We investigated shapes of a cosmetics container, a lipstick container, that reflect consumer preferences for appearance. The target users were women in their teens and 20s. Their preferences were estimated by administering a questionnaire to 64 participants that assessed their impressions of lipstick container shapes. Lipstick container shapes were then generated on the basis of the findings and produced by 3D printing. They were compared with those of actual products likely to reflect consumer preferences for appearance by 43 participants. The results indicate that the generated container shapes create a sense of luxury.

“Resolution” in music theory means the chord progression by the transition from dissonance to consonance. More profound satisfaction is provided to listeners by using the transition than consonant chord only. The authors assumed an analogy between temporal transitions in music and colors and found that the transition could improve consonance in some color arrangements. In this paper, abstract and concrete arrangements of three-color combinations are tested to find suitable ones to enhance the “resolution” effect. It is found that the effect is enhanced in curvy abstract arrangements like camouflage or wave patterns. It suggests that the “resolution” effect is enhanced in curvy or irregular geometries. It is observed that the effect occurs significantly in some animal characters and human faces. It suggests that the observers can feel strange on some animal characters and faces although the color combinations are consonant, and transitions can improve the consonance further.

“Resolution” in music theory means the chord progression by the transition from dissonance to consonance for providing more profound satisfaction to listeners than consonant chord only. The authors have investigated similar effects in temporal color transitions as an analogy of musical perception’s effect. In Part (1), the authors used various display patterns of three-colors combinations, measured the score of the color consonance with a sensory test, and revealed conditions where the “resolution” effect is enhanced. In Part (2), the authors focus on the conditions of gaze control. The fixation point appears as a small white dot on the black screen, and the observers are advised to fix their gaze onto the point. The white dot continues to appear while the observers watch the color arrangements and the transition. The experimental results suggest that the “resolution” effect appears more clearly when the color arrangements are captured in peripheral visual fields.

Generally, in industrial design, sketching work is performed to search for the suitable shape and color for a concept including various impression word combinations. In sketching work, rookie designers take more time to reach shapes and colors that match the concept due to lack of knowledge and experience than skilled designers. Therefore, in this research, we developed a glasses design generation system that infers and visualizes suitable shapes of glasses from impression words using deep learning, aiming at educational effects for rookie designers. The purpose of the study is to clarify the relationship between impression words and shapes of glasses. In addition, to verify the usefulness of the system, we conducted an inference experiment on shapes of glasses for impression words with dimension compression by principal component analysis. As a result, it was confirmed that suitable shapes of the glasses for some impression words was inferred.

The face pareidolia phenomenon, incorrect perception of something that is not a face as a face, affects human cognition and behavior. Although some face pareidolia studies focus on event-related potentials (ERPs) as the physiological indicator, most of them use stimuli uncontrolled context, spatial frequency, and contrast. The physiological response induced by this stimulus might be influenced by contexts other than face-likeness. The current study estimated the ERP component that affects the face pareidolia phenomenon by measuring EEGs while evaluating face-like abstract figures’ face-likeness. Multiple regression analysis was performed to predict the face-likeness evaluation based on EEG data. As a result, the N170 component at the left posterior temporal area and the N400 component at the right frontal area were related to the pareidolia phenomenon.

The influence of the COVID-19 pandemic has increased the opportunities of online meetings. In this study, we analyzed and modeled the consensus-building process in online and face-to-face meetings. As a result, we successfully constructed a model for the consensus-building process and visualized the structural changes using Bayesian network. This showed the possibility of discussing various consensus-building processes quantitatively. Furthermore, we compared the consensus-building process in the online case and in the face-to-face case using time to reach consensus, number of words per unit of time, and percentage of statements by participants. The findings suggest that in online meetings, the amount of statements per unit time may increase because the information that cannot be supplemented by non-verbal information is supplemented by words. In addition, it is easier to determine the participant who would lead the discussion in face-to-face meetings, while in online meetings, the participants tended to speak simultaneously.

The generation of attractive sentences for consumers is an important issue. In this paper, we propose a new method to generate title sentences that are more popular on YouTube, which is gaining large popularity in recent years. In the proposed method, the automatic generation of titles is regarded as a sentence style transfer task; The title sentence generated by the video creator is transformed into a more attractive one. Specifically, first, the latent variable is obtained by the back translation method. In the back translation method, it translates the original title into another language and retranslates it into Japanese. Then, style transfer is performed based on this latent variable to generate a more attractive title sentence. In addition, transfer learning is employed to address the problem of scarce training data. Subjective evaluation experiments have been conducted to show the effectiveness of the proposed method.

In this paper, we propose a new personalization method for knowledge graphs using a graph convolutional neural network. As personalized items user interests are treated. Hobbies are used as an example of interest in this paper. Personalized data is obtained from labeled concepts rather than text. In the proposed method, the knowledge graph constructed from the Japanese WordNet and ALAGIN word co-occurrence dictionaries and GCN (Graph Convolutional Networks) are employed. Whether or a user is interested in each concept in the knowledge graph is estimated by three elements: interest data; graph structure of the hobby as an example; and GCN. As a result of the evaluation experiment, the proposed method obtained good results on predicting unknown subjects of interests.

Laterality is a bias in which humans prefer one side over the other. Previous research has shown laterality in facial direction and facial aspects; right-facing faces are preferred over left-facing faces, indicating a rightward dominance. Right hemisphere is specialized for face recognition, and facial parts of a right-facing face are located on the left side; hence right-facing face is expected to have higher perceptual fluency. This study investigated the relationship between the object’s face-likeness and the preference laterality and the difference in perceptual fluency depending on the object’s orientation, using psychophysical experiment and EEG measurement. As a result, right-facing preference was revealed only for face-like objects. In addition, it was shown that the perceptual fluency of right-facing objects was higher than left-facing objects, regardless of face-likeness. These results suggest that preference laterality is due to the influence of a) facial information processing mechanism and b) cognitive bias acquired through experience.

We have been developing a cognitive assessment system based on an electroencephalography-based brain-machine interface (EEG-based BMI) that uses event-related potentials (ERPs) as a virtual “EEG switch.” ERP reflects a temporal change in attention and is also known as a potential biomarker for assessing cognitive functions for older people even with motor decline. We recorded EEG data from 26 healthy adult subjects (18-79 years old) who performed the target-selection task with the EEG switch and compared the result with that of a standard neuropsychological test, the “Trail Making Test” (TMT). We found a correlation between them, and the result showed that this system has the potential for early detection of cognitive decline in order people.

Sound symbolism is the non-arbitrary associations between pseudowords and meanings. It has been reported that these associations extend to the lexical stimulus (e.g., names and faces); however, it is not clear whether non-native English speakers feel the same way. Experimental results using the stimuli that have existing associations might be affected by language family. In this study, we created high-quality images of Caucasians with generative modeling and then conducted a questionnaire survey on Japanese native speakers. The results showed that names containing round-sounding phonemes could be associated with face images possessing feature amounts related to roundness. The same is true of sharpness. This tendency implies Japanese native speakers also feel the non-arbitrary associations between names and faces.

The purpose of this paper is to extract Kansei to rain and to derive a conceptual model for designing spaces that incorporate rain as a component. In order to achieve this purpose, we analyzed Instagram posts using a text mining method. First, we analyzed the co-occurrence of nouns and adjectives in all posts mentioning rain. Next, we conducted correspondence analysis to analyze the characteristic words related to rain in each season. Finally, a conceptual model for the spatial design of rain was proposed by adding theoretical considerations. This conceptual model allows us to understand relationship between spatial scales, phenomenon caused by rain, and activities in rain. This model is useful for designing architecture, green spaces, and landscapes that incorporate rain.

In this paper, we analyze the characteristics of dialogues with high reputations using the dialogue data of online counseling services. Specifically, the dialogue data in the fortune-telling service and the evaluation data of the fortune-teller by the user were used. We analyzed words that are often used in dialogue, the length of dialogue exchanges, and the frequency of utterances, and obtained the following four findings regarding dialogue by highly evaluated fortune-tellers. First, a role as a counselor rather than fortune-telling itself is more important. Second, interjection words tend to be used. Third, there is a tendency for many statements to be made in succession at the end of the dialogue. These findings, such as avoiding excessive use of technical terms, proper usage of empathic expressions, and advice given at the end of the dialogue, are thought to provide important guidance for future automatic dialogue system development.

In recent years, rough set theory has attracted attention as one of the data mining techniques. However, when there are tens or hundreds of decision rules extracted as a result of analysis, it is difficult to identify the combination of attribute values with high frequency of occurrence (core) and the inclusion relations among decision rules and attribute values. Therefore, the purpose of this study was to develop a notation that reflects the extraction of cores in decision rules and the inclusion relations between attributes obtained by formal concept analysis in the decision rule condition part. This enabled us to discover inclusion relations between decision rules and affinity between cores and attribute values, which were difficult to discover in rough set theory alone. We also verified the usefulness of the proposed notation by comparing the proposed notation with the conventional decision rule notation using two example problems as case studies.

In this study, we conducted an experiment that compared 27 schizophrenia patients and 29 healthy subjects in perception of emotions expressed by an NAO humanoid robot. We programmed the robot to express the eight basic emotions proposed by Plutchik (1984) which were anger, anticipation, expectation, joy, trust, fear, surprise, sadness, and disgust using Teshi et al. (2015)’s method of dynamically changing the robot’s eye color. We found a significant difference between the two groups of subjects in their judgements of the robot’s emotions. The results showed that the schizophrenic patients had difficulty in recognizing the robot’s emotions. In addition, the two groups showed different patterns of emotion misperceptions. The schizophrenic group’s misperceptions mistook positive and negative emotions. The misperception of schizophrenic patients may be caused by the similarity of colors and fast blinking cycles. We consider it resulting from the decline in information processing abilities of the schizophrenic patients.

Emojis are extensively used to convey emotional nuances non-verbally in computer-mediated communication. Previous studies have examined how emojis are used and how they change the expression of a sentence, as well as the subsequent impression of the reader. Regarding whether the writer’s emotion is understood correctly by readers, evidence shows that emojis can enhance specific emotions. However, the effects of variation in the kinds of sentence, emotion, and emoji on the shifting emotion conveyed have not been studied. In this study, we examined the various factors that affect the shift in emotions being communicated. Our results demonstrated that positive emotions can be communicated more correctly compared with negative emotions and that smiley emojis (a.k.a. emoticons) can communicate emotions more clearly compared with other kinds of emoji.

Currently, robots that are designed to communicate with people, such as service robots, are becoming popular. It is known that expressing emotions can make communication smoothly. Therefore, humanoid robots must have the means to express their emotions. In this study, we focus on the body movements of humanoids and aim to add emotions to them. In this paper, we propose RNNPB-Trans. We conducted experiments to generate movement of the wave and bow with emotion using RNNPB-Trans. The results showed that RNNPB-Trans could generate for waves that express emotion used for training, but could not for bows which is an unknown movement. It seems to be better to extract the features from the movement and then transform the movement features by RNNPB-Trans.

With the evolution of technology, the role of online product purchasing and information seeking has become more critical. Although there have been analyses of this process using questionnaires, none of these studies have used actual behavioral data. This study analyzes the information-seeking behavior that contributes to purchasing products on EC sites using users’ access log data captured by a research company. As a result of analyzing 100 pages of the website before visiting the EC site, it finds that it is significant to browse the one’s pages to purchase. The combination of the websites to be browsed is also essential, and it confirms that browsing three types of sites - search site, shopping mall EC site, and the one’s EC site - is significant for purchasing.