Ye Lim Rhie

Evaluating Representativeness of Qualitative Text Data in Identifying UX Issues

ABSTRACT With increasing recognition of user experience (UX), it became important to identify UX issues from customers’ opinions expressed in a form of text that is unstructured qualitative data. Since reliability of qualitative data has little been investigated, it requires much time and effort for researchers to define meanings and values from the qualitative data. Therefore, this article proposes a novel approach to measuring the stability of textual data by adopting semantic network analysis, as shown by our analysis on in-depth interview transcripts and online customer review data. Among the semantic networks generated from text, subnetworks were sampled from the original networks until the representativeness of each sample size was determined. Then, similarities between the subnetworks and the original network were calculated by applying a correlation analysis to determine whether stability occurs. The results indicated that variables of word frequency and network-level statistics affect stability of the network. The suggested method will be able to assist in identifying UX issues.

Understanding characteristics of user-generated content as a source of extracting user value

This study aims to identify and categorize the characteristics of user-generated content to obtain a profound understanding on the user data, and actively utilize the data in extracting user values. Based on literature review, characteristics of online reviews and photos on Social Network Services (SNS) were investigated to evaluate five categories for online reviews (interactive, imprecise, cultural, biased, and ambivalent), and four categories for SNS photos (affirmative, condensed, informative, and archival). These categories were reclassified on the three axes representing individual-social, explicit-implicit, and pleasant-unpleasant, which are well known dimensions of user-generated value. As a preliminary study, this categorization is expected to act as a stepping stone in understanding the user-generated content. It will also help Human-Computer Interaction researchers to select and use data consistent with the research purpose, and could be served as the indexes for measuring the characteristics of the data generated in social media and online websites.

Design Optimization Of Control Layout For Naval Mfc (Multi-Function Console) Using A Modified Layout Analysis Method

The purpose of this paper is to determine the optimal layout of controls for naval Multi-Function Console (MFC). Layout analysis procedure was modified with two major phases. (1) Possible layouts that satisfy four ergonomic principles, which are functional grouping, sequence-of-use, frequency-of-use, and importance of controls, were identified. This process involves interview and survey of actual operators for comprehensive understanding of context of use. (2) To evaluate the appropriateness of suggested layout, heuristic evaluation was conducted after performing the given tasks. As a result, additional constraints could be identified during iterative process to result in three optimal layouts. In heuristic evaluation, all of proposed layouts showed no significant differences (p > 0.05), and they were received satisfactory score. Therefore this study evaluated three optimal layouts that can be adopted in hard key panel of naval MFC. Results imply that suggested process can be adopted to construct optimal layout in various fields.

Development of affective modeling for toilet seat comfort

Since the comfort feeling and satisfaction of a toilet seat is important for the defecation, we investigated the design parameters and affective attributes for toilet seats, and developed the affective model that explains the level of customer satisfaction. To achieve, objective measurements of design parameters and total 19 number of relevant affective attributes including appropriateness of each design parameter were collected from literature review and user interview. Then, a subjective experiment was conducted on five different toilet seat samples. As a result, 6 factors of surface and fatigue, appropriateness of toilet size, accessibility, immaculateness, centroid, and physical tensity were revealed by factor analysis. An affective model for seat comfort was also developed by regression analysis, and the relationships between each factor and design parameters were investigated through correlation analysis. This study is expected to contribute in improvement of toilet seat comfort as a preliminary study.

Use of Hand Biometric Information in Gender Identification Integration of Anthropometric and Electromyographic Data

These days, mobile devices are expected to act as biometric readers due to their increasing number of sensors. However, few studies have suggested a method to identify the user’s demographic profile by collecting biometric information through smart devices, whereas demographic characteristics are an important factor that determines the overall preferences of the interface and content. Therefore, this paper suggests a method to identify the user’s gender based on the biometric information of users. Since the hand is one of the most frequently used body parts in touch interfaces, four hand dimensions and the EMG activities of four hand muscles were measured. The hand length, hand breadth, hand thickness, and hand circumference of 321 Koreans were collected as hand dimensions, and the EMG activities of 27 participants when using mock-ups of mobile phones were observed. As a result of discriminant analysis, hand dimensions could predict males and females with over 75% accuracy, while %MVC showed relatively lower accuracy. The results revealed that simple hand dimensions are a more efficient method to recognize the gender of users. In a further study, other biometric data could be utilized in sensing various types of demographic information considering that identification technologies are growing fast.

Development of an UX Assessment Model Based on Network Analysis

This paper introduces a novel approach to assess user experience (UX) by calculating relation of concepts based on network analysis. Since it is highly subjective and context-dependent, there was only marginal success in building quantification models for UX. This paper suggested UX assessment model to quantify the satisfaction level of user experience by assuming user values could be represented with network structure and centrality measures. As a case study, gesture interface for TV was investigated. Three core values were able to be identified, and an UX assessment model was proposed. In addition, the model was validated via quantitative study. The quantitative study collected objective and subjective performance measurements which are related to the peripheries of core values. The result revealed that identified user values significantly affected the overall level of satisfaction, whereas normalized link strengths were not much coincided with the correlation coefficients. This study suggested that an UX assessment model can be used to reveal important user values that affect user’s perceived satisfaction.

Observing the Smart TV-Viewing Experience by a Diary-Based Observation Method

This study aims to observe user’s experience longitudinally in a natural context of use by using DRM (Day Reconstruction Method). DRM is an observation technique that periodically surveys user’s experience and evaluates it systemically. In addition to emotions collected by existing DRM, we also investigated contextual information using structured questionnaires. Each questionnaire was conducted for one episode which is composed of specific actions that are taken, context of each episode, and detailed description of mood and emotion at that time of episode occurrence with detailed reason. For a case study, DRM for the smart TV-viewing was conducted. As a result, users showed different TV-viewing behaviors depending on the day of the week. DRM also showed that users typically did not use the majority of the technical functions provided by the modern TV sets popularly used today. Moreover, we figured out that many secondary activities were conducted concurrently together with TV-viewing. The study suggested that a diary-based observation technique proposed in this paper can be used to analyze the context efficiently and quantitatively without bias.