Search

The explosive growth in supercomputers capacity has changed simulation paradigms. Simulations have shifted from a few lengthy ones to an ensemble of multiple simulations with varying initial conditions or input parameters. Thus, an ensemble consists of large volumes of multi-dimensional data that could go beyond the exascale boundaries. However, the disparity in growth rates between storage capabilities and computing resources results in I/O bottlenecks. This makes it impractical to utilize conventional post-processing and visualization tools for analyzing such massive simulation ensembles. In situ visualization approaches alleviate I/O constraints by saving predetermined visualizations in image databases during simulation. Nevertheless, the unavailability of output raw data restricts the flexibility of post hoc exploration of in situ approaches. Much research has been conducted to mitigate this limitation, but it falls short when it comes to simultaneously exploring and analyzing parameter and ensemble spaces. In this paper, we propose an expert-in-the-loop visual exploration analytic approach. The proposed approach leverages: feature extraction, deep learning, and human expert–AI collaboration techniques to explore and analyze image-based ensembles. Our approach utilizes local features and deep learning techniques to learn the image features of ensemble members. The extracted features are then combined with simulation input parameters and fed to the visualization pipeline for in-depth exploration and analysis using human expert + AI interaction techniques. We show the effectiveness of our approach using several scientific simulation ensembles. [ABSTRACT FROM AUTHOR]

We propose a novel visual analytics approach based on the Latent Dirichlet Allocation (LDA) model for exploring and interpreting the dynamic evolution of networks. In this approach, we define networks as documents and relationships within networks as words. Using this definition, the LDA model is able to extract a list of structures that fuse relationships and connect the network features. We project networks described by the extracted structures with probabilistic assignments as points into a two-dimensional space via dimensionality reduction techniques. Users can identify evolution states in dynamic networks, including stable states, recurrent states, outlier states, and state transitions. To facilitate the interpretation of evolution states, we provide a novel small multiples view that shows how the extracted structures behave over time. We demonstrate the effectiveness of our work through case studies conducted on two real-world dynamic networks. [ABSTRACT FROM AUTHOR]

For a large complex industrial equipment with high-density sensors, exploring the potential influence of generated multiregion monitoring parameters on subsequent control links can be very meaningful to optimize the control process. However, the influencing mechanism and randomness between such numerous monitoring parameters and subsequently influenced parameters are intertwined, and each working condition of the control system has its unique running characteristics and control rules, which makes it challenging to analyze the correlations between these different categories of parameter sets effectively. In this paper, we propose a comprehensive approach that combines parameter fusion and canonical correlation analysis for this kind of high-dimensional industrial control data and constructs a visual analysis framework CAPVis that supports multi-perspective and multi-level exploration of canonical correlation patterns. For a single working condition, we visualize the intricate structure inside of the canonical correlation relationships with a particular tripartite graph and evaluate the redundancy and stability of these relationships with multiple auxiliary views. For multiple working conditions, we design different visual comparison strategies to comprehensively compare the many-to-many canonical correlation patterns from local to global. Experiments on real industrial control datasets and feedback from industry experts demonstrate the effectiveness of CAPVis. [ABSTRACT FROM AUTHOR]

In the field of deep learning, pre-trained BERT models have achieved remarkable success. However, the accompanying problem is that models with more complex structures and more network parameters. The huge parameter size makes the computational cost in terms of time and memory become extremely expensive. Recent work has indicated that BERT models own a significant amount of redundant attention heads. Meanwhile considerable BERT models compression algorithms have been proposed, which can effectively reduce model complexity and redundancy with pruning some attention heads. Nevertheless, existing automated model compression solutions are mainly based on predetermined pruning program, which requires multiple expensive pruning-retraining cycles or heuristic designs to select additional hyperparameters. Furthermore, the training process of BERT models is a black box, and lacks interpretability, which makes researchers cannot intuitively understand the optimization process of the model. In this paper, we propose a visual analysis system, BHPVAS, for pruning BERT models, which helps researchers to incorporate their understanding of model structure and operating mechanism into the model pruning process and generate pruning schemes. We propose three pruning criteria based on the attention data, namely, importance score, stability score, and similarity score, for evaluating the importance of self-attention heads. Additionally, we design multiple collaborative views to display the entire pruning process, guiding users to carry out pruning. Our system supports exploring the role of self-attention heads in the model inference process using text dependency relations and attention weight distribution. Finally, we conduct two case studies to demonstrate how to use the system for Sentiment Classification Sample Analysis and Pruning Scheme Exploration, verifying the effectiveness of the visual analysis system. [ABSTRACT FROM AUTHOR]

Gradient boosting decision tree (GBDT) is a mainstream model for advertisement click-through rate (CTR) prediction. Since the complex working mechanism of GBDT, advertising analysts often fail to analyze the decision-making and the iterative evolution process of a large number of decision trees, as well as to understand the impact of different features on the prediction results, which makes the model tuning quite challenging. To address these challenges, we propose a visual analytics system, GBDT4CTRVis, which helps advertising analysts understand the working mechanism of GBDT and facilitate model tuning through intuitive and interactive views. Specifically, we propose instance-level views to hierarchically explore the prediction results of advertising data, feature-level views to analyze the importance of features and their correlations from various perspectives, and model-level views to investigate the structure of representative decision trees and the temporal evolution of information gain during model prediction. We also provide multi-view interactions and panel control for flexible exploration. Finally, we evaluate GBDT4CTRVis through three case studies and expert evaluations. Feedback from experts indicated the usefulness and effectiveness of GBDT4CTRVis in helping to understand the model mechanism and tune the model. [ABSTRACT FROM AUTHOR]

Spatio-temporal evolution analysis has been a critical topic of air pollution research. However, there are still several difficulties caused by the large scale and dimensionality of the data. Specifically, First, traditional methods deal with such data by simplifying and abstracting, resulting in information loss. Second, most existing visualizations, generally focusing on overall evolution, ignore the exploration of multiple time scales and pattern transitions between subsequences. This paper presents AirPollutionViz, a visual analytics system that enables to analyze the spatio-temporal evolution in two manners: sequence mining and clustering analysis. Concretely, we propose sequence merging to shorten the sequence length and construct a weighted directed graph structure, which promotes efficient querying of sequence patterns by combination with dynamic time warping. We design a novel summary view to display the overview of pollution level changes, together with the improved node-link chart, to support the analysis of air pollution spatio-temporal evolution patterns. We also apply K-means clustering to pollutants, and a scatter plot and map reflect the spatial distribution aggregation. The system supports users' free exploration across multiple time scales with rich interactions. Case studies with three domain experts and a user study with ten users demonstrate the usefulness and effectiveness of AirPollutionViz. [ABSTRACT FROM AUTHOR]

Social networking services (SNSs) have become the primary means by which individuals express themselves. Consequently, the thoughts of individuals could be explored by analyzing primary topics on SNSs. In this study, we proposed and developed a novel system for visual analytics to address the following intriguing questions. When do topics change? Do they ever resurface? What do people typically discuss? Using document embedding and dimensionality reduction approaches, we abstracted dynamic topics as several points in a two-dimensional space. In addition, we provided other charts depicting words that appeared at certain moments and their time series dynamics over entire periods. In addition, we created a novel text visualization technique called semantic-preserving word bubbles to visualize words at a specific time. In addition, we demonstrated the efficacy of the proposed system utilizing Twitter data regarding early COVID-19 trends, Fukushima nuclear disaster trends, and user ratings of system usability. In general, we have presented this novel system to aid users in exploring and comprehending the transitions of contents uploaded on SNSs. [ABSTRACT FROM AUTHOR]

Deep reinforcement learning (DRL) has received widespread attention recently, where the control policies are trained through deep neural networks. Several visual analytics methods were proposed to reveal the internal mechanism of DRL. However, most of the current methods focused on analyzing DRL algorithms with a single agent. Due to the inherent non-stationarity of the environment and the complex interactions among multiple agents, understanding the learning process of multi-agent deep reinforcement learning is more challenging. This paper presents MADDPGViz, a visual analytics system to expose details of the training process from different aspects for a multi-agent deep deterministic policy gradient (MADDPG) model. MADDPGViz allows users to overview the model's training statistics and compare the large experience space of multiple agents during different training episodes to learn the functionality of agents' centralized critics. Users can further examine the dynamic interaction among multiple agents and environmental objects for a selected episode, to discover the learned cooperative strategies of agents from the microscopic aspect. Case studies are conducted in two different cooperative environments, to demonstrate the usability of the system. [ABSTRACT FROM AUTHOR]

Goods turnover is the core of digital warehouse operation, including many processes, such as receiving, picking, and packing of goods. Analyzing goods turnover data can generate valuable insights for optimizing warehouse management, thereby improving operation efficiency. However, most existing methods focus on partial processes, making it hard for warehouse managers to understand the operation state and the goods turnover patterns, which often require the analysis of the interrelated processes of goods turnover. In this paper, we abstract six types of goods turnover events to describe the warehouse operation workflow and present WarehouseLens, a visual analytics system to analyze goods turnover from an overall perspective. To understand the warehouse operation state, we propose a temporal visualization method consisting of a novel state calendar view and an improved circular heat map to reflect the trend and periodicity pattern of the operation state. To explore the goods turnover patterns, we provide an improved parallel coordinate plot for users to view the attribute distribution of goods to filter key goods and a tailored mode circle view to discover the frequent outbound mode of goods. Three case studies and expert interviews on a real-world warehouse dataset demonstrate the usefulness and effectiveness of WarehouseLens in revealing the warehouse operation state and goods turnover patterns. [ABSTRACT FROM AUTHOR]

Online learning is gradually becoming a popular way of learning due to the high flexibility in time and space. Reducing the high dropout rate is important to promote the further development of smart education. However, learners' learning is a dynamic temporal process, which is influenced by multiple factors synergistically. How to identify the key influencing factors of dropout in an interpretable way is still a challenging problem. In this paper, we propose a pattern identification method of dropout behavior, including the prediction of the dropout probability and the mining of potential impact factors, to gain a comprehensive insight into the dropout behavior hidden in the data. A CNN-LSTM model for dropout prediction is constructed, which can automatically extract features and learn the temporal dependence of dropout behavior. By introducing the counterfactual explanation, the dropout impacts of different learning behavior can be revealed quantitatively. Moreover, we design and develop an interactive visual analytics system, DropoutVis, for exploring learning behavior, extracting the various dropout patterns and providing a basis for formulating strategies. The effectiveness and usefulness of DropoutVis have been demonstrated through case studies with a real dataset. [ABSTRACT FROM AUTHOR]

Time-series representation is fundamental for various analysis tasks on time-series data, such as similarity analysis, clustering, and classification. Traditional representation methods produce indistinguishable representations based on statistical information. Given the good performance of deep learning techniques in feature extraction, we apply it to the task of time-series data representation. However, such methods are usually evaluated only through a few quantitative metrics that reflect the limited model performance and lack further explanation. In this paper, we design TSRNet, a deep neural network containing two weight-sharing auto-encoders to learn representations effectively. To better analyze the representation, we design an interactive visual analytics system TSRVis in collaboration with seven experts from the fields of visual analytics and machine learning. Experimental evaluations on 18 datasets verify the effectiveness of TSRNet in terms of clustering accuracy. Three case studies demonstrate that TSRVis is practicable for assisting analysts to perform overview, comparison, and point-of-interest analyses and assess the validity of representations from multiple perspectives. [ABSTRACT FROM AUTHOR]

This paper presents a hybrid prediction and search approach (HPS) for building visualization systems of big data. The basic idea is training a regression model to predict a coarse range on the dataset and then searching target records that satisfy the query conditions within the range. The prediction reduces the storage cost without preprocessing a data structure storing aggregate values of queriable attribute range combinations. Meanwhile, the search eliminates the prediction bias inevitable for machine learning models. Experiments on multiple open datasets demonstrate HPS's comparable query speed to existing techniques and the potential of continuous performance improvement by investing more hardware resources. In addition, the feature of returning original records instead of aggregate values brings better use flexibility, enabling to construct visualization systems with display/query functions that are unavailable for existing techniques. [ABSTRACT FROM AUTHOR]

Continuous research and development of novel tourism routes is necessary for tourism service providers to improve the tourist experience and industrial competitiveness. However, the route planning is cumbersome due to the time-consuming, extensive, and costly field study. Most of the existing route planning studies focus on recommending tourism routes for users based on attraction characteristics or tourist behavior features, which are generally unexplainable due to the black-box approaches they use. Other solutions allow users to customize itineraries through an interactive interface but often lack guidance from the aspect of route evaluation or destination image perception. In this paper, we thoroughly discuss the requirements and design tasks with domain experts and propose TriPlan, an interactive visual analytics system that provides intuitive planning guidance for tourism product developers. We design and improve multiple coordinated visualizations to facilitate analysis from the perspectives of overall route pattern and individual destination image. We also develop a hierarchical planning view to display the structural information of a plan. In addition, we introduce an automatic route optimization algorithm and multiple interactions to assist users in optimizing and adjusting the itineraries. Finally, we evaluate the usability and effectiveness of our system through three case studies and quantitative and qualitative interviews with the domain experts on real-world datasets. [ABSTRACT FROM AUTHOR]

Reading is a complex task that can provide valuable information about our perceptual and cognitive processes. To understand how people read, researchers have embraced the use of eye-tracking techniques. Recent research work studies the eye movements during reading of short sentences, and however, the extension of these findings to natural reading has not been yet studied in depth. The visual analysis of eye movement data has become an emerging field providing important means to support statistical analysis and hypothesis building. In this work, we focus on the visual analysis of the natural reading of a particular type of text, the micro-stories, which are short-length texts that condense a large amount of information. We present a novel visualization technique for analyzing eye movement data during the reading of micro-stories. In the design of the proposed technique, we consider all the characteristics defined for a typical reading experiment, integrating all of them into a single view. We also provide associated interactions to facilitate exploration. Our novel technique allows the analysis of eye movements during micro-story reading helping the experts to explore relationships among characteristics and to discover hidden relations that help to understand the cognitive process involved. [ABSTRACT FROM AUTHOR]

The striking results of deep neural networks (DNN) have motivated its wide acceptance to tackle large datasets and complex tasks such as natural language processing, facial recognition, and artificial image generation. However, DNN parameters are often empirically selected on a trial-and-error approach without detailed information on convergence behavior. While some visualization techniques have been proposed to aid the comprehension of general-purpose neural networks, only a few explore the training process, lacking the ability to adequately display how abstract representations are formed and represent the influence of training parameters during this process. This paper describes neural network training fingerprint (NNTF), a visual analytics approach to investigate the training process of any neural network performing classification. NNTF allows understanding how classification decisions change along the training process, displaying information about convergence, oscillations, and training rates. We show its usefulness through case studies and demonstrate how it can support the analysis of training parameters. [ABSTRACT FROM AUTHOR]

The availability of massive temporal multivariate data provides unprecedented opportunities to analyze anomaly pattern evolution. However, due to the complicated relationships and dynamic temporality, it is a challenging problem to efficiently reveal and perceive the evolution of abnormal patterns. Existing methods mostly focus on the use of automated algorithms to capture anomalies, which limits the understanding of the evolutionary law. In this paper, we propose a novel visual analytics system, PEVis, to help domain experts extract four types of abnormal evolution patterns and quantitatively analyze the contribution of different features to its occurrence. By aggregating different anomaly detection algorithms, the anomalous data that deviate seriously from other densely distributed points or with small isolated clusters can be identified. A strategy for perceiving temporal multivariate data is proposed to solve the conflict between exploring the global time-varying law and local data distribution in visualization, through aligning dimension reduction space of adjacent timestamp and designing a doughnut glyph to visually display data distribution and feature contributions. Experiments on two real-world datasets, namely air quality index data and consumer price index data, demonstrate that the proposed methods can effectively reveal the seasonal variation of abnormal phenomena. [ABSTRACT FROM AUTHOR]

Named entity recognition (NER) is a crucial initial task that identifies both spans and types of named entities to extract the specific information, such as organization, person, location, and time. Nowadays, the NER task achieves state-of-the-art performance by deep learning approaches for capturing contextual features. However, the complex structures of deep learning make a black-box problem and limit researchers' ability to improve it. Unlike the Latin alphabet, Chinese (or other languages such as Korean and Japanese) do not have an explicit word boundary. Therefore, some preliminary works, such as word segmentation (WS) and part-of-speech tagging (POS), are needed before the Chinese NER task. The correctness of preliminary works importantly influences the final NER prediction. Thus, investigating the model behavior of the Chinese NER task becomes more complicated and challenging. In this paper, we present CNERVis, a visual analysis tool that allows users to interactively inspect the WS-POS-NER pipeline and understand how and why a NER prediction is made. Also, CNERVis allows users to load the numerous testing data and explores the critical instances to facilitate the analysis from large datasets. Our tool's usability and effectiveness are demonstrated through case studies. [ABSTRACT FROM AUTHOR]

Deep reinforcement learning (DRL) has been widely used in autonomous control due to its superior performance. DRL-based autonomous control model (ACM) aims to train an agent to achieve self-control and learn optimal policy through pre-defined rewards. Despite the super-human performance, ACM is regarded as a black box, and the interpretation of its internal working mechanism remains a challenge to domain experts. In addition, adjusting the reward settings of ACM is also challenging due to the uncertain relationship between rewards setting and strategies. In this paper, we propose ACMViz, a visual analytics system to explore control strategies at different stages and reveal the relationship between rewards and action patterns. Focusing on controlling a lunar lander, ACMViz investigates different landing trajectories and action sequences to interpret the model and control the training. From our visual analytics of the action patterns, we diagnose and improve reward settings for different control targets. Through our case studies with deep learning experts, we validate the effectiveness of ACMViz. [ABSTRACT FROM AUTHOR]

Rare category detection is an important topic in data mining, which focuses on identifying the very first example from rare categories by requesting only a small number of labels. Most existing techniques for rare category detection are designed for tabular data sets, thus not suitable for graph data. Moreover, the few applicable techniques to graph data require prior information about data sets and will become useless without such information. In this paper, we introduce RCDVis, a visual analysis system designed to support prior-free rare category detection on graph data. RCDVis adopts a community detection-based algorithm RCDGD to identify candidate vertices of rare categories. Furthermore, the algorithm is tightly integrated with an interactive interface which supports the effective exploration of graph data and the labeling of rare categories. This paper (1) introduces the prior-free rare category detection algorithm; (2) describes the visualization and interaction designs of user interface; and (3) presents results from quantitative experiments and example usage scenarios to demonstrate the effectiveness of RCDVis. [ABSTRACT FROM AUTHOR]

Nonlinear programming is a complex methodology where a problem is mathematically expressed in terms of optimality while imposing constraints on feasibility. Such problems are formulated by humans and solved by optimization algorithms. We support domain experts in their challenging tasks of understanding and troubleshooting optimization runs of intricate and high-dimensional nonlinear programs through a visual analytics system. The system was designed for our collaborators' robot motion planning problems, but is domain agnostic in most parts of the visualizations. It allows for an exploration of the iterative solving process of a nonlinear program through several linked views of the computational process. We give insights into this design study, demonstrate our system for selected real-world cases, and discuss the extension of visualization and visual analytics methods for nonlinear programming. [ABSTRACT FROM AUTHOR]

Table tennis matches consist of many stroke sequences where two players hit the ball interactively and consecutively until one fails to hit the ball. Players usually employ many complicated playing techniques at each stroke in highly antagonistic, variable, and flexible matches. In-depth comparative analyses of players' stroke sequences are necessary to obtain insights into the technical playing patterns of players. Experts commonly use spreadsheets to browse and compare strokes one by one, and this process is tedious and prone to errors. Statistical analyses are limited to well-defined patterns (e.g., value distribution and relation significance) and fail to present complex and peculiar patterns. We collaborated with experts to dig out soft patterns of stroke sequences and proposed a novel interactive visualization system to present and compare the patterns. The main visualization challenge is to display the multivariate stroke sequence and the spatial variation patterns. We designed a glyph-based pattern view to solve the challenge. These comprehensible visualizations and coordinated views in the system allow efficient comparative analysis of stroke sequence patterns and are highly commended by domain experts, who have identified several new and interesting patterns using the system. We demonstrated the effectiveness and usability of the visualization system through case studies with table tennis experts.. [ABSTRACT FROM AUTHOR]

Understanding the relationship between urban areas is essential to the extraction of human mobility patterns. Most popular algorithms use word embedding technology to vectorize urban areas to calculate the similarity between them. However, noise generated by the traffic hub area is often ignored, such as crossroads and road toll stations, which affect the training time of the model and the accuracy of vectorization. In this paper, we propose a vectorized representation method of urban geographic areas based on the distribution of trajectories, combining with the filtering of traffic hub areas based on the law of similarity to eliminate their noise. A novel visual analytics system is designed, which enables experts to analyse the relationship between urban areas and explore human mobility patterns based on this vectorized representation method. Finally, we analyse the relevance to long-distance areas and campus crowd mobility patterns on real datasets, which proves the effectiveness of our method. [ABSTRACT FROM AUTHOR]

The advertising revenue of online shopping platforms comes from the users who click on the advertisements and purchases the advertised goods. Therefore, to accurately advertise and increase revenue, advertising analysts engage in discovering representative groups and their behavior patterns from the data of demographic attributes, shopping behaviors, and advertising click behaviors of a large number of users. Existing methods often represent user behaviors based on single-level user profiles. However, under different community granularity and time scales, user behaviors have different characteristics. In addition, the sequential relationship between advertisement clicks and other shopping behaviors is difficult to be accurately identified by the single-level analysis methods. Therefore, we cooperate with advertising experts and propose a multi-level visual analysis method based on the K-Means algorithm, which can better understand user behaviors from multiple community granularity and multiple time scales. We design two novel visualization diagrams and improve three traditional charts that can help analysts observe user characteristics at the three levels: user groups, user subgroups, and user individuals, as well as can analyze the time-series events such as advertising clicks and product purchases of representative users from multiple time scales. Furthermore, we implement a multi-view interactive prototype system MulUBA to help analysts put targeted advertisements and increase advertising revenue. Finally, we verify the effectiveness and usability of our approach by conducting three case studies and an expert evaluation on a real-world online shopping advertising dataset. [ABSTRACT FROM AUTHOR]

Decision makers often find themselves in situations where they need to consider time-varying values for multi-criteria decision-making. Skyline queries are one of the most widely used methods of approaching multi-criteria decision-making problems because they reduce the size of search space by excluding inferior data. However, skylines in time-series data fluctuate with changes in attributes. Moreover, the number of skyline points increases as the number of dimensions increases, and the skyline query itself does not provide any ranking method. Thus, users are required to direct a considerable amount of effort into analyzing and finding the best selection. To address these issues, we propose SkyFlow, a visual analytical system for comparing time-varying data to facilitate the decision-making process. We apply two datasets in our system and describe scenarios to demonstrate the effectiveness of SkyFlow. In addition, we conduct a qualitative study to highlight the efficiency of our system in assisting users to compare candidates and make decisions involving time-series data. [ABSTRACT FROM AUTHOR]

Current tracking technology such as GPS data loggers allows biologists to remotely collect large amounts of movement data for a large variety of species. Extending, and often replacing interpretation based on observation, the analysis of the collected data supports research on animal behaviour, on impact factors such as climate change and human intervention on the globe, as well as on conservation programs. However, this analysis is difficult, due to the nature of the research questions and the complexity of the data sets. It requires both automated analysis, for example, for the detection of behavioural patterns, and human inspection, for example, for interpretation, inclusion of previous knowledge, and for conclusions on future actions and decision making. For this analysis and inspection, the movement data needs to be put into the context of environmental data, which helps to interpret the behaviour. Thus, a major challenge is to design and develop methods and intuitive interfaces that integrate the data for analysis by biologists. We present a concept and implementation for the visual analysis of cheetah movement data in a web-based fashion that allows usage both in the field and in office environments. [ABSTRACT FROM AUTHOR]

Sensor networks composed of static and mobile sensors are applicable for situation monitoring. In this paper, we propose SensorAware, an interactive system for visualizing and exploring spatial–temporal data from static and mobile sensors. Our system follows the procedure starting with an overview, then zoom and filter, and finally details-on-demand. We use pixel-based time-series visualization to show overall readings from individual sensors, and the readings of mobile sensors are aggregated spatially to display the distribution of sensor readings. SensorAware provides cross-filtering and details-on-demand interactions, which allow users to investigate data at different levels of details from both spatial and temporal aspects. [ABSTRACT FROM AUTHOR]

An unbalanced panel is a dataset in which at least one subject is not observed some times. Moreover, each subject is recorded with irregular periods and intervals. Therefore, only short trend pattern pieces exist in the data. When applying existing prediction techniques, it is challenging to create a prediction model that reflects individual subject patterns. Also, uncertainties in the predicted results emerge since the overall trend of the data is unknown. In this paper, we present a Bayesian network to predict the future trends of subjects from the unbalanced panel data. We also present a new approach to estimate the predicted intervals of the predicted results. Moreover, we propose a visual analytics system that enables us to build a prediction model from unbalanced panel data. The visual analytics system also supports performance improvement in the already designed prediction model. We evaluate the effectiveness of our system while building a predictive model according to various data patterns. [ABSTRACT FROM AUTHOR]

A city contains a variety of different urban functions with referring to the purpose of land use to support the diverse needs of urban residents, such as residence, working and recreation. Exploring urban functional zones is a critical task, which provides valuable applications for business site selection, transportation management and urban planning. It is well known that location information and human mobility semantics both are significant factors for identifying urban functional zones, just considering one factor is not effective. However, most of existing techniques capture the transformation of functional zones and interpret the results based solely on location or mobility semantics, and lack the capacity to deal with the multifaceted features of urban data. To tackle these problems, in this paper, we propose an interactive visual analytics system for effectively exploring urban functional zones based on spatio-temporal OD data and Points of Interest data. We first adaptively partition the territory into region units based on adaptive blue noise sampling method, extract POI feature matrix (location information) as a prior knowledge, model multidimensional spatio-temporal OD data as a tensor for addressing multifaceted features and propose an augment tensor-based algorithm that enables users to simultaneously combine mobility semantics and inherent location information for identifying functional zones. In addition, we design a set of visual encodings to better understand and interpret the results in a visual and intuitive manner. This system has been demonstrated using two case studies with a real-world dataset of HangZhou city and domain-expert interviews. [ABSTRACT FROM AUTHOR]

Satellites detect the distribution of meteorological data worldwide. However, due to the orbital constraints, the satellite can only reach the same area again after one orbiting cycle. The interval between two detections in the same area is long, and the variation of meteorological data between the two detections is unknown. Moreover, meteorological satellite data are only located near the orbit in one cycle, while the global distribution of meteorological data is unknown. Our method allows to train a regression model with only few meteorological satellite data by taking advantage of the recent advances in deep learning. In detail, we train a model-agnostic meta-learning (MAML) model with data from ground stations instead of meteorological satellites and get the initial network parameters. Based on the initial network parameters trained by MAML, we train the regression models again for different areas. We sample the regression curves of all areas by time and get a time series of global meteorological data distribution. Through case studies conducted together with domain experts, we validate the effectiveness of our method. [ABSTRACT FROM AUTHOR]

Air pollution has become an important environmental issue, attracting more and more attention from many scholars and experts recently. Understanding air quality patterns in urban areas is essential for air pollution prevention and treatment. However, most existing studies usually cannot effectively capture air quality patterns from large-scale air quality data, due to lacking effective interaction approaches and intuitive methods that reveal sequential and multivariable information. In this paper, we present AirExplorer, a novel visual analysis system providing abundant interactive ways and intuitive views to help users explore the time-varying and multivariable patterns of air quality data. We design a time-embedded RadViz view that not only shows the relationship between data and multivariable attributes, but also puts the air quality temporal variations among the observation stations into perspective. Furthermore, we suggest a time-series querying algorithm, which combines hierarchical Piecewise Linear Representation and Dynamic Time Warping, to help users query time-series patterns of interest accurately by a sketch-based interaction. The experiment results based on the real dataset demonstrate that our method can help users understand the spatial-temporal multi-dimensional characteristics effectively and discover some potential laws of air quality patterns. AirExplorer with easy-to-use interactions can improve the efficiency of analyzing air quality data. [ABSTRACT FROM AUTHOR]

The urban transportation system is the footstone of a city's infrastructure, and the booming bike-sharing system has become a vital part of urban transportation. Understanding the bike-sharing system and traditional taxi system as well as their similarities and differences are essential for bike-sharing rebalancing, taxi dispatching, and urban planning. However, due to the sparseness of record data and the difference in service regions, the relationship between them is indeed obscure, and previous solutions mostly focus only on a single system. In this paper, we propose a visual analytics system to investigate the similarities and differences between bike-sharing and taxi systems. The service region for each bike station is created to fuse bike-sharing data and taxi data. We harness two three-order tensors to represent them in a unified framework to generate potential patterns by tensor decomposition. The visual analytics system integrates two spatiotemporal data sources by analyzing the patterns that are typical of each data source and the patterns that are common to both data sources to assist users in better discovering the relationships between the taxi system and the bike-sharing system. We demonstrate the effectiveness of our system through real-world case studies. The urban transportation system is the footstone of a city's infrastructure, and the booming bike-sharing system has become a vital part of urban transportation. Understanding the bike-sharing system and traditional taxi system as well as their similarities and differences are essential for bike-sharing rebalancing, taxi dispatching, and urban planning. However, due to the sparseness of record data and the difference in service regions, the relationship between them is indeed obscure, and previous solutions mostly focus only on a single system. In this paper, we propose a visual analytics system to investigate the similarities and differences between bike-sharing and taxi systems. The service region for each bike station is created to fuse bike-sharing data and taxi data. We harness two three-order tensors to represent them in a unified framework to generate potential patterns by tensor decomposition. The visual analytics system integrates two spatiotemporal data sources by analyzing the patterns that are typical of each data source and the patterns that are common to both data sources to assist users in better discovering the relationships between the taxi system and the bike-sharing system. We demonstrate the effectiveness of our system through real-world case studies. [ABSTRACT FROM AUTHOR]

Targeted advertising is a dominant form of online advertising. It considers advertisers' major concern of their customers, including the consumers' certain traits, interests and individual preferences. To promote the effectiveness of advertisement delivery, advertising analysts need to understand advertiser delivery behavior and problems in targeting structure. However, statistical methods cannot meet analytical requirements completely, and analysts have to spend a lot of time reading countless data reports. Concretely, there is no efficient tool accomplishing analysis tasks such as exploring targeting usage at different levels, discovering useful or abnormal targeting combination patterns, finding competition from user behavior. In this paper, we design and implement an interactive visual analytics system named TargetingVis to visualize targeted advertising delivery data to face the challenges. After conducting a detailed requirements analysis with the domain experts from Tencent Inc., we design TargetingVis with four linked views: a novel chord diagram for cross-level exploration of targeting relations, a view for delving into the analysis of targeting combination patterns, an auxiliary view for displaying data indicators and a view to help gain insights into the behavior of advertisers. Finally, we evaluate the usability and efficiency through experiments based on real-world datasets. [ABSTRACT FROM AUTHOR]

Higher-order information of moving objects is of great importance for processing in-case scenarios in emergency management. Multivariate higher-order information management is a crucial key to the success of emergency management since emergency management involves developing plans with a given set of multiple resources. Past studies focus on univariate higher-order information limiting the scope of applicability and usability. This paper proposes a set of visual analytical approaches supporting multivariate higher-order information for dynamically moving disasters. We introduce a robust Voronoi-based data structure supporting multivariate datasets and dynamic disasters and propose visual analytical approaches for effective emergency management. The proposed visual analytical suite facilitates interactivity and enables users to explore in-case scenarios with multivariate datasets and dynamic disasters. A case study with real datasets is given to explain the applicability, usability and practicability of the proposed system. [ABSTRACT FROM AUTHOR]

This paper presents SSRDVis, a visual approach to effectively summarize event sequences and interactively detect rare behaviors. SSRDVis is mainly composed of three components: (1) a sequence embedding module for learning effective feature vectors of sequences, (2) a sequence grouping and summarization module to find representative clusters and patterns in the dataset, (3) a rare detection module to discover and explain the rare cases. The sequences are embedded into vector space via "mixed-ngram2vec," which is adapted from "word2vec." Then, unsupervised learning models could be applied to group similar sequences and detect anomalies in the vector space. Furthermore, sequential pattern graphs are built to provide a compact and semantic summarization of sequences. These components work together to present both overall sequential patterns and abnormal behaviors in one visual interface. We have demonstrated the feasibility of our approach by applying it to analyze Web clickstreams. Experimental results have shown that our approach could help identify noticeable patterns from a large number of event sequences, especially for rare behaviors. [ABSTRACT FROM AUTHOR]

Electron Microscopy Data Bank (EMDB) provides biological researchers with a platform for uploading and downloading data obtained mainly by electron microscopy. With the continuous expansion of the total amount of data stored in EMDB in recent years, its influence has also been continuously improved. In order to make it easier for domain scientists to make better use of these data, we analyzed the EMDB meta-information structure and developed a visual analytics system, called VASEM, to provide convenient data exploration, searching, and analysis tools. Domain scientists can use our system to interactively filter massive EMDB data into relatively small subset for analysis and obtain subset depiction from visualization results. A modified parallel coordinates plot suitable for large-scale multi-dimensional ranking data visualization is also proposed for EM data subset depiction tasks. We present two usage scenarios to describe how the proposed system achieves the design goals. [ABSTRACT FROM AUTHOR]

Exploring relationships in complex datasets is one of the challenges in today's big data era. The graph-based visualization approach, which integrates the advantages of graph analysis theory and visualization technologies and combines machine and human intelligence, has become an effective means for analyzing various relationships in complex datasets. In this paper, we first introduce a graph-based visual analytics model for associated data. Then, we summarize seven typical visualization methods for associated data according to their layout features, including their node-link diagram, adjacency matrix, hypergraph, flow diagram, graphs with geospatial information, multi-attribute graph, and space-filling diagram and discuss their advantages and disadvantages. We describe current graph simplification and interaction techniques, including graph filtering, node clustering, edge bundling, graph data dimension reduction, and topology-based graph transformation. Finally, we discuss the potential challenges and developmental trends of the research direction. [ABSTRACT FROM AUTHOR]

The FY-3C satellite returns more than 250,000 occultation data each year, each of which records high vertical resolution profiles of various climatic parameters in the range of 0–1000 km above the Earth. These long-term, stable, and globally distributed observations can be used to improve climate modeling and analyze trends in spatial climate. However, traditional methods based on small samples meet challenges in analyzing the multi-dimensional occultation data in large volume. In this paper, we developed OccVis-a visual analytics system for multi-dimensional and multi-scale comparative analysis of occultation data. With a novel workflow, a series of data processing methods are proposed to support data correcting, patching, and clustering. Moreover, a matrix view with two modes is presented for overview, and a detail view along with time series view is provided for further analysis. Domain scientists can easily utilize our system to visually and interactively explore multivariable occultation data at different time and space scales. Finally, we conduct case studies in climate modeling of ionosphere and obtain several preliminary results to demonstrate the usage and effectiveness of our system. [ABSTRACT FROM AUTHOR]

Abstract: It is a feasible way to investigate the mental health status of college students through University Personality Inventory (UPI). However, the analysis of original UPI data is a difficult and tedious task, because the traditional software always requires the users to select and reload slices of data by hand. Also, the limited constant functions provided by the software cannot meet the various requirements of domain experts. In this paper, we propose VisUPI, a specialized visualization framework for UPI datasets, aiming at an in-depth analysis of the mental health status of college students. A circular view is firstly designed to layout the questionnaires and visualize the answers of individuals. Then, a decision tree model is employed to classify the investigated students, and a radial hierarchy chart is designed to present the relationship of different groups of students. According to the prior knowledge of UPI, we restructure the network relationship of specified questionnaires, and use the force-directed model to layout the questionnaires, enabling users to better perceive the answers of those students with serious mental health problems. Furthermore, multi-dimensional scaling is used to visualize the dissimilarity between different questionnaires, and the statistical answers are presented through bar charts in detail. Finally, the effectiveness and scalability of VisUPI are demonstrated through case studies with the real-world datasets and the domain-expert interviews.Graphical Abstract: [ABSTRACT FROM AUTHOR]

Abstract: Outlier detection and clustering are important to analyze trajectory. While many algorithms have been developed to tackle these issues, they lack the combination with visualization to enable the involvement of human intelligence during the analyzing process. We propose a visual framework called M3, which combines data mining algorithms with visualization technique through three coordinated views: Map, MST, and FSDMatrix. Map view displays the spatial information of trajectories. MST is a minimum spanning tree, which presents the relationships between trajectories. In MST, each node represents a trajectory; edges between nodes denote the Fréchet distance between trajectories. FSDMatrix shows a matrix of pairwise Free Space Diagram to assist in detecting outliers and clustering trajectories. The three views are interacted with each other. Through case studies, we discuss the applicability of our framework and demonstrate the convenience brought by it.Graphical abstract: [ABSTRACT FROM AUTHOR]

Abstract: Bike-sharing systems have grown tremendously worldwide in the recent years. Understanding the user activities in urban areas is invaluable, especially for bike rebalance and urban planning. However, it is difficult to directly capture the user activity patterns from the bike-sharing data due to its sparse and discontinuous characteristics. In the recent years, many methods have been explored to visualize the user activity patterns. Many previous methods focused on visually presenting the temporal and spatial distribution directly. In this paper, we construct a tensor based on the spatial, temporal, and user information of the bike-sharing data, and employ tensor factorization to extract latent user activity patterns. To facilitate the users to analyze and understand these patterns, a visual analytics system is designed to interactively explore these patterns from the spatial, temporal, and user dimensions and compare these patterns in/between cities. We demonstrate the effectiveness of our system via case studies with real-word datasets.Graphical abstract: [ABSTRACT FROM AUTHOR]

Abstract: Many prevalent visualization packages can be used to visualize the GDP data from different perspectives. However, it is difficult to integrate these visualizations and provide a comprehensive analysis to assist users get deeper insights into the various economic features of GDP data, due to its spatio-temporal and multidimensional attributes. In this paper, we propose a visualization tool for the analysis of spatio-temporal multidimensional GDP data, aiming at the combination of the extraction of economic clusters in a time period and the track of dynamic feature evolutions across time periods. MDS is first employed to reduce the multiple dimensions of GDP data, in which the attributes used to achieve similarity matrix are selected interactively by users, according to their requirements. The 2D coordinates obtained by MDS are further clustered based on a hierarchical clustering scheme, allowing the analysts to visually capture the economic features of interest in a time period. We also design a temporal visualization to visually present the dynamic changes of clusters, which largely helps users track the various evolutions of economic features. In addition, stability is defined to evaluate the disorder of clusters between adjacent time periods and used to map meaningful colors to different glyphs in the visualizations. A rich set of interactions are further provided to help users highlight and explore economic features of interest. We demonstrate the usefulness of our system in two case studies based on a real-world GDP data of China.Graphical abstract: [ABSTRACT FROM AUTHOR]

Interpersonal ties, such as strong ties and weak ties, describe the information carried by an edge in social network. Tracking the dynamic changes of interpersonal ties can thus enhance our understanding of the evolution of a complex network. Nevertheless, existing studies in dynamic network visualization mostly focus on the temporal changes of nodes or structures of the network without an adequate support of analysis and exploration of the temporal changes of interpersonal ties. In this paper, we introduce a new visual analytics method that enables interactive analysis and exploration of the dynamic changes of interpersonal ties. The method integrates four well-linked visualizations, including a scatterplot, a pixelbar chart, a layered graph, and a node-link diagram, to allow for multi-perspective analysis of the evolution of interpersonal ties. The scatterplot created by multi-dimensional scaling can help reveal the clusters of ties and detect abnormal ties, while other visualizations allow users to explore the clusters of ties interactively from different perspectives. Two case studies have been conducted to demonstrate the effectiveness of our approach. Graphical abstract: [ABSTRACT FROM AUTHOR]

Visualization has evolved into a flourishing research field in recent 30 years. There are substantial visualization methodologies and applications published every year. Most of literature surveys focus on reviewing the state-of-art techniques in a certain direction in-depth. In this work, we conduct a cross-section survey by taking all the latest literatures as a whole, to obtain insights into the ecology of Information Visualization and Visual Analytics field in 2016. Center around 70-related publications in the IEEE VIS, we perform a mixed quantitative and qualitative analysis to report the current research progress, including statistical overview as well as detailed research topics. Graphical Abstract: [ABSTRACT FROM AUTHOR]