Search

Recently, implicit neural representations (INRs) have demonstrated significant capabilities in handling 3D volume data, especially in the context of data compression. However, the majority of research has predominantly focused on structured grids, which are not commonly found in scientific domains, particularly in physics. To address this limitation, we propose an unstructured grid reduction method via implicit neural representation (UGINR). UGINR employs a divide-and-conquer approach; specifically, we segment the large-scale data into pieces based on values. Subsequently, we employ an INR network for each piece to learn its distinctive features. Finally, we integrate these individual networks to achieve the compression goal. To ensure compatibility with established research methods, we sample only the vertices of each cell in the unstructured grid. Through weight quantization, our model can achieve a high compression ratio. To illustrate the effectiveness of the proposed method, we conduct experiments on various datasets, demonstrating our approach's robustness in scientific visualization and large-scale data compression. [ABSTRACT FROM AUTHOR]

As the importance of research data gradually grows in sciences, data sharing has come to be encouraged and even mandated by journals and funders in recent years. Following this trend, the data availability statement has been increasingly embraced by academic communities as a means of sharing research data as part of research articles. This article presents a quantitative study of which mechanisms and repositories are used to share research data in PLOS ONE articles. We offer a dynamic examination of this topic from the disciplinary and temporal perspectives based on all statements in English-language research articles published between 2014 and 2020 in the journal. We find a slow yet steady growth in the use of data repositories to share data over time, as opposed to sharing data in the article and/or supplementary materials; this indicates improved compliance with the journal's data sharing policies. We also find that multidisciplinary data repositories have been increasingly used over time, whereas some disciplinary repositories show a decreasing trend. Our findings can help academic publishers and funders to improve their data sharing policies and serve as an important baseline dataset for future studies on data sharing activities. [ABSTRACT FROM AUTHOR]

Large-scale numerical simulations are capable of generating data up to terabytes or even petabytes. As a promising method of data reduction, super-resolution (SR) has been widely studied in the scientific visualization community. However, most of them are based on deep convolutional neural networks or generative adversarial networks and the scale factor needs to be determined before constructing the network. As a result, a single training session only supports a fixed factor and has poor generalization ability. To address these problems, this paper proposes a flow feature-enhanced implicit neural representation (FFEINR) for spatiotemporal super-resolution of flow field data. It can take full advantage of the implicit neural representation in terms of model structure and sampling resolution. The neural representation is based on a fully connected network with periodic activation functions, which enables us to obtain lightweight models. The learned continuous representation can decode the low-resolution flow field input data to arbitrary spatial and temporal resolutions, allowing for flexible upsampling. The training process of FFEINR is facilitated by introducing feature enhancements for the input layer, which complements the contextual information of the flow field. To demonstrate the effectiveness of the proposed method, a series of experiments are conducted on different datasets by setting different hyperparameters. The results show that FFEINR achieves significantly better results than the trilinear interpolation method. [ABSTRACT FROM AUTHOR]

In the scientific visualization and data analysis workflow, data transmission bandwidth and memory resources have become the main bottlenecks when handling large-scale volume data. As a direct and effective scheme, data reduction is generally used to decrease data movement overhead and memory usage. However, it is still a challenge to obtain visualization results from reduced data without losing too many features. This paper proposes a visualization scheme for large-scale volume data based on enhanced super-resolution generative adversarial networks (ESRGAN) and designs a reduction-restoration workflow. Firstly, in order to reduce memory footprint, we propose an error-controlled data reduction method to delete data in 3D space, which is based on octree. Secondly, rendered images with loss of details are generated by performing volume rendering on reduced data. Lastly, to obtain feature-lossless visualization results, we apply ESRGAN to restore the details of rendered images. Based on the above scheme, the dual goals of data reduction and visual feature retention can be realized. Finally, the effectiveness of the proposed method is demonstrated by evaluating the performance of data reduction and visual restoration. [ABSTRACT FROM AUTHOR]

Purpose: Research objects, such as datasets and classification standards, are difficult to be incorporated into a document-centric framework of citations, which relies on unique citable works. The Diagnostic and Statistical Manual for Mental Disorder (DSM)—a dominant classification scheme used for mental disorder diagnosis—however provides a unique lens on examining citations to a research object, given that it straddles the boundaries as a single research object with changing manifestations. Design/methodology/approach: Using over 180,000 citations received by the DSM, this paper analyzes how the citation history of DSM is represented by its various versions, and how it is cited in different knowledge domains as an important boundary object. Findings: It shows that all recent DSM versions exhibit a similar citation cascading pattern, which is characterized by a strong replacement effect between two successive versions. Moreover, the shift of the disciplinary contexts of DSM citations can be largely explained by different DSM versions as distinct epistemic objects. Practical implications: Based on these results, the authors argue that all DSM versions should be treated as a series of connected but distinct citable objects. The work closes with a discussion of the ways in which the existing scholarly infrastructure can be reconfigured to acknowledge and trace a broader array of research objects. Originality/value: This paper connects quantitative methods and an important sociological concept, i.e. boundary object, to offer deeper insights into the scholarly communication system. Moreover, this work also evaluates how versioning, as a significant yet overlooked attribute of information resources, influenced the citation patterns of citable objects, which will contribute to more material-oriented scientific infrastructures. [ABSTRACT FROM AUTHOR]

The data paper is an emerging academic genre that focuses on the description of research data objects. However, there is a lack of empirical knowledge about this rising genre in quantitative science studies, particularly from the perspective of its linguistic features. To fill this gap, this research aims to offer a first quantitative examination of which rhetorical moves—rhetorical units performing a coherent narrative function—are used in data paper abstracts, as well as how these moves are used. To this end, we developed a new classification scheme for rhetorical moves in data paper abstracts by expanding a well‐received system that focuses on English‐language research article abstracts. We used this expanded scheme to classify and analyze rhetorical moves used in two flagship data journals, Scientific Data and Data in Brief. We found that data papers exhibit a combination of introduction, method, results, and discussion‐ and data‐oriented moves and that the usage differences between the journals can be largely explained by journal policies concerning abstract and paper structure. This research offers a novel examination of how the data paper, a data‐oriented knowledge representation, is composed, which greatly contributes to a deeper understanding of research data and its publication in the scholarly communication system. [ABSTRACT FROM AUTHOR]