We characterize 16 challenges faced by those investigating and developing remote and synchronous collaborative experiences around visualization. Our work reflects the perspectives and prior research efforts of an international group of 29 experts from across human-computer interaction and visualization sub-communities. The challenges are anchored around five collaborative activities that exhibit a centrality of visualization and multimodal communication. These activities include exploratory data analysis, creative ideation, visualization-rich presentations, joint decision making grounded in data, and real-time data monitoring. The challenges also reflect the changing dynamics of these activities in the face of recent advances in extended reality (XR) and artificial intelligence (AI). As an organizing scheme for future research at the intersection of visualization and computer-supported cooperative work, we align the challenges with a sequence of four sets of research and development activities: technological choices, social factors, AI assistance, and evaluation.

Preference learning is often used to adjust the output of LLM after instruction tuning for more desirable outputs. However, since creating datasets for preference learning is challenging, we are considering automatic creation based on user history. As a new method for label estimation in this context, we propose a similar sample search method for preference learning datasets, which compares data with similar labeled samples. This method is based on the relational structure of preference learning datasets, where data are organized as pairs of Chosen and Rejected.

In digital advertising, repeated exposure to the same creative can lead to performance decline, a phenomenon known as creative fatigue. To support the analysis and mitigation of this issue, we developed an interactive visualization system that integrates both textual and visual features of advertisements. The system uses a pre-trained Japanese RoBERTa model to extract text embeddings and LPIPS to evaluate image similarity. Dimensionality reduction with t-SNE and clustering enables interpretable grouping of creatives. We define an effectiveness score based on ad frequency changes after introducing new creatives, and visualize fatigue trends using heatmaps. This system allows advertisers and researchers to intuitively explore how creative attributes relate to ad fatigue, supporting data-driven decisions for creative optimization.

Parallel coordinate plots (PCPs) are a prevalent method to interpret the relationship between the control parameters and metrics. PCPs deliver such an interpretation by color gradation based on a single metric. However, it is challenging to provide such a gradation when multiple metrics are present. Although a naïve approach involves calculating a single metric by linearly weighting each metric, such weighting is unclear for users. To address this problem, we first propose a principled formulation for calculating the optimal weight based on a specific preferred metric combination. Although users can simply select their preference from a two-dimensional (2D) plane for bimetric problems, multi-metric problems require intuitive visualization to allow them to select their preference. We achieved this using various radar charts to visualize the metric trade-offs on the 2D plane reduced by UMAP. In the analysis using pedestrian flow guidance planning, our method identified unique patterns of control parameter importance for each user preference, highlighting the effectiveness of our method.

Large-scale and objective painting analyses have recently gained attention. In particular, analyzing influence between individual painters requires substantial effort and is hard to reproduce due to subjectivity. Despite increasing demand for automatic estimation, this remains unresolved because such influence is complex and often directional, making it difficult to model. In this paper, we develop an interactive system that visualizes, manipulates, and analyses chains of painterly influence as a network. Using 32,401 paintings, the system infers directional links from color and brushstroke features. The resulting network based on color style features captures stylistic lineages such as landscape-focused and portrait-focused streams, while a multifaceted analysis of Picasso shows that Cézanne’s impact appears in brushwork rather than color. Our contributions are twofold: (1) the use of an evolutionary model to assign explicit direction to painter influence and support art historical interpretation, and (2) providing a visualization system that allows dynamic comparison of influence networks based on multiple image features.

In recent years, licensed nursery schools in Japan have been evaluated from various perspectives. The Ministry of Health, Labour and Welfare (MHLW) encourages local governments and operating organizations to undergo third-party evaluations to improve staff morale and gain parental trust. However, the results of these evaluations are usually presented as plain text for each item. This makes it difficult to compare and interpret the data intuitively. To address this issue, we developed a visualization system to support efficient analysis and understanding of third-party evaluation data. The system is designed for educational specialists from organizations and local governments, rather than childcare professionals. We collected evaluation data for licensed nursery schools in Bunkyo-ku, Tokyo, and applied clustering to group similar schools. This system maps the results using color coding to enhance visual clarity. We also implemented co-occurrence network analysis to visualize key expressions and reveal similarities and differences among nursery schools. This system improves the readability of evaluation content and enables intuitive comparative analysis across multiple nursery schools.

Abstract This paper describes SingDistVis, an information visualization technique for fundamental frequency (F0) trajectories of large-scale singing data where numerous singers sing the same song. SingDistVis allows to explore F0 trajectories interactively by combining two views: OverallView and DetailedView. OverallView visualizes a distribution of the F0 trajectories of the song in a time-frequency heatmap. When a user specifies an interesting part, DetailedView zooms in on the specified part and visualizes singing assessment (rating) results. Here, it displays high-rated singings in red and low-rated singings in blue. When the user clicks on a particular singing, the audio source is played and its F0 trajectory through the song is displayed in OverallView. We selected heatmap-based visualization for OverallView to provide an overview of a large-scale F0 dataset, and polyline-based visualization for DetailedView to provide a more precise representation of a small number of particular F0 trajectories. This paper introduces a subjective experiment using 1,000 singing voices to determine suitable visualization parameters. Then, this paper presents user evaluations where we asked participants to compare visualization results of four types of Overview+Detail designs and concluded that the presented design archived better evaluations than other designs in all the seven questions. Finally, this paper describes a user experiment in which eight participants compare SingDistVis with a baseline implementation in exploring interested singing voices and concludes that the proposed SingDistVis archived better evaluations in nine of the questions.

When using exploratory visual analysis to examine multivariate hierarchical data, users often need to query data to narrow down the scope of analysis. However, formulating effective query expressions remains a challenge for multivariate hierarchical data, particularly when datasets become very large. To address this issue, we develop a declarative grammar, HiRegEx (Hierarchical data Regular Expression), for querying and exploring multivariate hierarchical data. Rooted in the extended multi-level task topology framework for tree visualizations (e-MLTT), HiRegEx delineates three query targets (node, path, and subtree) and two aspects for querying these targets (features and positions), and uses operators developed based on classical regular expressions for query construction. Based on the HiRegEx grammar, we develop an exploratory framework for querying and exploring multivariate hierarchical data and integrate it into the TreeQueryER prototype system. The exploratory framework includes three major components: top-down pattern specification, bottom-up data-driven inquiry, and context-creation data overview. We validate the expressiveness of HiRegEx with the tasks from the e-MLTT framework and showcase the utility and effectiveness of TreeQueryER system through a case study involving expert users in the analysis of a citation tree dataset.

The serve is the first attack and the key to the flow of the match in volleyball. This paper presents a visualization technique for the process of practice for a well-controlled serve. In this study, we ask players to hit the serve toward a target and measure the distance from the drop point and the coordinates of the joints. Then, we apply dimensionality reduction to the acquired coordinates and draw them as a scatter plot, and visualize the correlation between serving posture and accuracy by color-coding the points according to serving accuracy.In addition, we are developing a 3D model visualization of the serve, a visualization of the height of the toss, drawing the trajectory of the joints, and visualization of the twist, tilt, and forward lean of the body, which enable detailed check of improvements in the serve. The results of these functions showed that the stability of the players’ serve and the moment when the players’ posture is stable during the serve differ among players.

Ensemble learning that combines multiple weak learners for enhanced performance, is widely used but suffers from low interpretability/explainability. This leads challenges not only in operational aspects like model maintenance and quality assurance but also in addressing societal needs such as fairness and privacy. To tackle this, we propose a new visualization method focusing on the relationship among weak learners in ensemble models to improve understanding of the model structure and its learning processes. In this paper, we defined the relation between weak learners based on a “common sample” in gradient-boosting decision trees, and a visualization method as a three-dimensional graph structure was proposed. Ensemble models trained with synthetic data sets that include typical distribution shifts and real-world open data sets were visualized. As a result, we demonstrated that this approach enables a more accessible understanding of the behavior and structure of ensemble models comprising multiple weak learners, facilitating the identification of overfitting and underfitting through visualization of changes during the training and validation processes.

Observation of precipitation changes is important for a variety of purposes such as predicting river levels. Previous studies for data visualization of precipitation and river water levels plotted graphs and color bars were many stations on a map. Instead of such visualizations on a map, we construct a graph to imitate a connected structure such as a tributary of a river in this study. Our method displays two pseudo-coloring sparklines at nodes of the graph as the stations. The method can visualize the time difference between the increase in precipitation upstream and the increase in river water level downstream. Users can observe precipitation and river water levels at different observation points. Our method uses a Delaunay diagram connecting gauging positions to interpolate and calculate precipitation at river level observation points. This avoids the discrepancy between observation points.In addition, we adjust the amount of visualized information by skipping the display of several observation points based on the similarity of the time-series data at each station, which is calculated by applying the dynamic time-stretching method. The visualization results show that downstream, once the water level rises, it tends to take longer for the water level to drop. In addition, the results show that a time lag occurs between the increase in precipitation and the rise in river levels in the mainstream, while tributaries have little time lag. In addition, data on rainfall and river levels at the same station over multiple periods and their relationship are plotted as scatter plots. The scatter plots make it easier to compare data from multiple periods at the same time than two-tone pseudo coloring sparklines.

The proximity status of the individual poses a significant infection risk since infectious diseases can be transmitted through contact with others. Implementing crowd control measures to alleviate congestion is crucial in preventing infection spread during large-scale events with numerous spectators. To assess effective crowd control, comparing pedestrian behaviors and employing appropriate methods is essential. In this study, we propose a method for visualizing the influence of pedestrian behavior on infection risk by representing proximity as a network.

With the enhancement of relatively inexpensive human motion capture technology using cameras and infrared sensors, the measurement of human body motion has been easier and its applications have expanded dramatically. We are using this technology to develop a visualization system for the assistance of dance skill improvement by visualizing differences and changes in the motions of novice dancers while theur repetitive practice of the same dance. This system applies a body size correction between the dancer and the instructor, a time correction that aligns the timing of the motions, and a spatial correction that aligns the dancer's position and body orientation. The system then applies a clustering process is to the body parts, and visualizes the clustering results. Users can then select multiple arbitrary motions and animate them to check the differences and changes in the dancer's motions. This paper firstly presents the processing flow and functions of the visualization system, and then introduces the results of the motion data of three novice dancers in three different dance genres to verify the effectiveness of this method.

Each player of competitive Karuta has a unique posture and a style of swiping the cards, which are important factors in winning a game that requires the swift acquisition of all 50 cards. It is essential to analyze the characteristics of a player's posture in order to achieve the swift acquisition. While previous studies on the analysis of posture of competitive Karuta required special devices to be attached to the body for measuring brain activity or analyzing wrist acceleration to determine the speed of a player. Such approaches were problematic since they may interfere with the natural movement of the players. To address this issue, we are developing a visualization system that displays the three-dimensional skeletal movements of multiple players in response to a read the poetry of Karuta. We utilize skeletal information extracted from videos using Mediapipe, developed by Google, to compare and analyze the movements of different players. This paper presents the results of our analysis, which compared the swiping movements of multiple players using our proposed system.

The market size of Internet advertising continues to grow. The creation and delivery of effective Internet advertising will become increasingly important in business. On the other hand, it is difficult to achieve an advertising effect if the target setting is inappropriate because Internet advertising is targeted and delivered to a specific group. Therefore, this study aims to discover effective Internet advertising target settings by evaluating Internet video advertising data with abandonment rate, click through rate and conversion rates and by developing a visualization system that takes into account combinations of advertising attributes. As a concrete example, we introduce the results of visualizing 224,355 anonymized records of advertisements distributed by Yahoo! JAPAN Ads and 89,400 anonymized records of advertisements distributed by LINE Ads.

Infectious diseases are typically transmitted through close contact with infected persons. The effective management of overcrowding is a crucial issue for events with a large number of attendees. Since the COVID-19 outbreak, analyzing people flow to recognize pedestrian behavior and walking patterns have been attracted studies. Visualizing crowded high-risk situations for infection at large gatherings is a complex task. It requires an approach that can effectively represent both spatial and temporal features while ensuring that the visibility of walking paths is not significantly compromised. To address these issues, we propose a novel approach for visualizing proximity as a network that represents the distance relationship between pedestrians. We developed the visualization system linking three components: Proximity Network, the walking paths of selected pedestrians from the network, and the temporal statistics of pedestrian traffic. Users of this system can freely select a group of pedestrians from Proximity Network and observe the paths of the selected pedestrians. This procedure enables better visibility of walking paths and an understanding of their spatio-temporal characteristics because only a smaller number of paths are drawn. This paper presents our case study of the proposed method for visualizing pedestrian proximity using real-world people flow data collected at an event site.

Graph layouts are useful for visualizing relationships between data entities. Nodes represent each entity, and edges represent the relationships between the nodes. Hierarchical graph visualization is known as an efficient method to represent the overview of large-scale graphs, where some clusters are generated depending on node properties and are drawn as a set of multiple nodes (meta-nodes). However, it is often difficult to determine the layouts of nodes and edges of large-scale graphs, including hierarchical graphs. Although many graph layout methods have been proposed so far, the quality of the layouts by many of such methods including force-directed layout methods strongly depends on the initial positions of the layouts. Therefore, it is often a difficult task to obtain desired layouts. This paper presents a layout optimization method for hierarchical graphs using a genetic algorithm. Specific metrics for hierarchical graph layouts are used to evaluate the generated layout. This method makes users easy to select a favorite layout from several layouts which are optimized based on the evaluation results. First, hierarchical graph layouts are generated by applying an existing method multiple times. Then, they are optimized using a genetic algorithm. Users can select their favorite layouts from several optimized layouts. Consequently, this method does not require adjusting initial positions of the layouts to obtain good layouts. The paper also introduces examples using human relationship graph datasets, including a co-authorship dataset.

New social problems caused by data bias have been an important issue in recent years. Data bias is sometimes difficult to determine using only quantitative methods, and it requires human decision-making to resolve. Visualization technology can help humans understand data bias and assist with this issue. This paper proposes a visualization method for detecting intersectional bias caused by multiple attributes. The method generates a hierarchical structure by classifying data items according to multiple attributes, and applies a hierarchical data visualization method equipped with band charts. It provides a visualization that gives an overview of the entire data and makes it easy to detect biases in a particular subset of the data. This paper introduces an application to the gender difference in the feeling of temperature with air conditioning systems, and presents the results of participant evaluations to verify the effectiveness of the proposed visualization over baseline implementation.

Food is one of the basic human needs and an important part of human culture. With the development of Internet, it has become increasingly common for people to share recipes on recipe websites. However, in most cases, these recipes contain only text information about the cooking procedures. It is difficult for users to have a big picture about the food materials and cooking methods at a glance. In this paper, we propose a tool for generating infographic from recipe procedure text, to supplement the pure text information on the recipe websites. This tool accepts one step text of the cooking procedures, and then segments the text stream into appropriate units. After that, it extracts the food material, cooking tool, cooking action segmentation and label them with F, T, Ac (F: Food material, T: Cooking tool, Ac: Cooking action) respectively. Then, it analyzes the grammatical relationship between the extracted words. Finally, it matches the words with prepared pictures and place them with appropriate positions in an infographic and represents it to users.

There have been automatic techniques that selects scatterplots from the ones generated from a multi- dimensional dataset as a powerful multi-dimensional data visualization method. One of the criteria used in this method supposes that plots are color-coded by a specific categorical variable in the multi-dimensional dataset. We extended this method and show the visualization results by color-coding the plots according to a specific numeric variable in the multidimensional dataset. This extended method represents a specific numeric variable by color, and generates a set of scatterplots in which all pairs of other two numeric variables are assigned to the two axes. Then, the method automatically selects interesting scatterplots that represent three numeric variables.

Photos can be treated as life logs of photo owners. Photos can be reliable information to estimate patterns of actions and movements of the owners. Based on this discussion, we are developing an interactive technique to explore the recommended tourist spots based on their past personal travel photos. The technique extracts a set of keywords from the photo set applying a generic object recognition and constructs a tree structure to support the exploration of the keywords. When a user selects a set of interesting keywords, the system provides travel information related to the selected keywords. Our previous paper already introduced the visualizations that demonstrate the appropriateness of the structure of the keywords. This paper focuses on the mechanism for interactive travel information retrieval of our system and user evaluations with this system.

Annotation is an upstream process for constructing training data for machine learning tasks. The reliability of annotation is very important for the reliability of machine learning. The annotations vary from worker to worker, and differences in these tendencies may impair the reliability of the data. This is especially relevant for tasks that depend on the subjectivity of the workers. This study aims to realize reliable annotation by observing the annotation results of workers. As a specific example, we applied the annotations of three workers who evaluated facial expressions by the Likert scale on 977 face images as a subject. We verified the reliability of the annotations from the visualization results.

Lyrics are one of the most important components of music, and they have a great impact on the appreciation of songs such as J-POP. Therefore, it is useful to classify and search songs based on lyrics. However, the impression of lyrics is subjective and may be influenced by musical elements other than lyrics, so the criteria for searching for lyrics required by users may vary from person to person. To address this issue, we are working on a research project to support active lyric search by visualizing the distribution of lyrics. Here, it is often difficult to appropriately calculate the distribution of the lyrics because lyrics have a higher degree of lexical freedom than articles and papers. In this study, we propose a method to visualize the distribution of lyrics calculated applying guided LDA (Latent Dirichlet Allocation) that interactively consumes guided words. This method facilitates the iterative visualization of lyric classification results based on the users' viewpoints. It also makes it possible to search for songs by focusing only on lyrics without taking other musical elements into account. Users can observe the differences in individuality and tendency of songs and artists, and the diversity of lyrics, by using the visualization results.

With the diversification of machine learning applications, the quality verification and comparison of training data has been an important process. For example, while performing transfer learning, verification the difference in the quality between the source and the target data can prevent the accuracy of the model from deteriorating. However, training datasets for deep learning is getting larger and larger, and analysis of such datasets is not always easy. As a solution to this problem, we are working on the visualization for training data validation. In this study, we apply dimensionality reduction to the training datasets and display them as scatterplots to realize a visual analysis that can easily detect differences in the quality. Our current implementation draws the regions where the points are concentrated as semitransparent polygons for each label in the scatterplot. Also, the implementation provides a slider to set a threshold for the interactive adjustment of polygon generation. This allows us to observe the differences in the distribution of labels among the training data.

“Skin Transparency” is an important keyword for women of all generations as one of the conditions for beautiful skin. Although, no one has clear definitions of “Skin Transparency”. As it stands, beauty experts are often invited to use visual methods in conducting skin transparency evaluation; however, it has not been still sufficiently clarified which visual properties are related to skin transparency. In this study, we aim to discover the relations between skin image features and sensory evaluations applying real human skin images. Specifically, we investigate “Skin Transparency” by comparing them using the Parallel Coordinate Plots. We observed their complex distributions by the visualization task.

Multi-layer network visualization techniques have been developed so that users can firstly overview the largescale network and then explore the interesting parts of the data. Meanwhile, local features of the networks are often more interesting rather than their overall structures. It often happens with particular kinds of applications such as social networks. We developed a visualization technique for such types of large-scale networks. The technique iteratively applies a graph sampling algorithm to extract small-scale sub-networks from a large-scale network and then visualize the features of the sub-networks as hierarchically arranged icons. User-specified sub-networks are then visualized by applying our own graph visualization technique. Using networks generated from Twitter data, we actually visualize small-scale networks using the proposed method.

Scatterplot selection is an effective approach to represent essential portions of multidimensional data in a limited display space. Various metrics for evaluating scatterplots, such as scagnostics, have been devised and applied to scatterplot selection. This paper presents a new scatterplot selection technique that applies multiple metrics. First, the technique calculates the scores of scatterplots with multiple metrics and then constructs a graph by connecting similar scatterplots. Next, it uses a graph coloring algorithm to assign different colors to similar scatterplots. We can extract a set of various scatterplots by selecting them that the specific same color is assigned. This paper introduces a visualization example with a retail dataset containing multidimensional climate and sales values.

There are a variety of graphs where multidimensional feature values are assigned to the nodes. Visualization of such datasets is not an easy task since they are complex and often huge. Immersive Analytics is a powerful approach to support the interactive exploration of such large and complex data. Many recent studies on graph visualization have applied immersive analytics frameworks; however, there have been few studies on immersive analytics for visualization of multidimensional attributes associated with the input graphs. This paper presents a new immersive analytics system that supports the interactive exploration of multidimensional feature values assigned to the nodes of input graphs. The presented system displays ”label-axes” corresponding to the dimensions of feature values, and ”label-edges” that connect label-axes and corresponding to the nodes. The system supports visual linking operations which controls the display of edges that connect a label-axis and nodes of the graph. This paper introduces visualization examples with a graph dataset of Twitter users and reviews by experts on graph data analysis.

The Semantic Differential (SD) Method is a rating scale to measure the semantics. Attributes of SD are constructed by collecting the responses of participant's impressions of the objects expressed through Likert scales representing multiple contrasting with some adjective pairs, for example, dark and bright, formal and casual, etc. Impression evaluation can be used as an index that reflects a human subjective feelings to some extent. Impression evaluations using the SD method consist of the responses of many participants, and therefore, the individual differences in the impressions of the participants greatly affect the content of the data. In this study, we propose a visualization system to analyze three aspects of SD, objects (images), participants, and attributes defined by adjective pairs. We visualize the impression evaluation data by applying dimension reduction so that, users can discover the trends and outliers of the data, such as images that are hard to judge or participants that act unpredictably. The system firstly visualizes the attributes or color distribution of the images by applying a dimensional reduction method to the impression or RGB values of each image. Then, our approach displays the average and median of each attribute near the images. This way, we can visualize the three aspects of objects, participants and attributes on a single screen and observe the relationships between image features and user impressions / attribute space. We introduce visualization examples of our system with the dataset inviting 21 participants who performed impression evaluations with 300 clothing images.

Users often choose songs with respect to special situations and environments. We designed and developed a music recommendation method inspired by this fact. This method selects songs based on the distribution of acoustic features of the songs listened by a user at particular places that have higher ordinariness for the user. It is important to verify the relationship between the places where the songs are listened to and the acoustic features in this. Hence, we conducted the visualization to explore potential correlations between geographic locations and the music features of single users. In this paper, we designed an interactive visualization tool methods and results for the analysis of the relationship between the places and the acoustic features while listening to the songs.

People flow has useful knowledge in various fields including traffic, disaster prevention, and marketing. Such information can be obtained by looking for the movement patterns of each person and the features of each place.We have proposed a visualization method that summarizes the paths of a large number of pedestrians in order to discover their useful information. However, we have not been working on visualization focusing on walking states whether people are moving rapidly or slowly. This paper presents a visualization method based on the walking states of a large number of pedestrian paths measured by using cameras. The visualization tool allows users to inspect walking people based on walking states from spatial and temporal views. The spatial view applies a heatmap and the temporal view adopts a stacked line graph. In addition, the user interface has a function to simultaneously visualize changes in the walking state distribution over time and walking paths.

Graph drawing readability metrics are routinely used to assess and create node-link layouts of network data. Existing readability metrics fall short in three ways. The many count-based metrics such as edge-edge or node-edge crossings simply provide integer counts, missing the opportunity to quantify the amount of overlap between items, which may vary in size, at a more fine-grained level. Current metrics focus solely on single-level topological structure, ignoring the possibility of multi-level structure such as large and thus highly salient metanodes. Most current metrics focus on the measurement of clutter in the form of crossings and overlaps, and do not take into account the trade-off between the clutter and the information sparsity of the drawing, which we refer to as sprawl. We propose an area-aware approach to clutter metrics that tracks the extent of geometric overlaps between node-node, node-edge, and edge-edge pairs in detail. It handles variable-size nodes and explicitly treats metanodes and leaf nodes uniformly. We call the combination of a sprawl metric and an area-aware clutter metric a sprawlter metric. We present an instantiation of the sprawlter metrics featuring a formal and thorough discussion of the crucial component, the penalty mapping function. We implement and validate our proposed metrics with extensive computational analysis of graph layouts, considering four layout algorithms and 56 layouts encompassing both real-world data and synthetic examples illustrating specific configurations of interest.

Light art represents various objects by a light stroke drawn in the air. It takes about 10 to 30 seconds to create a light art picture. We could create a light art movie by binding a set of pictures; however, it is a time-consuming task because we need a large number of frames to create a movie and difficult because a person would need to author a sequence of temporally consistent frames. To solve this problem, we are developing a virtual and interactive light-art-like system. This system extracts edges of human bodies from depth information and then applies a point-to-curve algorithm to mimic hand-drawn figures. Finally, the system applies neon-like drawing and a visual effect on the figures and displays them in real-time. This paper also introduces our artwork produced with this system, and a user evaluation that shows that the artwork conveyed happiness and excitement to the audience.

Scatterplot matrix and parallel coordinate plots are well-used multi-dimensional data visualization techniques. These techniques have a problem that they need a very large screen space when an input dataset has an enormous number of dimensions. To solve this problem, we propose a method for selecting important scatterplots from all scatterplots generated from input datasets and for drawing the scatterplots as ”outliers” and ”regions enclosing non-outlier plots.” The technique is useful for users to determine whether to delete outliers from the datasets and form mathematical models of non-outlier plots. This paper introduces an example of visualization using this technique with a retail transaction dataset and climate values.

This paper presents a visualization technique for multidimensional datasets containing real and categorical variables. Supposing multidimensional datasets containing real and categorical values, this technique displays a set of axes corresponding to the dimensions of real values. The technique evenly divides the axes into several ranges and displays component bar charts there. It brightly draws the component bar charts if association rules are applied at the corresponding ranges of the dimensions of real values. As a result, this technique highlights association rules so that users can discover important relationships between real and categorical variables in multidimensional datasets. This paper introduces an application of the presented technique to painting image collections. This application visualizes image features and categorical information of painting images and provides a user interface to browse the painting images associated with the multidimensional values. This paper also introduces user evaluation results of the user interfaces for painting image collections.

Visual cluttering is still a severe problem of large-scale network visualization techniques, and therefore many improvements on network visualization have been presented. Meanwhile, many network datasets in our daily life contain label information. We are developing a network visualization technique which overlays a node-link diagram for visualizing the connectivity and a set visualization for representing the label information. Our technique firstly places a set of nodes in an input dataset into a 3D space and provides an interactive mechanism to manipulate viewpoints in the 3D space. The technique then projects the nodes onto a 2D plane and generates a Delaunay triangular mesh connecting the nodes which have the user-specified label on the 2D plane. Finally, it displays the outer boundary of the triangular mesh to represent the region enclosing the nodes which have the label. In addition, supposing that multiple weighted labels can be assigned to the same node, our implementation draws nodes like pie charts to clearly represent the weights of labels assigned to the nodes. This paper shows examples of the visualization applying a co-authorship network dataset.

Virtual Reality (VR) provides immersive environments where users watch objects and scenes as if they were in front of users. This is a significant advantage while applying to 3D visualization systems. However, it is often difficult for novice users to make the best use of VR to reproduce the objects and scenes precisely and the comprehensively depending on the skills of users and the complexity of objects and scenes. One of the problems is that the allowed operations in VR are too flexible. Here, we propose a viewpoint selection method that allows every user to watch the objects and scenes easily and carefully in a VR space. The viewpoint is not fixed but can be shifted to the most favorable position according to the demand of users. Regarding the selected viewpoint as the initial position, users can easily go towards the target part with fewer movements. This system helps users to understand the complex objects in the VR space precisely with less time. We demonstrate our system on the observation of mode water pairs and discuss the comparative visualization results.

Mode water forms a 3D region of seawater mass, which has similar physical characteristics values. Research and observation of mode water have a long history in physical oceanography because analysis of mode water brings the understanding of various natural phenomena. There have been various definitions of mode water, and comparison of mode water regions extracted with such various definitions is an important issue in this field. This paper presents our study on comparative 3D visualization tool for the comparison of mode water regions. We extract pairs of outer boundaries of mode water regions as isosurfaces and calculate dissimilarity values between the pairs. The tool visualizes the multi-dimensional vectors of the dissimilarity values by Parallel Coordinate Plots (PCP) and provides a user interface to specify particular pairs of mode water regions so that we can comparatively visualize the shapes of the regions. This paper introduces our experiment on a comparison of mode water regions between an observation and a simulation datasets using the presented tool.

A large number of organizations and individuals use Social Networking Services (SNS) to spread announcements nowadays. Among SNSs, Twitter is particularly well-used to spread announcements and messages easily. Various commercial organizations including entertainers, fashion brands, and enterprises use Twitter to announce news and other information. Highlighted users who tweet actively on Twitter are called “influential users.” Many users around influential users often spread their messages by the “retweet” function. This study visualizes the relationship between the tweets by the influential users and the surrounding user groups applying dendrogram and heatmap. From the visualization results, we found groups of users surrounding the influential users, numbers of retweets, and intervals between original tweets and their retweets depend on contents of tweets and aspects of surrounding users.

Thanks to the recent spread of smartphones, tablets and digital cameras, people can take photos easily anytime, and anywhere. Therefore, photos can be a tool to record life logs. We can estimate patterns of actions and movements of people by analyzing their photos. Based on this discussion, we are developing a technique to recommend tourist spots based on the estimation of users' preferences of traveling plans from their past personal travel photos by using generic object recognition. We applied a generic object recognition system to acquire keywords of subject information taken in the photos and represented the co-occurrence of the keywords by a graph visualization technique. This paper presents our travel recommendation technique and a visual user interface that represents a graph with travel photos based on our graph visualization technique. This paper also introduces our case study with real travel photos.

Social media analysis is helpful to understand the behavior of people. Human behavior in social media is related to time and location, which is often difficult to understand the characteristics appropriately and quickly. We chose to apply virtual reality technologies to visualize the spatio-temporal social media data. This makes us easier to develop interactive and intuitive user interfaces and explore the data as we want. This paper presents our visualization of tweets of microblogs with location information. Our system features a three-dimensional temporal visualization which consists of the two-dimensional map and a time axis. In particular, we aggregate the number of tweets of each coordinate and time step, calculate scores and display them as piled cubes. We highlight only specific cubes so that users can understand the overall tendency of datasets. We also developed user interfaces for operating these cubes and panels which indicate details of tweets.

This paper proposes a new image retouch system "CrowdRetouch" which reflects users' tendency of image retouch for a set of similar photos. CrowdRetouch firstly asks initial users to manually retouch sample training images and then divides the initial users based on the image retouch parameters. It then applies regression analysis to each of user clusters to solve the relationship between the retouch parameters and image features and automatically retouches rest of similar photos based on the regression analysis results. After forming the user clusters, CrowdRetouch specifies the clusters of new users with a smaller number of training images by visualizing the learning processes, and therefore we do not need to require heavy preprocesses to the new users. CrowdRetouch realizes personalized automatic image retouching to a large number of photos while reflecting preferences of novice users. This paper introduces our user experiments which demonstrate the parameter visualization is effective for appropriate learning of users' preferences.

We1 often want to track back the conversation logs, such as meeting or group chats which users were late to participate, and timelines of social network services. It often happens to be difficult to find important remarks if the logs are very long. Micro-blogs (e.g. Twitter) is a typical example. We often miss to find important remarks if we have large number of followers. Also, it often happens that many topics are mixed in a single timeline; this situation may also make us difficult to find important remarks. This paper presents a visualization tool which briefly displays the flow of conversations. This tool provides animations of flow of topics and speakers/writers applying a force-directed visualization technique.

We previously presented a crowd-powered digital contents evaluation system. This system shows a lot of pictures to the answerers and ask them to input the evaluations. It preferentially selects pictures which are predicted to be highly or poorly evaluated to the answerers, based on our assumption that high or poor evaluations are more informative results comparing with moderate evaluations. We have applied an interactive genetic algorithm in our system to select such pictures. This paper presents a technique for ranking and visualization for the evaluation results collected by our system. The presented technique calculates scores of all contents and uses for the ranking. Here, it may happen that some pictures are shown to no answerers while using our evaluation system. Our technique presented in this paper estimates the evaluation of such pictures shown to no answerers, and finally complete the ranking of all the pictures. The paper also presents the visualization tool for the ranking of pictures, and our experiment to demonstrate the effectiveness of our technique.

Fluid dynamics simulation is often repeated while changing conditions, and therefore we need to compare a large amount of results. In order to compare results under different conditions, it is effective to overlap the streamlines generated from each condition in a single 3D space. Streamline is a curved line which represents a wind flow. This paper presents a technique to automatically select and visualize important streamlines suitable for comparison of the simulation results. In addition, we present an implementation to observe the flow fields in virtual reality spaces.

We have previously presented a visualization technique which represents multi-dimensional data as collection of lowdimensional parallel coordinate plots. This paper presents a general-purpose extension of the visualization technique which represents multi-dimensional data as a combination of the scatterplots with parallel coordinate plots. We aim to automatically select a small number of pairs of variables which are estimated that they bring interesting visualization by scatterplots. This paper also presents an application of this visualization technique to Multi-objective optimization of manufacturing design. Our multi-dimensional data visualization technique effectively assists us to understand the distribution and correlation of design variables and objective functions in multi-objective optimization processes.

Heatmap is one of the effective representations for time-varying data visualization. It may require large display spaces when an input dataset contains large number of data items or time steps. We may often want mechanisms to interactively filter non-important data items or time steps, so that we can form appropriate sizes of heatmaps and focus on important data items or time steps. This paper presents a heatmap-based time-varying data visualization technique featuring an interactive mechanism to display meaningful data items and time steps. This technique firstly calculates distances between arbitrary pairs of data items, and constructs a dendrogram consisting the data items. It then generates clusters of the data items and displays the data items belonging to the specified sizes of clusters in the heatmap, so that we can focus on groups of similar or correlated data items. It applies a similar mechanism to a set of time steps so that we can remove outlier time steps from the heatmap. Our implementation features two scatterplots, which represent distribution of data items and time steps respectively, and slider widgets to interactively adjust the thresholds of the clustering process. We can intuitively understand how clusters of data items or time steps are constructed, by looking at the scatterplots while operating the sliders.

People flow information brings us useful knowledge in various industrial and social fields including traffic, disaster prevention and marketing. However, it is still an open problem to develop effective people flow analysis techniques. We suppose compression and data mining techniques are especially important for analysis and visualization of large-scale people flow datasets. This paper presents a visualization tool for large-scale people flow dataset featuring compression and data mining techniques. This tool firstly compresses the people flow datasets using UniversalSAX, an extended method of SAX (Symbolic Aggregate Approximation). Next, we apply natural language algorithms to extract movement patterns. Finally, we visualize trajectories of people flow and extracted features to represent popular walking routes and congestions.

Regression analysis has been widely applied to various academic and industrial fields. Applications of regression analysis include medical problems such as health estimation, environmental problems such as disaster prediction and energy consumption estimation, and business/economic analytics. Here, accuracy and quality of regression analysis strongly depend on relevancy between input explanatory variables and actual objective functions. It often happens that several explanatory variables are well correlated with objective functions while others do not well correlated, and therefore accuracy of regression analysis may improve by removing unnecessary explanatory variables. This paper presents a scatterplot-based regression analysis tool. This tool visualizes the distribution of errors between actual and estimated values of objective functions, and provides user interfaces to explore the relationships between explanatory variables and the errors. This paper introduces examples of the visualization results using the presented tool with actual and estimated revenues at a store.

Directed graphs are used to represent variety of information, including friendship on social networking services (SNS), pathways of genes, and citations of research papers. Graph drawing is useful to intuitively represent such datasets. This paper presents an edge bundling and a node layout technique for tightly and mutually connected directed graphs. Our edge bundling technique includes three features: ordinary bundling of edges connecting common pairs of node clusters, convergence of multiple bundles connecting to the same node cluster, and shape adjustment of two bundles connecting the same pair of node clusters. This paper includes a case study with a directed paper citation graph.

There have been many collective knowledge on the Web, such as evaluation of restaurants, hotels, and manufactured products. Even though each of the participants on such Web sites usually just evaluate the small number of contents, these kinds of crowd-powered contents evaluation services bring us fruitful information. Visualization is a useful tool to carefully observe the evaluation results and discover complex trends of the evaluation. This paper presents our study on visualization of the crowd-powered contents evaluation. Firstly we developed a contents evaluation technique applying an interactive genetic algorithm, which presents contents estimated to be highly or poorly evaluated. Then we had a case study with various appearances of female face images to collect the evaluations. Finally, we visualized the result by applying an image browser CAT. This paper discusses how the visualization result depicts the trends of the evaluation on appearance of women.

Many public offices and companies manage their energy consumption by Building Energy Management System (BEMS). It is not an easy task to determine whether the past energy consumption was really necessary or just wasted. Visualization for energy consumption is useful to understand the situations of energy consumption to determine their necessity. This paper presents a visualization tool for energy consumption with BEMS. The tool firstly divides the daily variation of the energy usage and environmental measurements (e.g. Temperature and humidity) into the meaningful number of patterns. It displays "long-term polyline chart" to represent the frequency of the daily pattern so that users can easily focus on particular dates at particular places. It also displays " one-day polyline chart" to represent the daily variation of the recorded values of the particular dates and places specified by users' click operations. The paper introduces the examples of visualization to demonstrate the effectiveness of the presented tool, with a real dataset of business office building.

Survey of research papers is not an easy task for novice researchers, because they are not always good at finding all appropriate keywords for the survey. Moreover, it is not easy for them to understand positions of papers in their research fields instantly, even when they use famous search engines like Google Scholar, it may often take a long time for them to find scholarly literature. On the other hand, many researchers have presented citation visualization techniques for surveying research papers. However, it is still often difficult to observe the complicated relations across multiple research fields or traverse the entire relations in their interest. In this paper, we proposed a visualization technique for citation networks applying topic-based paper clustering. Our technique categorizes papers applying LDA (Latent Dirichlet Allocation), and constructs clustered networks consisting of the papers.

Many people (especially women) tend to take relatively longer time for shopping. This paper presents a system for product retrieval inspired by psychology of women's shopping activity, and an implementation of the system for apparel products. Our study supposes products which pre-defined keywords are assigned, and prepares icons representing the combination of the keywords. The system intuitively displays various icons in a display space to demonstrate the diversity of the products. When a user selects an interested icon, the system switches the display to a set of images corresponding to the selected icon, so that the user can visually compare the similar products. The system also features user interfaces to input the preference of users, and reflects the input to the evolutionary computation which adjusts the selection of icons to their preferences. It acts real shopping behavior because we often firstly look over the shops to understand the diversity of products, and then close up the particular groups of the products. This paper introduces an experiment which demonstrates the preferable assistance of the shopping behavior of women who are interested in various products.

Web browsing history of people who use internet everyday represents processes they learned. Visualization of summarized browsing history can contribute to learn what we did. This paper proposes HistoryPaper, a system to extract representative Web pages from browsing history of users, and arranges them like a newspaper. The system helps users to remember what they did in a particular day. HistoryPaper firstly clusters browsed Web pages based on their contents. Then, it calculates the priority of Web pages from search terms and rality of accessed Web pages, and selects the representative page of each cluster. HistoryPaper arranges the representative pages by utilizing a layout algorithm following the magazine style applied in many news sites. This paper presents the numeric evaluation of the layout results to demonstrate the reasonableness of the algorithm.

We commonly represent time-varying values as polyline charts or heatmaps; however, both type of techniques are difficult to simultaneously observe short-term features of time-varying values and cluster transitions. This poster proposes storyline-based visualization technique for consecutive numerical time-varying data. Storyline is a visualization technique to show associative feature among elements over time. Our technique measures similarity of each elements and draw similar elements as proximity storyline. The technique also reflects differential values on storyline as a visual variable to emphasize the amount of line changes.

The author's laboratory has many joint projects on analysis and visualization of temporal daily life data, such as transactions of credit cards, Web access logs, walking paths at public spaces, and measurements for building management (including population, temperature, and electric power). We often discover interesting behavior from such visualization results, and feel several results demonstrate Japanese specific behavior and mentality. This talk introduces some of our visual analytics results which denote such Japanese specific behavior and mentality. Due to the space limitation, this manuscript just introduces two of the examples, though the talk may introduce more examples.

H.265/HEVC standard, promising up to twice the compression efficiency over H.264/AVC standard is suitable for encoding UHD videos and has garnered much attention since its inception. With increasing amount of devices supporting UHD, real-time H.265/HEVC encoder and decoder are needed to complete the UHD media ecosystem. In this paper, we present a highly parallelized HEVC software encoder suitable for broadcasting or network streaming applications. Real-time Main 10 profile encoding of 4K UHD videos at 60fps is achieved through an efficient parallel encoder platform comprising of CPUs interconnected by high speed network. The encoder core is optimized with data parallelism and GPU assisted motion estimation. Utilizing temporal parallelism of GOP and spatial parallelism of picture through slicing to encode, scalability and flexibility are achieved. Results show that our encoder system is about 15794 times faster than HEVC test model HM 11.0 and 13 times faster than ×265, an open source HEVC encoder.

Multi-objective optimization tools have been applied in various academic and industry fields. It is often difficult to optimize all the objectives since they often cause trade-offs. It is also difficult to figure what kinds of trade-offs actually cause. We think visualization of multi-objective optimization results assists users to intuitively understand the distributions of their solutions. This paper proposes a visualization of explanatory variable and objective function spaces in the separate views, so that users can easily understand their relevancy. Also, our tool features the linkage mechanism between the two views. When a user selects certain ranges of the values by a mouse click operation, the tool highlights all the corresponding individuals. This mechanism is useful for users to narrow down the results. We expect the tool assists the understanding of the behavior of the optimization processes and improvement of the future processes.

Protein is the major component of the organism. It has a unique three-dimensional structure determined by its amino acid sequence. A concave (pocket) on the surface of a protein is known to be the best target for a drug to react. We started analyzing how" drug ability" of proteins related to the location of amino acids in a pocket. For as tarter of the study, this paper presents a visualization tool for distance analysis between pockets and the amino acid residue. Provided that a protein surface is described by a triangular mesh, this tool first identifies pockets on the protein surface, specifies the deepest point and outer loops of the pocket, and calculates distances between atoms of an amino acid residue and the deepest point or the outer loops of the pocket. The tool then visualizes the statistics of the distance calculation results by poly line charts and the distribution by scatter plots. This paper proposes a biological interpretation of the visualization results.

Most time-varying data in our daily life is multi-variate. Moreover, most of such time-varying data contains both numeric and categorical values. It is often meaningful to visualize both of them as they are often correlated. We aim to visualize every value in such time-varying data in a single display space so that we can discover interesting relationships among the values of the time-varying data. This paper presents a heat map-based time-varying data visualization technique which displays both numeric and categorical values in a single display space. The technique assigns time to the horizontal axis of the display space, and vertically arranges the series of colored belts corresponding to the time-sequence values. It generates one belt for a numeric value, and multiple belts for a categorical value. It clusters the belts according to the similarity of color sequences, and re-arranges the belts based on the clustering result. This paper shows an example of the visualization result applying a time-varying multi-variate marketing dataset.

Thanks to recent evolution of portable music players featuring large storage spaces, we tend to carry large number of tunes. It often makes us more bothering to look for tunes which we want to listen to. On the one hand, we usually listen to the tunes on the music players by manually selecting playlists or album names, rather than manually selecting each tune one-by-one. This paper presents GRAPE, a playlist visualization technique being used as user interfaces on the music players. GRAPE presents a set of tunes as a gradation image, by assigning colors to the tunes based on their musical features, and placing them onto a display space by applying Self Organizing Map (SOM). This paper describes the processing flow of GRAPE, and introduces user evaluations to demonstrate the effectiveness of GRAPE.

It is a common problem for many people to keep their skins fine, and there have been huge number of cosmetics products. We have developed of image recognition and geometric modeling technique for impression analysis of human skins [1]. The technique firstly extracts parameters of micro-geometry of human skins from real photographs, as shown in Figure 1. It then constructs similar micro-geometry of human skins by a polygon-based shape modeling technique with the parameters including radii of pores and directional distribution of furrows, extracted from the real photographs, as shown in Figure 2. This poster introduces the impression analysis results of the skin images generated by our skin simulation technique.

A Photomosaic arranges many small photographs to represent a large image. Our study applies the photomosaics to a photograph browser CAT. Our implementation displays photomosaics while zooming out, and individual photographs while zooming in. Here, many photograph browsing software displays a set of photographs in the order of their times. To maintain this order of photographs, our photomosaic generation technique firstly arranges the given set of photographs in the order of those times, and then retouches so that the set of photographs forms a photomosaic-like scene. This paper presents our technique for photomosaic generation, and a user evaluation to discuss what kinds of photographs are preferable to be applied. We think this discussion should be fruitful for our future development of automatic photograph selection for photomosaic generation.

Multiple-objective optimization problems have multiple objective functions to be optimized simultaneously. It is often difficult to optimize all the objectives since they often cause trade-offs. We produce a visualization tool which displays both explanatory variables and objective functions for multi-objective optimization.

Protein is the major component of the organism. A hole (pocket) on the surface of a protein is known to be the best target for a drug to react. We started analyzing how "drug ability" of proteins related to the locations amino acids in a pocket. This poster presents a visualization tool for a distance distribution analysis between two types of amino acids in a pocket and proposes the biological interpretation of the visualization results.

Thanks to the recent evolution of movie- and sensor-based human tracking technologies, we can obtain and accumulate a set of walking paths ("trajectories" in this paper) of people over a long period in various places. Such people flow datasets are useful for many fields, including analyses of customer behavior, effectiveness of advertisements, and operational efficiency. This paper presents a linked visualization system to assist in the discovery of new knowledge by analyzing the accumulated people flow datasets, and a case study using this system. In this study we suppose the people flow datasets consist of a set of trajectories and temporal flow quantity. The system consists of two visualization components: classified trajectory visualization, and temporal flow quantity visualization. The former component classifies trajectories into several patterns applying the spectral clustering algorithm, and visualizes the patterns by colors on a physical space. The latter component displays temporal flow quantity of the above patterns applying a piled polygonal chart. This paper introduces a case study applying a movie-based human tracking dataset to the presented system.

Analysis and monitoring of system logs such as transaction logs and access logs is important for various objectives including trend discovery, update effort determination, and malicious behavior monitoring. However, it is not always an easy task because these logs may be massive, consisting of millions of records containing tens of variables, and therefore it may be difficult or time-consuming to discover significant knowledge. This paper presents a visual analytics tool which enables us to effectively observe system logs. The tool recommends variables that can reveal interesting discoveries and provides feature-based filtering that selects meaningful items from the visualization results. This paper also presents the result of experiments for non-professional users.

Multidimensional data visualization is an important research topic that has been receiving increasing attention. Several techniques that use parallel coordinate plots have been proposed to represent all dimensions of data in a single display space. In addition, several other techniques that apply scatter plot matrices have been proposed to represent multidimensional data as a collection of low-dimensional data visualization spaces. Typically, when using the latter approach it is easier to understand relations among particular dimensions, but it is often difficult to observe relations between dimensions separated into different visualization spaces. This paper presents a framework for displaying an arrangement of low-dimensional data visualization spaces that are generated from high-dimensional datasets. Our proposed technique first divides the dimensions of the input datasets into groups of lower dimensions based on their correlations or other relationships. If the groups of lower dimensions can be visualized in independent rectangular spaces, our technique packs the set of low-dimensional data visualizations into a single display space. Because our technique places relevant low-dimensions closer together in the display space, it is easier to visually compare relevant sets of low-dimensional data visualizations. In this paper, we describe in detail how we implement our framework using parallel coordinate plots, and present several results demonstrating its effectiveness.

While cooking is a daily activity in our life, many family cooks feel bothered by deciding the menu each and every time because of being required to cook both tasty as well as well-balanced meals for their families. We have been developing a meal recording system, MEAL (Menu Evaluation and Attribute Log), which supports those cooks by easily managing and retrieving meal records on mobile platforms. MEAL displays automatically generated icons as composed representations of information based on food data, so users can easily identify each data item on a small screen. The interface has two main views, a retrieval view and an analyzing view for food records in the user's own food database. The retrieval view displays retrieval results with photos of food and symbolic icons. The analysis view supports visualizing all sets of food data represented by icons. It allows users to easily understand their cooking habits for their family. Since the information gets optimized, the symbolic icons lead to higher visibility on small screens.

A gene network is constructed with genes as nodes, and interactions between genes as edges so as to reveal unknown gene functions and relationship. However, nodes and edges of gene networks are usually very numerous. Because of that, it may be difficult to understand relations between genomic functions and gene-gene interactions, if it is visualized by traditional techniques. This paper presents our technique on visualization of gene networks and gene ontology(GO), which summarizes gene functions and attributes. The technique represents the functions defined by GO as colors of nodes, and bundles edges depending on the gene functions to ease visual complication of the network.

We have various interesting time-varying data in our daily life, such as weather data (e.g., temperature and air pressure) and stock prices. Such time-varying data is often associated with other information: for example, temperatures can be associated with weather, and stock prices can be associated with social or economic incidents. Meanwhile, we often draw large-scale time-varying data by multiple polylines in one space to compare the time variation of multiple values. We think it should be interesting if such time-varying data is effectively visualized with their associated information. This paper presents a technique for polyline-based visualization and level-of-detail control of tagged time-varying data. Supposing the associated information is attached as tags of the time-varying values, the technique generates clusters of the time-varying values grouped by the tags, and selects representative values for each cluster, as a preprocessing. The technique then draws the representative values as polylines. It also provides a user interface so that users can interactively select interesting representatives, and explore the values which belong to the clusters of the representatives.

It is not always easy to quickly understand musical structure of orchestral scores for classical music works, because these works contain many staves of instruments. This paper presents Color score, a technique for visualization and condensation of musical scores. Color score supports two requirements for composers, arrangers and players: overview and arrangement. Color score divides each track of the score into note-blocks, and determines their roles. Color score then displays all the note-blocks in one display space to provide the overview, so that novice people can quickly understand the musical structures. In addition, Color score supports vertical condensation which reduces the number of displayed tracks, and horizontal condensation which saves the display space. It is especially useful as hints to rearrange music for smaller bands.

This paper presents "Lyricon", a technique that automatically selects multiple icons of tunes block-by-block, and effectively displays the icons. Here, Lyricon selects icons based on not only musical features, but also lyrical keywords. In other words, Lyricon can reflect not only the features of the tunes but also the story of lyrics on its icon selection. Users can understand both impression of the sounds and the content of the lyrics, and they can choose songs which is suitable for their feeling based on the visual impression of the icons. Besides, embedding Lyricon on GUIs of music players is convenient to play specific parts of songs.

We often want to select tunes based on our purposes or situations. For example, we may want background music for particular spaces. We think interactive evolutionary computing is a good solution to adequately recommend tunes based on users' preferences. This paper presents MusiCube, a visual interface for music selection. It applies interactive genetic algorithm in a multi-dimensional musical feature space. MusiCube displays a set of tunes as colored icons in a 2D cubic space, and provides a user interface to intuitively select suggested tunes. This paper presents a user experience that MusiCube adequately represented clouds of icons corresponding to sets of users' preferable tunes in the 2D cubic space.

We present a summarization and visualization technique for large-scale traffic path data. The research aims to visually distinguish the amount of similar traffic, by representing the similar traffic as bundles of lines. Our technique firstly quantizes the collection of paths, then categorizes the segmented paths, and finally renders the bundles of the segments. Our implementation also provides a graphical user interface (GUI) that allows users to interactively explore the various types of data, so that they can adjust the degree of summarization by controlling parameters in the GUI. The technique can visualize various kinds of path data recorded as chronologically ordered positions which form sequential segments, acquired from movies, sensors, and computer simulations. One of the features of the technique is that it can effectively visualize paths in the place where there are not expressly constructed ways. This paper demonstrates the effectiveness of the technique by applying it to two types of path data, where one is acquired by Radio Frequency Identification (RFID) sensors, and the other is extracted from a movie.

Image browsing techniques thus become increasingly important for overview and retrieval of particular images in large-scale collections. At the same time, there are various sets of images which are associated with multi-dimensional or multivariate datasets. We believe that image browsing for such datasets should be inspired from multi-dimensional data visualization techniques. This paper presents Image Cube, a scatter plot-like browser for image collections associated with multi-dimensional datasets. Image Cube locates a set of images into a display space assigning a pair of dimensions to X- and Y-axes. It suggests preferable pairs of dimensions by applying Kendall's rank correlation and Entropy on the display space, so that users can easily obtain interesting visualization results. This paper presents a case scenario that a user finds a preferable car from an image collection by using Image Cube.

We often want to select tunes based on our purposes or situations. For example, we may want background music for particular spaces. We think interactive evolutionary computing is a good solution to adequately recommend tunes based on users' preferences. This paper presents MusiCube, a visual interface for music selection. It applies interactive genetic algorithm in a multidimensional musical feature space. MusiCube displays a set of tunes as colored icons in a 2D cubic space, and provides a user interface to intuitively select suggested tunes. This paper presents a user experience that MusiCube adequately represented clouds of icons corresponding to sets of users' preferable tunes in the 2D cubic space.

Browsing techniques for large image collections become increasingly important for locating and retrieving images. We had a questionnaire result saying that many photograph collectors think that time and location are useful information to organize, browse, and retrieve images. This paper proposes MIAOW (Memorized Image Album Organized by When/Where); a 3D image browser which represents hierarchically categorized photographs based on their shooting locations and times. MIAOW utilizes a 3D space with an orthogonal coordinate system to place a set of photographs; it assigns two axes to the shooting locations of the photographs, and the other axis to their shooting time. Supposing that all images have shooting locations (longitudes and latitudes) and times, MIAOW hierarchically categorizes the set of images in preprocessing step. MIAOW then places all clusters of the images onto the XY-plane of the 3D space as nested rectangular regions, while it attempts to avoid overlapping, minimize the occupied area, and reflect the locations. It also places the same set of clusters onto the XZ- and YZ-planes. avoids any overlaps, minimizes the occupied area, and reflects the time and result of the placement on XY-plane. MIAOW provides an orientation and zooming user interface, so that users can use to easily navigate between location and time spaces and zoom into interested clusters of photographs. This paper demonstrates our user experiments showing that the users required less time to search for specific photographs by using MIAOW rather than using an existing browser.

We represent time-varying data as polyline charts very often. At the same time, we often need to observe hundreds or even thousands of time-varying values in one chart. However, it is often difficult to understand such large-scale time-varying if all the values are drawn in a single polyline chart. This paper presents a polyline-based 3D time-varying data visualization technique. The technique places a set of polylines in the 3D space, where the X-axis denotes time, the Y-axis denotes values, and the polylines are arranged along the Z-axis. It provides two views: the first viewpoint has a view direction along Y-axis, and the second viewpoint has a view direction along Z-axis. The technique displays the overview of the data from the first viewpoint, and the detail of the specific parts of the data from the second viewpoint. It also detects frequent or outlier patterns by applying SAX (Symbolic Aggregate approXimation), and indicates them so that users can discover such characteristic patterns. This paper shows several interesting visualization examples to demonstrate the effectiveness of the presented technique.

There have been two types of Web visualization techniques: visualization of Web sites themselves based on such as link structures or lexical contents, and visualization of browsers' behaviors. We think that integration of such two visualization techniques is very useful for Web site management, and therefore we are currently studying on visualization of access pattern and link structure on a single screen. This paper presents a Web visualization technique using our own multiple-category-embedded graph visualization technique. The presented technique constructs link structures using crawler software, and access patterns from access log files. It then integrates them and visualizes by our graph visualization technique. We expect that users can visually understand the relationship between access patterns and link structures, and utilize the knowledge for design and management of Web sites. This paper shows our case study and discusses typical access patterns we observed by the technique.

Many graphs used in real-world applications consist of nodes belonging to more than one category. We call such graph ldquomultiple-category graphsrdquo. Social networks are typical examples of multiple-category graphs: nodes are persons, links are friendships, and categories are communities that the persons belong to. It is often helpful to visualize both connectivity and categories of the graphs simultaneously. In this paper, we present a new visualization technique for multiple-category graphs. The technique firstly constructs hierarchical clusters of the nodes based on both connectivity and categories. It then places the nodes by a new hybrid space-filling and force-directed layout algorithm to clearly display both connectivity and category information. We show layout results using our hybrid method and compare it with other methods, and present a case study using an active biological network dataset.

Occlusion is an important problem to be solved for readability improvement of 3D visualization techniques. This paper presents an occlusion reduction technique for cityscape-style 3D visualization techniques. The paper first presents an algorithm for occlusion reduction. It generates bounding boxes of 3D objects on the 2D display space, moves them to reduce their overlap, and finally reversely projects their movements onto the 3D space. The paper then presents an application of the algorithm to our own hierarchical data visualization technique, and a music browser based on the technique. The paper also shows several numerical evaluations that denote the effectiveness of the presented technique.

We have various interesting time series data in our daily life, such as weather data (e.g., temperature and air pressure)and stock prices. Polyline chart is one of the most common ways to represent such time series data. We often draw multiple polylines in one space to compare the time variation of multiple values. However, it is often difficult to read the values if the number of polylines gets larger. This paper presents a technique for visualization and level-of-detail control of large number of time series data. The technique generates clusters of time series values,and selects representative values for each cluster, as a preprocessing. The technique then draws the representative values as polylines. It also provides a user interface so that users can interactively select interesting representatives,and explore the time series values which belong to the clusters of the representatives.

The recent revolution of digital camera technology has resulted in much larger collections of images. Image browsing techniques thus become increasingly important for overview and retrieval of images in sizable collections. This paper proposes CAT (Clustered Album Thumbnail), a technique for browsing large image collections, and its interface for controlling the level of details (LOD). As a preprocessing, this new system applies tree-structured clustering to images based on their keywords and pixel values, and selects representative images for each cluster. When a user specifies one or multiple keywords, CAT extracts a branch of the tree structure that contains clusters described by the user-specified keywords. A hierarchical data visualization technique is developed to display the tree structured organization of images using nested rectangular regions. Interlocked to the zooming operation, CAT selectively shows representative images while zooming out, or individual images while zooming in.

We previously applied our own hierarchical data visualizationtechnique for structure-activity relationship (SAR)analyses of biochemical data. The study applied a recursivepartitioning to store the drugs as hierarchical data,based on their chemical structures, and visualized the hierarchyof drugs. Though the activity data of drugs is usuallymulti-dimensional, our previous work did not represent themulti-dimensional values onto one display space. This paperpresents a technique for visualizing hierarchical multidimensionaldata, and its level-of-detail (LOD) control,for visualization of multi-dimensional bioactive chemicaldata. The technique is an extension of our hierarchicaldata visualization technique, where the extended techniqueplaces leaf-nodes as well as the original hierarchical datavisualization technique, and represents multi-dimensionalvalues by dividing the icons of the leaf nodes.

This paper presents a technique for the representation of large-scale hierarchical data which aims to provide good overviews of complete structures and the content of the data in one display space. The technique represents the data by using nested rectangles. It first packs icons or thumbnails of the lowest-level data and then generates rectangular borders that enclose the packed data. It repeats the process of generating rectangles that enclose the lower-level rectangles until the highest-level rectangles are packed. This paper presents two rectangle-packing algorithms for placing items of hierarchical data onto display spaces. The algorithms refer to Delaunay triangular meshes connecting the centers of rectangles to find gaps where rectangles can be placed. The first algorithm places rectangles where they do not overlap each other and where the extension of the layout area is minimal. The second algorithm places rectangles by referring to templates describing the ideal positions for nodes of input data. It places rectangles where they do not overlap each other and where the combination of the layout area and the distances between the positions described in the template and the actual positions is minimal. It can smoothly represent time-varying data by referring to templates that describe previous layout results. It is also suitable for semantics-based or design-based data layout by generating templates according to the semantics or design.

Visualization of distributed processes is useful for the management of large-scale distributed computing systems. Reactivity and scalability are especially important requirements for such visualization of distributed processes. We have proposed the visualization technique "Data Jewelry Box" algorithm, which satisfies both of the above requirements. The technique can be applied for the visualization of distributed processes, however, the algorithm has a problem that may yield much different data layouts even among very similar datasets. This is a serious issue for the seamless visualization of time-varying data. To solve the problem, we propose the extension of "Data Jewelry Box" algorithm. The extension places data elements referring positions of the previous data layout, so that the extension can yield similar layouts among similar datasets. We introduce the extended algorithm, and propose the visualization system for distributed computing systems using the extended algorithm.

One of the most effective techniques for developing efficient isosurfacing algorithms is the reduction of visits to nonisosurface cells. Recent algorithms have drastically reduced the unnecessary cost of visiting nonisosurface cells. The experimental results show almost optimal performance in their isosurfacing processes. However, most of them have a bottleneck in that they require more than O(n) computation time for their preprocessing, where n denotes the total number of cells. We propose an efficient isosurfacing technique, which can be applied to unstructured as well as structured volumes and which does not require more than O(n) computation time for its preprocessing. A preprocessing step generates an extrema skeleton, which consists of cells and connects all extremum points, by the volume thinning algorithm. All disjoint parts of every isosurface intersect at least one cell in the extrema skeleton. Our implementation generates isosurfaces by searching for isosurface cells in the extrema skeleton and then recursively visiting their adjacent isosurface cells, while it skips most of the nonisosurface cells. The computation time of the preprocessing is estimated as O(n). The computation time of the isosurfacing process is estimated as O(n/sup 1/3/m+k), where k denotes the number of isosurface cells and m denotes the number of extremum points since the number of cells in an extrema skeleton is estimated as O(n/sup 1/3/m).

In this paper, we propose a novel formfactor calculation algorithm for acceleration radiosity solutions in complex environments. Our basic algorithm is an improved version of Spencer's (S.N. Spencer, ‘The hemisphere radiosity method: a tale of two algorithms’, in Photorealism in Computer Graphics, Spencer, 1992, pp. 127–135) and Van Wyk's (G.C. Van Wyk Jr., ‘A geometry-based insolution model for computer-aided design,’ Ph.D. Thesis, The University of Michigan, 1998.) methods, which fail to remove hidden surfaces for relatively large patches and cause large discretization errors in formfactors. We also demonstrate that our technique is superior to the hemi-cube method in terms of the computation time. Moreover, we parallelize our approach on a parallel computer with shared memory, and obtain a high performance with our radiosity rendering system. Our method divides a hemisphere-base into regions, and assigns a region to each processor. The approach can be applied to geometrical data generated by CAD systems, and is evaluated in terms of the computation time, the visual effects, and the parallelization performance. © 1998 John Wiley & Sons, Ltd

An isosurface can be efficiently generated by visiting adjacent intersected cells in order, as if the isosurface were propagating itself. We previously proposed an extrema graph method (T. Itoh and K. Koyamada, 1995), which generates a graph connecting extremum points. The isosurface propagation starts from some of the intersected cells that are found both by visiting the cells through which arcs of the graph pass and by visiting the cells on the boundary of a volume. We propose an efficient method of searching for cells intersected by an isosurface. This method generates a volumetric skeleton. consisting of cells, like an extrema graph, by applying a thinning algorithm used in the image recognition area. Since it preserves the topological features of the volume and the connectivity of the extremum points, it necessarily intersects every isosurface. The method is more efficient than the extrema graph method, since it does not require that cells on the boundary be visited.

A high-performance algorithm for generating isosurfaces is presented. In our method, guides to searching for cells intersected by an isosurface are generated as a pre-process. These guides are two kinds of cell lists: an extrema graph, and sorted lists of boundary cells. In an extrema graph, extremum points are connected by arcs, and each arc has a list of cells through which it passes. At the same time, all boundary cells are sorted according to their minimum and maximum values, and two sorted lists are then generated. Isosurfaces are generated by visiting adjacent intersected cells in order. Here, the starting cells for this process are found by searching in an extrema graph and in sorted boundary cell lists. In this process, isosurfaces appear to propagate themselves. Our algorithm is efficient, since it visits only cells that are intersected by an isosurface and cells whose IDs are included in the guides. It is especially efficient when many isosurfaces are interactively generated in a huge volume. Some benchmark tests described in this paper show the efficiency of the algorithm.

A high-performance algorithm for generating isosurfaces is presented. In this algorithm, extrema points in a scalar field are first extracted. A graph is then generated in which the extrema points are taken as nodes. Each arc of the graph has a list of IDs of the cells that are intersected by the arc. A boundary cell list ordered according to cells' values is also generated. The graph and the list generated in this pre-process are used as a guide in searching for seed cells. Isosurfaces are generated from seed cells that are found in arcs of the graph. In this process, isosurfaces appear to propagate themselves. The algorithm visits only cells that are intersected by an isosurface and cells whose IDs an included in cell lists. It is especially efficient when many isosurfaces are interactively generated in a huge volume. Some benchmark tests described show the efficiency of the algorithm.>