Social media platforms, particularly Twitter, generate vast volumes of user-generated content that reflect real-time human emotions and opinions. Emotion analysis of such data offers significant value in areas including mental health monitoring, public sentiment tracking, and opinion mining. In this paper, we present a machine learning-based approach for automatically classifying emotions expressed in tweets into six distinct categories: joy, sadness, anger, fear, love, and surprise. The proposed system employs Natural Language Processing (NLP) techniques for text preprocessing, including tokenization, stopword removal, and lemmatization using the NLTK library. Feature extraction is performed using the Term Frequency-Inverse Document Frequency (TF-IDF) vectorization method with unigram and bigram representations. Three supervised machine learning algorithms are evaluated — Logistic Regression, Multinomial Naive Bayes, and Linear Support Vector Machine (LinearSVC) — and compared based on accuracy, precision, recall, and F1-score metrics. Experiments are conducted on the publicly available dairai/emotion dataset comprising over 20,000 labeled tweets. Results demonstrate that Logistic Regression with TF-IDF achieves the highest classification accuracy, outperforming the other models. The study confirms that classical machine learning models, when combined with effective text preprocessing and feature engineering, provide a reliable and computationally efficient framework for multi-class emotion detection in short social media text.

Preparing crime records, First Information Reports (FIRs), and charge sheets by hand continues to be a serious obstacle in policing due to human mistakes, procedural slowdowns, and inconsistent record-keeping. To overcome these shortcomings, this study presents an AI-powered framework that handles FIR creation, crime categorization, IPC-based charge sheet drafting, and punishment estimation automatically. The framework combines machine learning, natural language processing, and legal data analytics to identify offense categories, link them to suitable IPC sections, and produce structured legal paperwork. Logistic Regression paired with text vectorization methods allows dependable crime classification, while predictive analytics assists in estimating sentences using past case records. This method boosts operational productivity, lowers dependence on manual effort, and guarantees consistency and transparency across law enforcement processes. Future developments may involve blockchain-based data protection and live integration with national crime databases.

The accurate state of charge (SOC) estimation of lithium-ion batteries (LIBs) is crucial for ensuring the safe operation of electric vehicles. This article proposes a two-stage Kalman filter (TS-KF) method for SOC estimation of LIBs. In the first stage, a Kalman filter (KF) algorithm is designed to obtain an adaptive estimate for fitting coefficients of open circuit voltage and SOC relationship, which facilitates its use in the measurement equation of the second stage. In the second stage, an adaptive cubature Kalman filter (ACKF) algorithm based on fractional-order model (FOM) is designed for SOC estimation, where FOM comprises a pair of resistance-constant elements and a Warburg element (FO-RCW). Multiswarm cooperative particle swarm optimizer (MCPSO) is employed to identify parameters of the FO-RCW model based on experimental data. An augmented state equation, which includes SOC and parameters, is constructed using the identified parameters as initial values and the augmented vector method. Then a TS-KF method is proposed to obtain a high-precision SOC estimation, and achieve parameter estimation in real-time. The effectiveness and superiority of TS-KF method is validated via experiments.

Kinematically redundant parallel mechanisms (KR-PMs) exhibit complex error coupling and error amplification due to the existence of redundant branches, which poses enormous challenges to kinematic calibration. To address the key issue of insufficient excitation of error parameters caused by measurement noise and local convergence in traditional pose optimization algorithms, this study proposes a hybrid GOA-IOOPS measurement pose optimization algorithm for the kinematic calibration of the 2PUPR-PRPU KR-PM. First, based on the error mapping Jacobian matrix, a kinematic error model integrating error parameters of redundant branches is established using the closed-loop vector method and numerical differentiation. Innovatively, the grasshopper optimization algorithm (GOA) is fused with the iterative one-by-one pose search (IOOPS) algorithm: GOA undertakes global exploration to avoid local optima, while IOOPS performs local refinement to maximize the observability index O 3 . For parameter identification, a regularized nonlinear least squares method based on the Levenberg-Marquardt (LM) algorithm is adopted to balance convergence speed and robustness. Numerical simulations confirm that the GOA-IOOPS algorithm outperforms the traditional IOOPS algorithm and random selection method. Prototype experiments verify that the proposed calibration method significantly improves the positioning accuracy of the 2PUPR-PRPU KR-PM.

Abstract This study investigates the application of the Damage Locating Vector (DLV) method for detecting and localizing damage in two-dimensional truss structures using modal parameters obtained from two sources: ideal data from Modal Analysis in SAP2000 and simulated experimental data from Impact Tests processed via the Frequency Domain Decomposition (FDD) method. Numerical simulations were performed on a 2D truss model under various single and multiple damage scenarios, represented by reductions in member cross-sectional areas. Damage localization was evaluated using the Weighted Stress Index (WSI), where values below one indicated potential damage. The results show that the DLV method effectively identified damage when using modal data from SAP2000, even for minor damage levels of 10%, while lower accuracy was observed when using data derived from simulated Impact Tests. Nevertheless, damaged elements generally exhibited lower WSI values than undamaged ones, confirming the method’s sensitivity and robustness under different data conditions. This research contributes to the advancement of vibration-based damage detection by validating the DLV method’s applicability to 2D truss systems. The findings provide practical implications for nondestructive monitoring and preventive maintenance of civil infrastructure, supporting safer, more reliable, and sustainable structural systems.

Initial Orbit Determination for Too-Short-Arc Targets from Space-Based Observations via a Normal Vector Method

Spatial Fourier processing is commonly implemented using 4f optical architectures, in which spatial filtering and imaging are carried out on physically distinct Fourier and image planes. Our work proposes a metasurface-based imaging system. By introducing a quadratic phase factor at the object plane to pre-modulate the optical field, and integrating a metasurface with a filtering function and a focusing phase profile at the spectral plane, the system simultaneously executes frequency-domain filtering and spatial-domain imaging. This enables four-channel (x, y, 45°, and 135°) polarization-resolved edge imaging in a single exposure. Finally, by capturing images from four polarization channels and employing the Stokes vector method for polarization-state retrieval, the system reconstructs the original polarization information. This synchronous tuning scheme for spatial-domain and Fourier-domain information offers a novel, to the best of our knowledge, solution for compact optical processors and on-chip polarization imaging.

This study evaluated the early outcomes of hyperopic small incision lenticule extraction (SMILE) using the VisuMax 800 femtosecond laser in a Korean cohort, focusing on safety, efficacy, refractive predictability, and centration accuracy. This retrospective study included 11 eyes from 7 patients who underwent hyperopic SMILE using the VisuMax 800 between December 1, 2024, and October 31, 2025, at Onnuri Eye Hospital in Jeonju, Korea. All eyes completed at least 3 months of postoperative follow-up. Preoperative and postoperative evaluations included uncorrected and corrected distance visual acuity (UDVA and CDVA), manifest refraction, tomography, and wavefront aberrometry. Outcomes were analyzed using the standard nine graphs for refractive surgery. Astigmatic changes were assessed using the Alpins vector method. Optical zone decentration was calculated using postoperative Placido-Scheimpflug topography. The median preoperative spherical equivalent (SEQ) was +2.50 D [interquartile range (IQR), 1.94 to 4.12]. At 3 months, eight eyes (72.7%) were within ±0.50 D and all eyes (100%) were within ±1.00 D of the residual SEQ. The median efficacy index was 0.90 (IQR, 0.80 to 0.95). CDVA was unchanged in 63.6% of eyes and decreased by one line in 36.4%; no eye lost two or more lines. The median safety index was 0.86 (IQR, 0.80-1.00). RMS higher-order aberrations increased by +0.38 μm (IQR, 0.09 to 0.52). The median vertex-based decentration was 0.064 mm (IQR, 0.036 to 0.091). Decentration demonstrated a significant positive correlation with RMS HOA (ρ = 0.709, p = 0.014). Hyperopic SMILE using the VisuMax 800 produced favorable early clinical outcomes with high centration accuracy as well as shorter surgical time. Larger prospective studies with longer follow-up are necessary to determine long-term stability.

In this study, we present the AIM Review Tool, a modern web-based application that integrates active and supervised machine learning to accelerate the screening of publications for systematic reviews. AIM Review combines advanced text vectorization methods with machine learning models executed directly in the web browser, enabling rapid and privacy-preserving analysis. Unlike existing tools, AIM Review uniquely incorporates nested cross-validation and semi-automated screening strategies, enhancing both efficiency and precision in evidence synthesis. Using six real-world case studies across various topics, we demonstrate substantial workload reductions through active learning, with the percentage of publications not requiring screening while achieving ≥95% recall (WSS95%) ranging from 20% to 95%. Supervised learning pipelines trained on a subset of screened records predicted the relevance of unscreened publications with balanced accuracies between 75% and 87%. AIM Review provides a flexible, scalable, and accessible solution for large-scale literature screening and can be readily integrated into existing manual workflows.

More quantitative evidence is necessary on the link between livelihood resilience and livelihood adaptive capacity (LAC) in disaster resettlement. This study used 459 field research data collected from Ankang Prefecture, southern Shaanxi, China, examining how livelihood resilience influences adaptive capacity in the context of disaster-induced relocation. The resilience of rural household livelihood systems is described in terms of two components, general resilience, and specific resilience, which are quantified using the space vector method from systems engineering. The awareness, ability, and action framework is used to measure the LAC of rural households, and quantile regression is applied to explore the impact of livelihood resilience on LAC. Guided by the Sustainable Livelihoods Approach (SLA) and awareness, ability, action framework, we differentiate between cognitive, resource-based, and behavioral dimensions of adaptive capacity. The space vector method further reveals that individual adaptive capacity is reinforced by community-level resilience. It is found that: 1. livelihood resilience has a significant positive effect on high LAC levels, with the strongest effects observed at lower quantiles; as livelihood resilience increases, LAC also increases significantly. For rural households with low LAC levels, the impact is not significant; 2. general resilience and education have significant positive effects on all levels of LAC, with high levels being the most affected; 3. specific resilience has a significant negative effect on the lowest level of LAC only, and no significant effect on other levels. This study deepens our understanding of the relationship between livelihood resilience and LAC in the context of disaster resettlement, while testing the relationship between the two provides a methodological contribution to the study of disaster resettlement and community development.

Objectives . This study focuses on the development and investigation of a generalized nonlinear Support Vector Machine (SVM) method incorporating an adaptive transformation of the feature space. Its aim is to improve computational efficiency while maintaining high classification accuracy. The binary classification problem is used as a case study. The main objective of the research is to quantitatively evaluate the performance of the proposed approach when compared to classical SVM models using fixed kernel functions, and to analyze how the transformation parameters affect classification quality. Methods . The proposed approach involves a preliminary transformation of the input data using a learnable nonlinear mapping with a fixed structure. This mapping is implemented as a composition of elementary functions and is parameterized by a limited number of trainable weights which allows control over model complexity. A linear SVM with L2 regularization is applied after the transformation. The model is trained using conventional, unconstrained numerical optimization methods. The classification quality is evaluated using the Accuracy metric averaged over 10-fold cross-validation. The work also studies the behavior of the model with varying feature space dimensionality. In addition, computational complexity is analyzed in terms of the number of operations and inference time required on test datasets. Results . Numerical experiments demonstrate that the proposed model significantly reduces classification time when compared to a polynomial-kernel SVM, while maintaining a comparable level of accuracy. The runtime analysis confirms that the proposed approach scales much better than traditional kernel methods. At the same time, the structure of the model remains interpretable and can be further adapted to the specifics of the application domain. Conclusions . The method developed provides an efficient alternative to traditional kernel-based algorithms. Through the use of a parameterized transformation of the feature space, the method enables adaptability, interpretability, and scalability, making it promising for practical applications in machine learning tasks.

A relative configuration vector method for solving geometrically exact beam problems

Dynamical asymptotic analysis to a (3+1)-dimensional B-type Kadomtsev–Petviashvili equation: The superposition formulas of rational solutions and interaction solutions under the bilinear vector method

Interpreting genomics deep learning models remains challenging. Existing feature attribution methods largely focus on scoring individual bases or extracting global DNA motifs from one-hot encoded inputs, leaving them unable to assess broader genomic features such as chromatin accessibility or sequence annotations. Concept attribution methods offer an input-agnostic global interpretation framework, yet they have not been systematically applied to interpret neural network applications in genomics.We present the first application of concept attribution to interpret genomics deep learning models by adapting the Testing with Concept Activation Vectors (TCAV) method. We introduce Testing with PCA-projected Concept Activation Vectors (TPCAV), which improves upon the original method by using a PCA-based decorrelation transformation to address the correlated and redundant embedding features common in genomics models. We also introduce a strategy for extracting concept-specific input attribution maps. We evaluate our approach by interpreting influential biological concepts across a diverse set of genomics models spanning multiple input representations and prediction tasks.We demonstrate that TPCAV provides more reliable DNA motif interpretation than TCAV and is comparable to TF-MoDISco on one-hot coded DNA-based transcription factor binding prediction models. Beyond motif interpretation, TPCAV enables robust interpretive analysis of more general concepts such as repetitive elements and chromatin accessibility and generalizes to tokenized foundation models as well as models incorporating chromatin signal inputs. We further show that TPCAV can identify representative transcription factor binding sites associated with specific concepts, motivating downstream investigation of distinct binding mechanisms. Overall, TPCAV provides a flexible and robust complement to existing model interpretation techniques.

<title>Abstract</title> Gyrocompassing is the process of using the rotation of the Earth to determine heading. This approach is the state-of-the-art solution when precise and robust heading is needed, such as in ship navigation. Currently, this procedure is performed using a high-grade inertial measurement unit, consisting of an accelerometer and a gyrometer. To circumvent parasitic motions from the ship, the Earth’s rotation is not directly measured by the gyrometer. The commonly adopted solution is to measure the rotation of the gravity vector in the inertial frame. A curated, straightforward method for gravity vector fitting based on singular value decomposition was developed here to provide a baseline algorithm. This method was compared with a second algorithm based on inertial trajectory fitting. Our results show how a trajectory-fitting algorithm can improve robustness and accelerate gyrocompassing compared with a traditional gravity analysis algorithm.

Background and Objectives: To evaluate the magnitude, axis and age-related changes in corneal astigmatism in patients before cataract surgery. Materials and Methods: In this retrospective, cross-sectional, and observational study, data from 2152 eyes that underwent phacoemulsification were evaluated. Keratometric values were obtained using the IOL Master 500 device. The frequency, magnitude and axis of corneal astigmatism were determined. The astigmatism axis was categorized as with the rule (WTR), against the rule (ATR), and oblique astigmatism. Quantitative analysis was performed using the power vector method (J0 and J45). The distribution and characteristics of corneal astigmatism data according to age were analyzed. Results: The mean age of the patients was 70.56 ± 8.88 years (range 40-94 years) and 1010 (46.9%) were males. Mean corneal astigmatism, J0 and J45 values were 0.96 ± 0.72, 0.05 ± 0.51, 0.01 ± 0.30 diopters (D), respectively. The most common range of magnitudes was 0.50-0.99 D with 38.8%, followed by <0.50 D (25.3%), 1.00-1.49 D (20.3%), and 1.50-1.99 D (8.7%). The cubic regression curve showed a U-shaped nonlinear relationship between age and corneal astigmatism (p < 0.001). The most common type of astigmatism was WTR with 43.4%, followed by ATR with 37.5% and oblique astigmatism with 19.1%. With the increase in age, the astigmatism axis gradually changed from WTR to ATR. There was a linear trend in the rate of these types of astigmatism across age groups (p < 0.05). Additionally, in patients under 65 years of age, WTR astigmatism was negatively correlated with age, while in patients 65 years of age and older, ATR astigmatism was positively correlated with age (r = -0.217, p < 0.001; r = 0.153, p < 0.001, respectively). Linear regression analyses revealed that the J0 value decreased significantly with age, whereas J45 showed no significant relationship. Specifically, J0 decreased by 0.014 D per year of age (95% confidence interval [CI], 0.011-0.016; p < 0.001). Conclusions: The results obtained in this study may provide information to guide surgeons in the management of astigmatism and the choice of toric intraocular lens in cataract surgery.

Based on screw theory, this study analyzes the rehabilitation motions of the hip, knee, and ankle joints in a seated posture, determines the required degrees of freedom (DOFs) for each motion, and experimentally measures their respective motion ranges. Utilizing the application of screw theory in statics, a 2RPS-UPU reconfigurable parallel mechanism is designed, featuring three motion modes: 2R1T, 2T1R, and 2R2T. The corresponding rehabilitation motions for each mode, along with the overall rehabilitation workflow, are described in detail. Inverse kinematic equations for each motion mode are derived using the closed-loop vector method, and the corresponding workspaces are analyzed to confirm their alignment with rehabilitation requirements. Furthermore, Jacobian matrices for the three modes are obtained, and performance evaluations are conducted with respect to dexterity and stiffness. An improved multi-objective genetic algorithm is then employed to optimize key structural parameters, aiming to enhance global dexterity, stiffness, and workspace coverage. The optimization results indicate a 30.7% increase in workspace, a 12.3% improvement in global dexterity, the global maximum deformation index is reduced by about 40%. By enabling multi-mode switching and structural optimization, the proposed reconfigurable mechanism offers a highly adaptable and efficient solution for lower-limb rehabilitation, demonstrating strong potential for clinical application.

This study used DFT to investigate the mechanism and stereoselectivity origin of isothiourea-catalyzed acylation of planar chiral paracyclophanols with isobutyric anhydride via the projection of orbital coefficient vector (POCV) method.

The study of accompanying vehicles is a hot topic in the field of intelligent transportation. Because of the multiple selectivity of the traffic path and the loss of sampling in traditional companion vehicles discovery, the method based on path similarity mining will result in the omission of the companion candidates. This paper recognizes that the upstream and downstream relevance of trajectory intentions in traffic is similar to the contextual relevance of text semantics, as inspired by the semantic similarity of texts. Simultaneously, taking into account the generalization and tolerance of semantic processing, a companion vehicle discovery method based on “trajectory semantics” similarity is proposed. First, this paper proposes a trajectory semantic vectorized representation method trajectory semantic to vector (TS2vec), which realizes the low‐dimensional dense vectorization of the trajectory in the context of dynamic time slicing of the trajectory, fusion of the temporal and spatial characteristics of the trajectory, and text information. Then, based on the “trajectory pair,” this paper proposes the trajectory pair bidirectional GRU (TPBi‐GRU) model. This paper constructs forward and reverse subnetworks using the trajectory pair set—which is made up of the actual trajectory and ts sampled trajectories—realizes parameter transfer and contribution during training; gains a thorough understanding of trajectory semantics; and mines the internal relationship between vehicles more effectively. Finally, given the difference in the degree of contribution of the road shape in forming the adjoint pattern, and the sensitivity of the attention mechanism to local features, the attention mechanism is used to weigh the key nodes that affect the trajectory shape in order to obtain a more accurate trajectory representation. The experimental results show that the method in this paper can discover local and overall concomitant patterns more effectively and effectively overcome the interference of multiple selectivity of traffic paths on concomitant pattern mining.

The frontier of quantum computing (QC) simulation on classical hardware is quickly reaching the hard scalability limits for computational feasibility. Nonetheless, there is still a need to simulate large quantum systems classically, as the Noisy Intermediate Scale Quantum (NISQ) devices are yet to be considered fault tolerant and performant enough in terms of operations per second. Each of the two main exact simulation techniques, state vector and tensor network simulators, boasts specific limitations. This article investigates the limits of current state-of-the-art simulation techniques on a test bench made of eight widely used quantum subroutines, each in different configurations, with a special emphasis on performance. We perform both single process and distributed scaleability experiments on a supercomputer. We correlate the performance measures from such experiments with the metrics that characterise the benchmark circuits, identifying the main reasons behind the observed performance trends. Specifically, we perform distributed sliced tensor contractions, and we analyse the impact of pathfinding quality on contraction time, correlating both results with topological circuit characteristics. From our observations, given the structure of a quantum circuit and the number of qubits, we highlight how to select the best simulation strategy, demonstrating how preventive circuit analysis can guide and improve simulation performance by more than an order of magnitude.