suggested that there are two types of criteria for judging facial stiffness in schizophrenia: the emotional expression of "anger" and a mixture of features of "anger," "sad," and "calm.

In recent years, significant research has been conducted on the use of deep learning for prediction of facial attractiveness. These studies are expected to have various applications such as recommendation systems and face beautification. Therefore, it is crucial to improve the prediction accuracy. In this study, to improve the accuracy of facial attractiveness prediction, several convolutional neural network-based models were built using sex-specific datasets. Then, their accuracies were compared. The results showed that VGG19 and VGG16 had the highest accuracies for the male and female face datasets, respectively. A detailed confirmation of the factors necessary for prediction is expected to contribute to the construction of models based on human perceptual characteristics. These models maybe utilized in various engineering applications.

The purpose of this study is to determine the amount of information that is interesting to users on a landing page for the purpose of purchasing a product. In this study, we clarified the amount of information that attracts the interest of users and what kind of information changes the interest of users. Specifically, we created three types of landing pages with different amounts of information for three products, and conducted an impression evaluation questionnaire. The three types of information on the landing pages were: (1) product image, product description and catch copy, (2) (1) with additional campaign information, and (3) (1) with additional product sales results and satisfaction levels. Using the impression evaluation data obtained after the questionnaire survey, we conducted a factor analysis for each product and analyzed the factors that influenced the evaluation. In addition, we clarified the website design that made a difference in the willingness to purchase the products and the evaluation items that caused the difference using t-test. As a result, it was found that users' willingness to purchase increased when information such as campaign information and actual results were included, and that reliability affected the willingness to purchase, especially for products that are used for a long time.

Regardless of going online or physical shopping, expectation disconfirmation due to the discrepancy between pre-purchase and post-purchase evaluation may occur from time to time. Thus, this study takes the tumbler as an example and aims to explore the reasons behind, as well as finding a suitable solution. It is hypothesized that the change of sensory dominance in product interaction from purchase to real use might lead to expectation disconfirmation. More specifically, if the user-product interaction during purchase can be better designed to ensure the consistent sensory dominance with real use, there is a chance to avoid expectation disconfirmation. A set of five scenarios of user-product interaction was first proposed by investigating the daily-use behavior of tumbler users. 15 female participants with the experience of tumbler use were then recruited to experience the traditional product interaction during purchase first. After at least three weeks, they returned to experience the new manner of user-product interaction during purchase. Subsequently, each participant took a tumbler home to use it for four consecutive weeks, during which she was required to provide subjective responses at the end of each week. Results showed that there is no sufficient evidence that ensuring consistent sensory dominance during purchase with real use can avoid expectation disconfirmation. Nevertheless, the findings through interviews support the need for doing so.

In the informatized era in which the graphic language has been the advantaged communication, design and digital culture construct cultural ecology of diversified design together, which promotes that the thinking, concept and aesthetic style of design present unique cultural value. Design public welfare posters with graphic elements full of emotions, namely perceptual graphic elements, based on perceptual image elements with visual innovation and the spreading effect of modern public welfare posters and combining with the updating of media and application of digital technology. The research included the investigation on the relationship of the new media literacy as an intermediary variable and the interest and confusion of aesthetic model. The research aimed to achieve the following goals: Research what effect the AI artistic creation will have on appreciators and whether the effect is great different from that from the creation of human artists. Standing by the needs of mass audience, evacuate new ideas and new modes for public welfare posters design, attempt to surpass the two-dimensional formal graphics, speed up the transformation upgrading of traditional public welfare posters. The result shows that it’s greatly critical to study the psychology of mass audience of public welfare posters and it’s also important to give full consideration to perceptual elements, visual experience and technological media.

Product recommendation systems have been instrumental in online commerce since the early days. Their development is expanded further with the help of big data and advanced deep learning methods, where consumer profiling is central. The interest of the consumer can now be predicted based on the personal past choices and the choices of similar consumers. However, what is currently defined as a choice is based on quantifiable data, like the product features, cost, and type. This paper investigates the possibility of profiling customers based on the preferred product design and wanted affects. We considered the case of vase design, where we study individual Kansei of each design. The personal aspects of the consumer considered in this study were decided based on our literature review conclusions on the consumer response to product design. We build a representative consumer model that constitutes the recommendation system's core using deep learning. It asks the new consumers to provide what affect they are looking for, through Kansei adjectives, and recommend; as a result, the aesthetic design that will most likely cause that affect.

The negative attitudes towards robots scale (NARS) was applied widely in the robot-human interaction research field. However, robots' various occupations and roles have not been discussed when studying the negative attitudes towards robots. In this research, we aim to explore whether the occupation of robots could influence people's negative attitudes towards robots. The service robot and security robot, two types of robots that may be widely used, were used in the NARS-related study for the first time. We conducted the online questionnaire research, including three separate parts: negative attitudes towards robots, negative attitudes towards service robots, and negative attitudes towards security robots. The results of the online survey collected from 114 participants (54 females and 60 males) highlighted differences among the scores of people's negative attitudes towards service robots and the negative attitudes towards robots or security robots. People show the lowest negative attitudes towards service robots. There are no significant differences between the negative attitudes towards robots and security robots. This research supported the hypothesis people show different levels of negative attitudes towards different types of robots in the occupational division. These results provide a helpful indication for the study and design of robots with various occupations in the robot industry.

This paper proposes a method of conveying a driver’s intention to yield by blinking automotive lighting. Generally, headlight flashing indicates a driver’s intention to yield to another driver or pedestrians in Japan. However, this signaling method can have several meanings, including warning other drivers of road dangers such as crashed cars, or informing other drivers of the presence of a pedestrian. There is a possibility of a misunderstanding, increasing the risk of accidents. Therefore, we have investigated an effective method for facilitating communication in traffic. This study focuses on nonverbal communication in which blinking lights convey the driver’s intentions to others. In this study, visibility is not considered. Nine blinking patterns that changed the light colors, waveform, and blinking cycle were presented to fifteen healthy participants, and visual impressions were evaluated to identify the most suitable blinking patterns. These nine patterns indicate a driver’s intention to yield to pedestrians and other drivers. The results indicated that a blue-green blinking pattern in which luminance changes with a triangular wave pattern with a one-second cycle was best conveyed the intention to yield.

Some assistive devices (ADs) aim to ease elders performing daily activities by changing postural and strength requirements. Elders usually have cognitive lacks too, making AD usability perception an important issue. Usability perceptions arise from the affordances conveyed by the AD: design features should act as signifiers transmitting opportunities of how to use the AD. This study assesses the perception of jar openers usability. Eye-tracking (ET) data from 56 subjects were used. Rendered images of 6 ADs were shown, in 2 versions: with and without rubber on the grip area. Each slide showed the 6 ADs, each AD image being an area of interest (AoI) to be tracked. The participants ranked the ADs in the next usability affordances: comfort, effort level, easiness to grip, easiness to use, lid slippery and robustness. For each affordance, they were eye-tracked while deciding the best AD, and afterwards they ranked the other ADs. The ranks were transformed into scores, and their correlation with ET parameters (decision time and number of fixations on each AoI) were studied, along with the effect of the rubber on the scores. Heat maps were also analyzed to identify the signifiers of the ADs that attract attention on usability. The results showed the potentiality of ET to study the perception of ADs usability affordances, and that the addition of rubber on the grip area of the jar openers or the material and shape of the lid area have an effect on their understanding and use.

In a KE study to determine the emotional design features of a traditional product from Turkey, expert designers’ opinions were collected for the span the semantic space and span the space of properties steps of the KE methodology and an additional discussion came up. It was noted that the expert designers raised several questions about KE. They seemed to approach the potential KE results with caution because it was perceived as an intervention to their creative processes.

During the interviews the following arguments came out: Design problems are irrational and wicked and do not have only one solution. What KE offers is scientific, but it may interfere with the creative process needed for designing. It can be useful for novice designers, but some experts think that they already have the intuitive knowledge of Kansei.

After discussing with other KE researchers, about the similar experiences working with expert designers, there was the need to explore the issue in more detail.

First, interviews were conducted with 4 expert designers. A scale was developed to assess their attitudes towards using KE results in design problems. Data was collected with an online questionnaire from 49 experts on the scale. The findings can be used in creating new strategies to introduce KE to design experts in a way that they can be convinced to use KE in their further design processes.

Both the disciplines of design and statistics have promoted projects and research with clear objectives in their field, but for the other discipline, they have been difficult or challenging to fully understand.

In design, there are a large number of projects that provoke a reaction in spectators or users, since they have a spectacular scope and impact, but at a statistical level, their results add little value. On the other hand, statistics in some of the models and applications that are often used, the requirements are highly complex and numerous. This makes it difficult to put theory into practice since experiences or experiments that are so complex and difficult to manage cannot be carried out. In addition, later the communication process for non-experts is difficult to understand due to a large amount of information, as well as on occasions poorly designed presentations.

After understanding the limitations that the two disciplines have, their ability to work together and make the weaknesses of each of them a more complete and holistic solution is evident. Kansei engineering is also a good example since it is a complex design tool, and the only way to advance it, incorporating the use of data, is with the collaboration between designers and statisticians. In this paper, the Data Collection Toolkit is presented as a result of applying Kansei Engineering to unite these two disciplines including for each step some methodologies, resources, and tools for designers.

In this study, we regarded the idea that supplementary variables and Multiple Correspondence Analysis are promising for analysis and visualize complicated relations in Kansei analysis. It could merge several different information tables and then project them into a map. Applying this advantage would make the overall view of Kansei. Design elements, samples, and associated Kansei words were shown in an MCA map. Leather patterns of children’s lower leg orthotics are the objective of this Kansei evaluation. The leather surface was simulated with 3D CG with physically based rendering methods.

The aim of this paper was to systematically explore the knowledge landscape of papers presented at KEER conferences over the last decade. We collected all papers published in conference proceedings between 2010 and 2020. We (i) used a text mining pipeline to extract, clean, and normalize keywords from the Title and Abstract fields, and (ii) created a co-occurrence network reflecting the relationships between keywords. The network was then characterized at different levels of granularity (static analysis vs. time slice analysis and whole network vs. node-level analysis). The exploratory analysis showed a stable expansion of the network over time. The cluster structure revealed several groups of keywords that did not change over time and reflected both domain-specific and method-specific topics of Kansei research.