Given a massive dataset, data visualization (DV) could efficiently express the insights and summaries behind the massive raw data by employing vivid visual representations. To create suitable DVs, users are required to get a comprehensive understanding of the raw data and then transfer their ideas into DVs by composing a suitable and accurate specification in some declarative visualization languages (DVLs, e.g., Vega-Lite). A specification is a JSON object defining the properties of the DVs, like the selected data, the transformations, the visual details, and so on. Due to its complicated grammar and details, DV has quite a steep learning curve, even for data analysts. In this paper, we propose a new task named FeVisQA, referring to Free-form Question Answering over data Visualizations. More specifically,-given a raw dataset, a related DV (in the form of a specification), and a question, FeVisQA aims to predict a textual answer automatically. As a particular case of the general CodeQA (i.e., QA over general programming code like Python and Java) task, FeVisQA enables people to better comprehend data and its DVs by conducting logical reasoning when answering these questions. Since FeVisQA has not been studied in the literature, we first construct a benchmark dataset containing 152 datasets, 14,406 DVs, and 83,890 QA pairs. To tackle this new task, we design a novel neural network named FeVisQANet with advanced multi-modal encoder and adaptive decoder structures, and we also design a novel multi-step framework called VisQA for Multi-modal Large Language Models (MLLMs) based on Retrieval-augmented Generation (RAG) technology. Extensive experiments on our constructed datasets validate the rationale and effectiveness of this proposed FeVisQA task and the proposed model. While research on QA over text and table, machine reading comprehension, and CodeQA develops rapidly, prior works have yet to draw attention to question-answering over DVs. This study connects two important subareas, QA from the natural language process area and DV from the data engineering area. We hope this new dataset and model can serve as a helpful benchmark that would benefit the development of both fields.

The time–frequency (TF) characteristics of seismic signals make it possible to accurately identify the boundaries of natural gas. In particular, the transient-extracting transform (TET) improves the time resolution of the unstable signal TF representation (TFR) by extracting only the points on the group delay (GD) trajectory. However, the GD estimator of TET lacks certain accuracy for frequency-varying signals. To overcome this problem, this article proposes an improved GD (IGD) estimator. First, in the short-time Fourier transform (STFT) domain, based on the GD estimator of the frequency-varying signal, the GD estimator problem is transformed into a fixed-point problem, and the IGD estimator is constructed. Then, combining the advantages of TET, an improved TET was proposed. Finally, through the simulated geological model test of ITET, the root mean square error of the model’s reflectivity structure is only 0.001. Combined with the analysis of seismic data, the effectiveness of this method in the positioning of reflectivity structure is verified, which can provide strong support for subsequent seismic interpretation.

Time-frequency analysis (TFA) is widely used to describe the time-frequency (TF) features of seismic data, thanks to its superior capability in analyzing nonstationary signals. Among the commonly used TFA techniques, the time-reassigned synchrosqueezing transform (TSST) and transient-extracting transform (TET) provide a TF spectrum with high-time resolution for seismic signals. However, both TSST and TET cannot simultaneously emphasize both sharpening and reconstruction performance. This article proposes a transient matching squeezing transform (TMST) that integrates these two performances. First, drawing on the idea of TET, define the group delay (GD) band center to trace the GD trajectory of the signal. Then, explore the properties of the GD estimator and theoretically construct a matching GD (MGD) estimator based on these properties. Finally, incorporating the idea of TSST employs the MGD estimator to squeeze the dispersed TF coefficients to the GD band center, thereby sharpening the original TF spectrum while preserving the reconstruction of the original signal, as validated by numerical simulations. Furthermore, the application of the simulated geological model and field seismic data shows that the proposed TMST can effectively reveal geological features, making it a powerful tool for seismic signal analysis.

The time-reassignment method and time-synchrosqueezing transform show a good ability in impulse-like signal analysis. This kind of method calculates the group delay (GD) estimators at the spread TF locations first and then relocates the spread TF energy into the estimated GD trajectories to yield a high-concentration TF representation. However, computing the GD estimators for every TF energy point can lead to inaccurate location and energy diffusion. To address this issue, a new feature extractor called the second-order GD equation is proposed, which focuses only on the TF points on the GD to characterize the frequency-varying models withN-order amplitude and second-order phase. The theoretical analysis of the second-order GD equation is highlighted. By combining a fixed-point iterative algorithm with the extracting transform, we introduce a novel weighted transient-extracting transform based on the solutions of the second-order GD equation. This transform enhances TF distribution concentration while retaining reconstruction capability. Numerical simulations demonstrate that our proposed method improves the average performance for TF concentration by 3% across various noise levels and enhances the accuracy of GD location by over 0.2 within a SNR range of -1 dB to 20 dB, compared to current state-of-the-art TF analysis methods. Finally, the proposed TF transform is applied to analyze seismic data for low-frequency shadow attributes and thin layer characterization. The results clearly illustrate its effectiveness in seismic processing and interpretation.

Time-frequency (TF) analysis (TFA) is a vital spectral decomposition tool for reservoir characterization due to its superiority in analyzing nonstationary signals. Two newly developed TFA methods, the synchroextracting transform (SET) and transient-extracting transform (TET), provide a concentrated TF representation for seismic signals. However, the SET and TET are respectively only valid for harmonic signals with slow-varying instantaneous frequencies and transient signals of which the TF ridge almost parallels the frequency axis. This article proposes a centroid-oriented extracting transform (COET) to analyze mixed signals containing harmonic and transient features. First, we construct an objective function in the short-time Fourier transform (STFT) domain to link the most relevant signal features via the TF centroid. Then, the main direction of energy diffusion is defined by the TF centroid to separate the signal into harmonic and transient modes. Finally, a divide-and-conquer strategy is utilized to locally minimize the objective function in the frequency and time directions, respectively, thereby extracting the STFT coefficients at all optimal TF points while keeping the lossless reconstruction of the original signal. Such a COET provides both high time and frequency concentrations with strong noise immunity, as numerical signals elaborate. The application of field seismic data further indicates that the proposed COET clearly highlights the detailed information of different reservoir distributions, which is a promising tool for seismic spectral decomposition.

A time–frequency (TF) analysis (TFA) method plays a significant role in seismic signal processing. However, traditional seismic TFA techniques always provide blurred TF representations (TFRs) with low temporal resolution. The transient-extracting transform (TET) is an effective TFA tool that improves the temporal resolution of TFR of nonstationary signals. However, the TET suffers from nonreassigned point (NRP) problem, which can result in unreliable seismic interpretations. In this article, a novel TF transform is proposed to address this problem, named the time-reassigning transform (TRT). In the short-time Fourier transform (STFT) domain, this TRT searches for local maximum points of the spectral energy along the time direction instead of searching for group delay (GD) fixed points such as the TET. At these points, the STFT coefficients are reassigned into a sparser TFR while eliminating the NRP problem, as verified by two synthetic signals and a simulated seismic trace. Finally, application analysis on a geological model and field seismic data shows that the proposed TRT is a potential seismic data analysis tool.

Sedimentary cycle division is a critical task within sequence stratigraphy and provides significant guidance value in the exploration of gas resources. For the division of sedimentary cycles by analyzing the time-frequency (TF) curves of seismic data, the key lies in high-resolution highlighting of changes in dominant frequency induced by variations in layer thickness. To improve the energy concentration of TF analysis (TFA) methods, this article proposes an adaptive peak extracting transform (APET) for characterizing sedimentary cycle patterns. In this APET, we first establish a local peak criterion within the short-time Fourier transform (STFT) framework to obtain instantaneous frequency (IF) trajectories. Subsequently, a time-varying window function is constructed by Rényi entropy to determine the optimal window length at each moment. Finally, the TF coefficients are reallocated to the retrieved IF within the optimal window. The local peak criterion and the time-varying window function effectively enhance the TF resolution. The synthesized signals verify that the proposed APET exhibits highly focused TF representation (TFR). In addition, we apply APET to several typical sedimentary cycle models and field seismic data, indicating that the proposed method is a potential tool for sedimentary cycle analysis.

Thin interbed analysis has guiding significance for oil and gas exploration, and it is feasible to precisely predict their thickness variation through the time-frequency (TF) features of seismic signals. In this article, an auto-reassigning transform (ART) is proposed to analyze seismic thin interbeds. This ART retrieves the instantaneous frequency (IF) trajectory by local maxima, then determines the TF coefficients within the effective window by the 3σ criterion, and finally reassigns these coefficients automatically to the retrieved IF trajectory. Therefore, such ARTs provide a more concentrated TF representation with reconstruction capability. Synthetic signals and seismic data are employed to demonstrate the effectiveness of the proposed ART by comparing it with state-of-the-art TF analysis methods.