1: egoDetect: Visual Detection and Exploration of Anomaly in Social Communication Network [poster] [PDF]
| Jingwen Zhang | University of Electronic Science and Technology of China |
| Jiansu Pu | University of Electronic Science and Technology of China |
| Hui Shao | University of Electronic Science and Technology of China |
| Yuwei Zhang | University of Electronic Science and Technology of China |
| Tingting Zhang | University of Electronic Science and Technology of China |
| Shaolun Ruan | University of Electronic Science and Technology of China |
| Yunbo Rao | University of Electronic Science and Technology of China |
| Yadong Wu | Sichuan University of Science & Engineering |
Abstract:
Up to now, there are still many challenges in social anomaly detection.
Supervised methods need tag data, but it is hard to gain.
And unsupervised methods can’t guarantee the accuracy of anomaly
detection. Moreover, Dunbar and Zhou discovered that an ordinary
person’s social network is hierarchical [1, 3]. In addition, the egocentric
network allows experts to learn about the topology and have
an intuitive understanding of the ego’s network. Besides, while
the types of social data are diverse, it is hard to design a suitable
visualization model to detect all of them.
Combining the above questions and thinks, we have designed
a novel visualization system, egoDetect, which combine the time
series and can explore anomalies from both global and local perspectives.
We use the temporal LOF algorithm to filter the data.
Compared to the existing work [2, 4], it can detect anomalies in the
data of social networks without tags. Besides, inspired by the solar
system and the social brain hypothesis, we have designed a novel
glyph to explore an ego’s topology and the relationship between
egos and alters. It can help experts have an intuitive analysis on
egos. We also add friendly and intuitive interactions to help experts
quickly get the information they want.
2: Visualization of Worker Activity in Construction Site [poster] [PDF]
| Hiroaki ISHIOKA | SHIMIZU CORPORATION |
Abstract: In this study, we examined methods to visualize the position data and action data of workers in a construction site into XYZ values and color value in CAD software. We proposed methods to make site manager understand work flow in site and to consider next plan. From comparison of multiple visualization method, we concluded a method, perspective views from directly above of the result of inputting XYZ position coordinates data into XYZ coordinate values and inputting time data into color value as each worker comparison, can be effectively grasped work flow in construction site.
3: JobPlot: Visual Analysis of Abnormal User Behavior Detection in Large-Scale Distributed System [poster] [PDF]
| Shaolun Ruan | Kent State University |
| Qiang Guan | Kent State University |
Abstract: We design and implement JobPlot-an visual analysis tool to explore millions of job-instance batch logs on the cloud computing platform. Our tool is designed to address user-specific behaviors that may cause computing cluster issues. Through the tool, the administration team can be guided to understand the state of system. The visualization system illustrates a novel technique to visually identify the user’s abnormal operation instead of reading through the raw data collected from the cloud infrastructure.
4: Gaze Visualization Embedding Saliency Features [poster] [PDF]
| Sangbong Yoo | Sejong University, South Korea |
| Seokyeon Kim | Sejong University, South Korea |
| Daekyo Jeong | Sejong University, South Korea |
| Yejin Kim | Sejong University, South Korea |
| Yun Jang | Sejong University, South Korea |
Abstract: Visual information such as gaze movement and visual stimuli are clues to analyze human attention intuitively. However, it is not easy to analyze how visual stimuli affect gaze since existing techniques focus excessively on the eye movement data. In this paper, we propose a novel gaze visualization for analyzing eye movements using saliency features as visual clues to express the visual attention of an observer. The visual clues that represent visual attention are analyzed to reveal which saliency features are prominent for the visual stimulus analysis. We visualize the gaze movement data with the saliency features to interpret the visual attention.
5: A Case Study of Data Visualization and Storytelling Workshop for Middle School Students [Honorable Mention Award] [poster] [PDF]
| Puripant Ruchikachorn | Chulalongkorn University,Punch Up |
| Patchar Duangklad | Punch Up |
| Thanisara Ruangdej | Punch Up |
Abstract: Visualization and storytelling skills can be trained during early school years. We aimed to teach visualization knowledge through a workshop whose participants were 100 middle school students across Thailand. With several tangible materials and an assigned topic, a group of five students collected data, mostly from a primary source, produced a visualization and gave a short presentation on data insights. Despite different backgrounds, all groups shared similar visualization types. Concrete and abstract data representations in pictographs, bar charts, and pie charts were popular.
6: Interactive Diffusion Tensor Imaging Fiber Data Visualization via Multiple Devices [poster] [PDF]
| Sitong Fang | Nanjing Normal University |
| Guang Yang | Nanjing Normal University |
| Hailong Wang | Nanjing Normal University |
| Lijun Wang | Nanjing Normal University |
| Yuzhe Xiang | Nanjing Normal University |
| Genlin Ji | Nanjing Normal University |
| Richen Liu | Nanjing Normal University |
Abstract: Diffusion Tensor Imaging (DTI) reveals subtle abnormalities associated with stroke, multiple sclerosis, schizophrenia and dyslexia, which has a broad application prospect in the medical field. The densely sampled 3-D DTI fiber tracts in biological specimens have high geometric, spatial and anatomical complexity. To provide users with more immersive and conve- nient interactions in exploring DTI fibers, we design specific interactions based on the APIs of Leap Motion and Oculus Quest. Leap Motion and Oculus Quest are devices focusing on hand tracking and 3-D somatosen- sory interactions. We design four different interaction modes for users to analyze the data in different interaction stages and scenarios, in order to better explore the DTI fibers which users are interested in by Leap Motion gestures. They are Normal Mode, Box Basic Interaction Mode, Box Logic Operation Mode, and Cluster Exploration Mode. Compared with the ex- plorations through traditional input devices, the evaluation tests show that the proposed approach is more intuitive and efficient in 3-D explorations and provides an immersive experience for users to explore the DTI fiber data.
7: Visual Deep Learning Models Analysis for Air Pollution Predictions [Best Poster Award] [poster] [PDF]
| Hyesook Son | Sejong University |
| Seokyeon Kim | Sejong University |
| Hanbyul Yeon | Sejong University |
| Miyeon Lee | Sejong University |
| Yejin Kim | Sejong University |
| Yun Jang | Sejong University |
Abstract: The output of a deep learning model delivers different predictions depending on the input of the deep learning model. In particular, the input characteristics might affect the output of a deep learning model. In this paper, we propose a visualization system that can analyze deep learning model predictions according to the input characteristics with air pollution data. The input characteristics include space-time and data features, and we apply temporal prediction networks (LSTM, GRU), and spatiotemporal prediction networks (ConvLSTM) as deep learning models. We interpret the output according to the characteristics of input to show the effectiveness of the system.
8: An Interactive Paper Summarization System through Topic Network Visualization [poster] [PDF]
| Hyunwoo Han | Ajou University |
| Hyoji Ha | Ajou University |
| Jaejong Ho | Ajou University |
| Hyeonsik Gong | Ajou University |
| Junyup Hong | Ajou University |
| Soojung Lee | Ajou University |
| Juwon Hong | Ajou University |
| Kyungwon Lee | Ajou University |
Abstract: This study proposes a web system with visualization tools to help users easily explore and summarize topic keywords and contents of specific papers in which they are interested. The system is composed of three views; 1) A view that shows the original paper uploaded by user, 2) A bubble chart view that displays the distribution of topic keywords by each section of the paper, 3) A topic network view that shows the relation of keywords related to the selected nodes from the bubble chart. Through these views, users can not only understand what topic keywords are important in each section of the paper but can also quickly identify the content of the paper by checking the flow of the keywords' network.
9: Scatter Cube for Scatterplot Matrix Navigation in Virtual Environment [poster] [PDF]
| Yong Li | School of Digital Media & Design Arts, BUPT |
| Tiemeng Li | Beijing Key Laboratory of Network System and Network Culture |
Abstract: Scatterplot Matrix is a common model of multi-dimensional data visualization at present, which can visually present the relationship between the dimensions of the data set. In the navigation of scatterplot matrix, how to better understand the transformation process of scatter in the navigation of scatterplot matrix is a great challenge. In this work, an interactive method of scatter cube is developed to navigate in the scatterplot matrix. Three-dimensional cube is used as the visual carrier of scatterplot matrix navigation. The two-dimensional scatter points are projected to three-dimensional for transportation, and projected back to two-dimensional at the end. Spatial thinking is used to enhance users’ cognition of dimension transformation in the process of scatterplot matrix navigation，so that users can better explore the relationship between the dimensions of multidimensional data.
10: eduDig: Machine learning visual diagnosis base on student behavior and performance prediction [poster] [PDF]
| Tingting Zhang | University of Electronic Science and Technology of China |
| Jiansu Pu | University of Electronic Science and Technology of China |
| Yuwei Zhang | University of Electronic Science and Technology of China |
| Yulu Xia | University of Electronic Science and Technology of China |
| Haixing Dai | University of Georgia, Athens, Georgia, United States |
| Jingwen Zhang | University of Electronic Science and Technology of China |
| Hui Shao | University of Electronic Science and Technology of China |
| Shaolun Ruan | University of Electronic Science and Technology of China |
Abstract: eduDiag, a visual analytic system, is presented to analyze student behaviors with academic performance based on smart card data. We used the random forest model to predict students' GPA scores, and selected the characteristics of students' behavior data in school, including access to the library, canteen consumption records, fetching water records and bathing records. However, the results of the model are not very satisfactory, we need visual technology to help us more easily analyze our data and diagnose our model more conveniently, so as to quickly find a breakthrough to improve the performance of the model.
11: Uncertainty Visualization on Geospatial Data for Nitrogen Leaching [poster] [PDF]
| Babak Samani | University of Nebraska-Lincoln |
| Saeideh Samani | University of Nebraska-Lincoln |
| Hongfeng Yu | University of Nebraska-Lincoln |
| Haishun Yang | University of Nebraska-Lincoln |
Abstract: Nitrogen (N) is an essential nutrient for many crops, including corn and soybean. However, its leaching into groundwater is a severe cause of concern for environmental and public health. The amount of N-leaching is closely linked to soil water drainage and rainfall. Prediction of N-leaching in cropping systems is critical to the improvement of crop management through the reduction of N-leaching. Visualizations can help understand uncertainty in the prediction of Nleaching in soil. The uncertainty in N-leaching prediction originates from uncertainty in many parameters, such as weather predictions, soil properties, and the information entered by a user (e.g., N fertilizer). We have developed a platform to assist in comprehending the relationship between various input parameters and N-leaching. Our platform can reveal N-leaching with uncertainty analysis and visualization of different parameters.