Journal of Visualization最新文献

Large language models (LLMs), such as ChatGPT, have demonstrated outstanding performance in various fields, particularly in natural language understanding and generation tasks. In complex application scenarios, users tend to engage in multi-turn conversations with ChatGPT to keep contextual information and obtain comprehensive responses. However, human forgetting and model contextual forgetting remain prominent issues in multi-turn conversation scenarios, which challenge the users’ conversation comprehension and contextual continuity for ChatGPT. To address these challenges, we propose an interactive conversation visualization system called C⁵, which includes Global View, Topic View, and Context-associated Q&A View. The Global View uses the GitLog diagram metaphor to represent the conversation structure, presenting the trend of conversation evolution and supporting the exploration of locally salient features. The Topic View is designed to display all the question and answer nodes and their relationships within a topic using the structure of a knowledge graph, thereby display the relevance and evolution of conversations. The Context-associated Q&A View consists of three linked views, which allow users to explore individual conversations deeply while providing specific contextual information when posing questions. The usefulness and effectiveness of C⁵ were evaluated through a case study and a user study.

Graphical abstract

Abstract

Carotid artery plaques can cause arterial vascular diseases such as stroke and myocardial infarction, posing a severe threat to human life. However, the current clinical examination mainly relies on a direct assessment by physicians of patients’ clinical indicators and medical images, lacking an integrated visualization tool for analyzing the influencing factors and composition of carotid artery plaques. We have designed an intelligent carotid artery plaque visual analysis system for vascular surgery experts to comprehensively analyze the clinical physiological and imaging indicators of carotid artery diseases. The system mainly includes two functions: First, it displays the correlation between carotid artery plaque and various factors through a series of information visualization methods and integrates the analysis of patient physiological indicator data. Second, it enhances the interface guidance analysis of the inherent correlation between the components of carotid artery plaque through machine learning and displays the spatial distribution of the plaque on medical images. Additionally, we conducted two case studies on carotid artery plaques using real data obtained from a hospital, and the results indicate that our designed carotid artery plaque analysis system can effectively assist clinical vascular surgeons in gaining new insights into the disease.

Graphical Abstract

Since the birth of software, fault localization has been a time-consuming and laborious task. Programmers need to constantly find faults in software through program logging, assertions, breakpoints, and profiling. In order to improve the debugging efficiency, many fault localization methods based on test cases have been proposed, such as program spectrum-based methods, and slice-based methods. However, these methods are far from the logic of actual debugging and still require programmers to use traditional methods. However, programmers cannot access the execution process of the program, they need to constantly modify breakpoints and repeatedly check variable values, which makes fault localization very time-consuming. After interviewing five experts in the field of visualization and software testing, we designed SFLVis to provide users with a new method to improve the efficiency of fault localization. We designed an algorithm to obtain the process of program execution and combined it with existing fault localization methods. The goal is to show users the execution results of test cases, source code logic, and the level of suspicion of statements, and reproduce the execution process of test cases. We designed rich interactive features to help users explore SFLVis and correlate information from various views to improve the efficiency of fault localization. To verify the effectiveness of SFLVis, we conducted a case study using the program in the Siemens Suite dataset and conducted group experiments and related interviews with 20 volunteers. The results show that SFLVis can effectively improve programmers’ efficiency compared with existing fault localization methods.

Graphical abstract

Abstract

The integrity of plate structures in numerous facilities and vehicles is essential for ensuring safety. Guided wave testing is a prominent non-destructive testing (NDT) technique, especially for wide plate or long pipe structures. It can be related to tomography techniques to visualize defect information. One way to obtain data for tomography is through experimentation. However, a numerical approach, such as a computational simulation, could also be a feasible option because it can efficiently handle various defect cases. In this study, a dynamic analysis was performed to acquire the guided wave signal on a plate containing a wall-thinning defect, for which previous studies were insufficient. Acquired signals are compared to each other, and studies have demonstrated that wall-thinning defects can be visualized. This approach to signal data acquisition is expected to enhance the efficiency of data collection in several fields, such as machine learning implementation in NDT.

Air pollution data provides important information on air quality, which can be used to assess the impact of atmospheric pollution on human health, the environment, and the economy, as well as to develop corresponding policies and measures to reduce pollutant emissions and improve air quality. In this paper, we propose a novel approach to improve the resolution of meteorological data via masked image modeling (PMIM) to generate high-resolution air pollution data. In order to apply the image masking modeling to process air pollution data, we convert the data format and use radial basis function visualization to generate smooth distribution maps of air pollution data. To generate high-resolution air pollution data, we design several different masking strategies and use the masked image modeling to simulate the reconstruction process from low-resolution grid data to high-resolution grid data, obtaining the reconstructed high-resolution grid images. Finally, we use the mapping relationship between the pixel colors of the reconstructed images and the air pollution data to generate high-resolution air pollution concentration data. In order to verify the effectiveness of the proposed method, we conduct comparative experiments using different masking strategies and test air pollution data of different resolutions. The results show that our method has good applicability and effectiveness in different situations.

Graphical abstract

Abstract

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.

Graphical abstract

Abstract

Maintaining the normal flight of drones is crucial for drone operators. Analyzing the operation status of drones and adjusting flight parameters are essential to achieve this goal. However, as drone technology continues to evolve, the volume and complexity of spatio-temporal data related to drone flight status have grown exponentially. The complexity of this data poses a challenge to effective visualization, which can impact operators’ analysis and decision-making. Currently, there is limited research on identifying flight attributes from a large collection of drone time series data. Two challenges were identified: (1) visual clutter from spatio-temporal data; (2) effective integration of time and space properties. By collaborating with domain experts, we addressed two challenges with DSTVis, a novel interactive system for operators to visually analyze spatio-temporal data of drones. For Challenge 1, we designed dynamic interactive views by abstracting and stratifying spatio-temporal data, enabling effective exploration of large amounts of data. For Challenge 2, a two-dimensional map is utilized to integrate time information and assist users in comprehending the spatio-temporal properties. The effectiveness of the system is evaluated with a usage scenario on a real-world historical dataset and received positive feedback from experts.

Graphic abstract

Abstract

Air quality analysis helps analysts understand the state of atmospheric pollution and its changing trends, providing robust data and theoretical support for developing and implementing environmental policies. Air quality data are typically represented as multivariate time series, which poses challenges due to the large amount of data, high dimensionality, and lack of labeled information. Analysts often struggle to discover internal relationships and patterns within the data. There is still significant room for improvement in related data mining and exploration methods, as issues such as perceptual burden and low efficiency must be addressed. To assist analysts in atmospheric pollution analysis, we propose an air quality visualization scheme based on feature extraction of multivariate time series data. We utilize the automated data modeling capability of deep learning and intuitive data visualization to help analysts explore and analyze complex air quality datasets. To extract features of air quality data effectively, we transform the multivariate time series feature extraction task into an automated deep learning self-supervised task and propose a feature extraction method called CTDCN for multivariate time series. Finally, we design and implement a visualization and analysis system for air quality multivariate time series. This system helps analysts discover potential information and patterns in air quality data, providing support and a foundation for informed decision-making. The system offers rich visualization views, allows users to change data modeling parameters, and interactively analyze and extract insights from the data through multiple views. Extensive experiments on UEA public datasets confirm CTDCN’s superior feature extraction capabilities, while case studies and user studies validate the effectiveness and practicality of our visualization approach.

Graphical abstract

Abstract

The multi-layer indices decision-making of complex industrial processes is the key to reducing costs and improving production efficiency. With the development of the Industrial Internet, a large number of industrial streaming data and intelligent algorithms have brought opportunities for optimizing plant-wide production indices. However, due to the strong dynamic and coupling of the production process, the intelligent system based only on the optimization algorithm cannot give practical data analysis suggestions and decision results, so a human–computer interactive visual analysis and index decision system are urgently needed. This paper combines multi-layer indices decision-making algorithms with 3D digital twin visual analysis technology to propose an intelligent decision-making system for mineral processing production indices based on 3D digital twin interactive visualization (DTIV). The DTIV system provides users a 3D digital twin modeling view from the production park, workshop, and equipment scenes. It adopts visualization technology that seamlessly integrates 3D and 2D to help users obtain indices decision input information and hidden data features from real-time stream data with different spatiotemporal data characteristics. In addition, the DTIV system also combines a multi-layer indices optimization decision-making algorithms engine and designs a human–machine interaction indices decision interface and indices decision execution visual analysis interface to improve users’ production perception and decision-making ability. Through our collaboration with domain experts, carefully designed interviews, and prototype system evaluation in a beneficiation plant, the effectiveness and usability of the system have been proven.

Graphic Abstract

This study investigated unsteady pressure measurements on a pitching OA309 airfoil at a Mach number of 0.1 using a fast-responding pressure-sensitive paint (fast PSP). Two commonly used data acquisition methods applicable to fast PSPs, namely the real-time intensity-based method and the single-shot lifetime-based method, were separately used to obtain the pressure distributions on the upper surface at a reduced pitching frequency (k = πfc/U_∞) of 0.074. The signal-to-noise ratio, influences of model motion, and temperature-induced errors associated with the two methods were compared to explore the advantages and disadvantages of the methods. The real-time intensity-based method outperformed the single-shot lifetime-based method in pressure measurements on moving models with very low speeds. Flow separation and reattachment were identified according to the temporal- and spatial-resolved pressure fields obtained through the real-time intensity-based method; finally, the effects of the pitching amplitude and the leading-edge vortex generators were studied. The results showed that flow separation was postponed as the pitching amplitude increased, while flow reattachment occurred earlier on the airfoil equipped with leading-edge vortex generators.

Graphical abstract