Yoshinori Hijikata

Exploring the Relationship between User Activities and Profile Images on Twitter through Machine Learning Techniques

Social media profile images are one of many visual components of users. Moreover, user activities such as posting or chatting are regarded as self-expression behaviors. In this study, we examine Japanese Twitter users to explore the relationship between user activities and profile images. Logistic regression analysis is used to statistically identify and quantify relationships, leading us to conclude that several profile image categories significantly correlate with user activities. Furthermore, we use machine learning techniques (logistic regression, random forest, and support vector machine) to predict whether or not a user belongs to a specific profile image category. Each models performance is evaluated and compared for all profile image categories. Primary results show that users whose profile image includes others faces are more likely to use a replying function but less likely to add url links to their tweets, and that it is the easiest for machine learning models to find their category from their user activities. In short, our findings indicate that visual expression correlates with social media user behavior.

Controlling virtual body size to reduce discomfort caused by inappropriate interpersonal distance

Maintaining personal space is important for person to lead a comfortable social life because an inappropriate interpersonal distance often renders discomfort; however, constantly maintaining an appropriate interpersonal distance is difficult. Controlling the visual distance---rather than the physical distance---solves this problem. Therefore, we propose an interpersonal distance control method with a video see-through system, consisting of a head-mounted display and an RGB-D camera. The system controls interpersonal distance by using simple image processing to change the size of persons in the head-mounted display view. To confirm the effectiveness of this method, we conducted experiments that proved that a reduction in virtual body size of persons in the user view reduced discomfort felt by a user. This means that even the prototype system with simple image processing, can reduce discomfort caused by inappropriate interpersonal distance.

Study on the Relationship between Profile Images and User Behaviors on Twitter

In recent years, many researchers have studied the characteristics of Twitter, which is a microblogging service used by a large number of people worldwide. However, to the best of our knowledge, no study has yet been conducted to study the relationship between profile images and user behaviors on Twitter. We assume that the profile images and behaviors of users are influenced by their internal properties, because users consider their profile images as symbolic representations of themselves on Twitter. We empirically categorized profile images into 13 types, and investigated the relationships between each category of profile images and users behaviors on Twitter.

A Cross-Cultural Analysis of Trust in Recommender Systems

User system trust is critical to the uptake of recommendations, and several factors of trust have been identified and compared. In this paper we present a cross-cultural, crowdsourced study examining user perceptions of nine factors of trust and link the observed differences to trust development processes and cultural dimensions. While some factors consistently instil trust, others are preferred only in certain countries. Our findings and the discovered links are important for design of trusted recommender systems.

Proposal of Network Generation Model based on Latent Preference Topic

People select whom to follow on social networking sites based on the topics that interest them. In this paper, we propose a new generation model for complex networks to mimic peoples following behavior. In our proposed model, a node selects a target node to make a directed link based on the latent topic. We examine the features of the networks generated by our model through computer simulation. In the simulations, we calculate the average path length, clustering coefficient, and power exponent, which are representative evaluation indices of the network, and check whether they satisfy the properties of complex networks.

Understanding how user activities are related to profile images on Twitter through regression analysis

In this study, we explore the relationship between types of user activities (e.g., tweeting, replying) and categories of profile images of Japanese Twitter users. We divided profile images of Japanese Twitter users into thirteen categories, and examine how user activities are related to the categories through logistic regression analysis. We find that several types of user activities are significantly related to some categories of profile images (e.g. users in associate category, which is a group of users whose profile images includes others faces, are found to prefer replying to others). Furthermore, we build logistic regression models that predict who belongs to a specific category or not. The model can accurately predict who belongs to the target category in some categories (e.g. associate category). Our results imply that profile images can be clues to know usage patterns of Twitter users.

Investigation on dynamics of group decision making with collaborative web search

In this paper, we present results of investigation on the dynamics of group decision making - how people discuss and make a decision-with collaborative web search. Prior works proposed systems that support group decision making with web search but have not examined the influence of discussion behaviors especially on the satisfaction levels with the final conclusion. In this study, we conducted a set of experiments to observe discussion behaviors and the consequent satisfaction with the conclusion using our experimental system and a set of questionnaires. The task for each participant was to make a decision on a restaurant. Our primary results revealed (1) the similar activities across all groups at the beginning and the end of the group discussion, (2) a lack of correspondence between the satisfaction with the conclusion and the time spent to reach the conclusion, and (3) the presumption that a member who actively engaged in the activities that were visible for the other members was likely to be voted as a leader in the group discussion beyond the discussion. Finally, we discussed how to implement intelligent systems that aid group decision making.

Extension of Smartphone by Wearable Input/Output Interface with Floor Projection

In this paper, we propose an extension of smartphones with a wearable input/output interface with floor projection (WIIFP). WIIFP has advantages in comparison with smartphones. First, WIIFP allows the user to access information without retrieving the device. Thus, information access is quick and easy in comparison with typical smartphone. Second, floor projection is large and public. Therefore, WIIFP performs well at pasting temporal information on floor and sharing information with other users. Thus, we propose interaction between WIIFP and smartphones to take advantages of WIIFP and comfort mobile computing. Then, we illustrate utilization of WIIFP, which extends the function of smartphones describing a specific scenario. We studied the usability of the extension of smartphones by WIIFP. We obtained feedback of using the smartphone and WIIFP. The combination of the smartphone and WIIFP was found to be suitable for multi-tasking and multi-users.

Adding Search Queries to Picture Lifelogs for Memory Retrieval

A picture lifelog is a type of lifelog that consists of pictures, mainly taken by the user. Recently, users have been able to easily create picture lifelogs because many portable devices such as smart phones have a camera. When a user sees a picture in their picture lifelog, it is sometimes difficult to recall the events related to the picture. Therefore, we proposed to combine search queries on a picture lifelog in order to support memory retrieval. Search queries are input into a web search engine to satisfy a users need for information. Recently, because of the prevalence of smart phones, the opportunity to input search queries has increased to anytime and anywhere. Search queries are stored in a cloud user database such as Google search history. In addition, those search queries imply what the user was thinking at the time. We investigated whether search queries enable a user to recall their thoughts regarding picture lifelogs. Thus, we conducted an experiment to ascertain whether search queries reminded a user of past events. As a result, we reveal that displaying a picture with search queries performed around the time it was taken tends to improve users memories better than its time, location, or emails sent during that time.

Floor Interaction with Wearable Projection Interface Using Hand and Toe

We conducted a user study to unveil the usability of a wearable input/output interface using hands and toes for operating applications such as telephone calls and e-mails. Specifically, subjects performed tasks in the states “hands-free,” “having baggage in their dominant hands,” and “having baggage in both hands” using the proposed system and a smartphone. Then, we evaluated the usability according to a questionnaire, an interview, and the task completion time. The results indicate that hand and toe input in the proposed system were accepted when performing simple button operations such as answering the phone. In addition, hand input in the proposed system was accepted when performing scroll operations such as reading an e-mail. However, when performing accurate button operations such as text entry tasks, hand and toe input in the proposed system were seldom accepted.