LLM-driven tools have significantly lowered barriers to writing SQL queries. However, user instructions are often underspecified, assuming the model understands implicit knowledge, such as dataset schemas, domain conventions, and task-specific requirements, that isn't explicitly provided. This results in frequently erroneous scripts that require users to repeatedly clarify their intent. Additionally, users struggle to validate generated scripts because they cannot verify whether the model correctly applied implicit knowledge. We present Cerebra, an interactive NL-to-SQL tool that aligns implicit knowledge between users and LLMs during SQL authoring. Cerebra automatically retrieves implicit knowledge from historical SQL scripts based on user instructions, presents this knowledge in an interactive tree view for code review, and supports iterative refinement to improve generated scripts. To evaluate the effectiveness and usability of Cerebra, we conducted a user study with 16 participants, demonstrating its improved support for customized SQL authoring. The source code of Cerebra is available at https://github.com/zjuidg/CHI26-Cerebra.

Trajectory data record spatiotemporal information and related attributes of moving objects. High-quality trajectory data can accurately reflect behavior patterns, providing a reliable foundation for traffic management, logistics optimization, and smart city planning. However, existing research on trajectory data quality management remains limited, lacking interactive tools for addressing quality issues and making it difficult for users to explore and resolve problems in trajectory datasets. To address this gap, we present TrajectoryCurer, an interactive visual analytics system designed for trajectory data quality management. Through expert interviews and synthesis of preprocessing operations from 30 relevant articles on trajectory visualization and management, we construct a taxonomy of 19 trajectory data quality issues across six dimensions. Based on this taxonomy, we design a multilevel visualization approach encompassing Trajectory View, Point View, Map View, and Data View, enabling users to observe and address quality issues across multiple dimensions through an integrated interface. We demonstrate the effectiveness and usability of TrajectoryCurer through two usage scenarios and expert evaluations.

Animated geographic visualization is a common way to communicate spatial processes and narratives. However, authoring such animations often requires trade-offs among learnability, control, and production effort. To explore a structured authoring approach, we analyzed 64 source videos containing geographic animation segments and synthesized a design space for geographic animation targets, attributes, and effects. Based on this analysis, we designed GeoAnimation, a domain-specific declarative grammar and interactive prototype for authoring common point, line, and region animations. We examined the prototype through two use cases and expert interviews with two visualization researchers. The results provide preliminary evidence that GeoAnimation is feasible for the studied authoring tasks, while broader usability, expressiveness, and performance require further evaluation.

Articles containing geographic information are widely distributed and commonly used in daily life, frequently incorporating geographic visualizations as illustrations. However, the creation of such articles remains cumbersome, necessitating authors to switch between authoring text and illustrations, thereby disrupting immersive writing. Our interviews corroborated this observation and revealed the primary challenge in the traditional process stems from the low synchronization frequency between text and geographic visualizations during creation, coupled with weak visual links, forcing users to mentally maintain this synchronization and thereby increasing their cognitive burden. In response, we developed GeoAuthor, which facilitates the interactive creation of geographic articles by automatically synchronizing text creation with geographic visualizations with rich visual links. This bidirectional approach ensures that the written content and visual representations remain consistent and mutually informative throughout the creation process. Our evaluation demonstrated the efficacy of GeoAuthor, indicating its capacity to streamline the process of creating geographic articles.

Deep learning has advanced time series forecasting, but periodicity drift, in which cycle timing and phase vary over time, remains a challenging problem. Existing methods predominantly model these sequences on fixed time grids, suffering from a limited ability to accommodate phase-related variation. To address this limitation, we propose POEM, a phase-aware forecasting framework based on latent feature rotation using the special orthogonal group in two dimensions, denoted by SO(2). POEM aims to reduce the phase-related variability by learning a phase-correction coordinate and applying an invertible SO(2)-based rotation to paired latent features. To extrapolate this correction coordinate, Directional Phase Increment Attention (DPIA) retrieves historical phase increments from similar temporal contexts and integrates them into future phase corrections. Experiments demonstrate that POEM achieves competitive performance, while qualitative visualizations suggest that the learned phase-aware transformation makes latent trajectories more regular.

G-triplexes are noncanonical DNA structures formed by three guanine tracts, but their sequence design rules for small-molecule binding remain unclear. In this work, we carried out a systematic sequence screening to optimize G-triplex sequences for improved interaction with methylene blue (MB), and applied the best sequence in an electrochemical biosensor for melamine detection. Over 80 candidate sequences were designed by varying loop nucleotides, 5'-flanking bases, and strand length, and were evaluated by square wave voltammetry. The optimal sequence F9 (5'-ATGGGAGGGTGGG-3') achieved the highest current suppression (ΔI/I0 ≈ 0.81), outperforming all tested G-triplex and several well-known G-quadruplex sequences. Circular dichroism confirmed that F9 folds into a parallel G-triplex with a melting temperature of 65.1 °C, which increased to 70.0 °C upon MB binding. Electrochemical titration gave a 1:1 binding ratio with an association constant of 9.37 × 105 M-1. Molecular dynamics simulations showed that MB binds to the 5'-face of the G-triplex by π-π stacking, with a computed binding free energy of -151.9 kJ/mol. A hairpin probe (MelaPin) combining the F9 sequence with a poly-T melamine recognition region was then constructed. The resulting biosensor showed a linear response over 1-100 μM melamine with a detection limit of 0.74 μM, and a recovery of 96.0 ± 0.8% in spiked infant formula. This work provides a practical framework for G-triplex sequence design and shows that the G-triplex/MB system can serve as a simple, label-free signal module for electrochemical sensing.

Sign Language Translation (SLT) is a challenging cross-modal task requiring joint modeling of manual articulations and non-manual signals. Existing gloss-free SLT methods effectively capture gestural dynamics but often underutilize facial expressions, which play crucial grammatical and disambiguating roles. This limitation can cause semantic degradation when distinct concepts share similar manual configurations. To address this issue, we propose FEA-SLT (**F**acial-**E**xpression-**A**ware **S**ign **L**anguage **T**ranslation), a gloss-free end-to-end framework that uses facial dynamics as semantic anchors for resolving manual ambiguity. FEA-SLT employs a domain-transferred facial encoder to extract expression-sensitive representations and integrates them with manual features through a linguistically constrained *Facial-Expression-Aware Fusion* (FEAF) module. FEAF captures reciprocal dependencies between manual and facial channels via bidirectional modulation, enhancing syntactic fidelity. Experiments on PHOENIX14T and CSL-Daily show that FEA-SLT achieves state-of-the-art BLEU performance among gloss-free methods, while targeted analyses confirm improved translation of facial-sensitive utterances. Code is available at [https://github.com/TuGuobin/FEA-SLT](https://github.com/TuGuobin/FEA-SLT).

Large language models are increasingly deployed as tool-augmented agents to acquire information beyond parametric knowledge. While recent work has improved long-horizon tool-use reasoning, most approaches focus on tasks with a single correct answer. In contrast, many real-world queries require discovering a comprehensive set of valid answers, a setting known as Multi-Answer QA. This setting raises two challenges: fine-grained credit assignment over long search trajectories and reward alignment for sustained exploration beyond easy high-frequency entities. We propose SPADER, a reinforcement learning framework for long-horizon tool use in Multi-Answer QA. SPADER includes Step-wise Peer Advantage (SPA), a critic-free step-level credit assignment mechanism that aligns parallel trajectories by decision step and estimates advantages from peer returns. It also includes a diversity-aware exploration reward that promotes long-tail entity discovery by upweighting rare findings and downweighting redundant ones. Experiments on QAMPARI, Mintaka, WebQSP, and QUEST show that SPADER generally improves recall and overall F1 over prompting-based agents, outcome-supervised RL methods, and recent step-level supervision approaches. Our code and model weights are available at https://github.com/KhanCold/spader.

Publishing time series datasets raises substantial privacy concerns, as the underlying patterns (e.g., trends, values) can lead to the disclosure of individual identification. Mitigating these concerns remains challenging due to difficulties in pinpointing specific privacy-leaking patterns and protecting them without significantly compromising the analytical utility of the published data. Existing methods remain vulnerable to identity attacks utilizing diverse temporal patterns and may compromise data utility for subsequent analytical tasks. To address these limitations, we collaborated with domain experts to summarize a taxonomy of privacy risks in time series data and developed TSEditor, an interactive editing system. TSEditor integrates coordinated views for multi-perspective analysis of privacy risks and introduces six editing operations for targeted modifications, providing visual feedback. We demonstrate the effectiveness and usability of TSEditor through two case studies, an expert interview, a model evaluation, and a user study.

Data validation is a crucial step in data analytics workflows that assesses and ensures the reliability of data flowing into analytical processes. One common approach to data validation involves defining validation rules, which provide explicit constraints and conditions that data must satisfy. However, creating accurate and effective validation rules remains challenging for many practitioners. This challenge stems from the need for practitioners to understand both data structures and their domain-specific semantic relationships. Recent studies have proposed automated approaches to generate validation rules by deriving patterns from data properties. However, these approaches generate rules with limited interpretability and lack support for rule verification and modification, making the rules difficult to understand and adapt. To address these limitations in current validation rule authoring approaches, we present RuleScope, an interactive system for authoring data validation rules through semantic-aware rule generation, visualization, and refinement. RuleScope employs an LLM-based workflow to generate interpretable rules by analyzing data semantics and incorporating domain knowledge. To facilitate rule comprehension, we design a matrix-based visualization that helps users understand rules and analyze validation results. Additionally, RuleScope enables users to interactively refine rules. We evaluate the LLM-based workflow through model evaluation on datasets from different domains and assess RuleScope's usability and effectiveness through two case studies and a user study.

Non-stationary time series forecasting is challenged by evolving distribution shifts that static models struggle to capture. While Mixture-of-Experts (MoE) architectures offer a promising paradigm for decoupling complex drift patterns, existing approaches are limited by fixed expert pools and memoryless routing, hampering their ability to adapt to abrupt regime shifts. To address this, we propose , a framework that unifies architectural evolution with temporal continuity during learning phase. By detecting distribution shifts via Maximum Mean Discrepancy (MMD), we dynamically instantiate heterogeneous experts and prune redundant ones to optimize capacity. Additionally, a temporal memory router leverages recurrent states and an anomaly repository to ensure stable, context-aware expert selection without requiring test-time updates. Experiments on nine benchmarks demonstrate state-of-the-art performance, reducing MSE by 10.4\% and MAE by 7.8\%. Code is available at https://anonymous.4open.science/r/Dynamic-TMoE.

Root cause analysis (RCA) is critical for maintaining the availability and efficiency of cloud computing systems. However, identifying root causes from the large-scale, high-dimensional monitoring data generated by these complex environments is a significant challenge. Current approaches often rely on time-consuming manual analysis to ensure flexibility and reliability, while recent automated methods lack the crucial insights provided by domain experts. To bridge this gap, we propose RCInvestigator, a visual analytics system that facilitates interactive root cause investigation by establishing a tight collaboration between human experts and machine analysis. Our approach addresses three key challenges: a) modeling databases for the root cause investigation, b) inferring root causes from large-scale time series, and c) building comprehensible investigation results. We demonstrate the effectiveness and utility of RCInvestigator through two real-world case studies, which received positive feedback from domain experts.

Large Language Models (LLMs) demonstrate exceptional capabilities in factual question answering, yet they sometimes provide incorrect responses. To address this issue, knowledge editing techniques have emerged as effective methods for correcting factual information in LLMs. However, typical knowledge editing workflows struggle with identifying the optimal set of model layers for editing and rely on summary indicators that provide insufficient guidance. This lack of transparency hinders effective comparison and identification of optimal editing strategies. In this paper, we present KEditVis, a novel visual analytics system designed to assist users in gaining a deeper understanding of knowledge editing through interactive visualizations, improving editing outcomes, and discovering valuable insights for the future development of knowledge editing algorithms. With KEditVis, users can select appropriate layers as the editing target, explore the reasons behind ineffective edits, and perform more targeted and effective edits. Our evaluation, including usage scenarios, expert interviews, and a user study, validates the effectiveness and usability of the system.

Skill-based prompting has become a practical mechanism for improving large language model (LLM) agents, yet existing skill acquisition methods often treat skills as experience summaries, memory entries, or direct summaries of successful demonstrations. This creates a mismatch for weaker student agents: when a student fails because it lacks task knowledge or operational strategy, its failed trajectory may not contain enough evidence to infer the missing behavior, while the teacher trajectory may be too implicit to be internalized as reusable guidance. We propose SKILL-KD, a contrastive skill distillation framework that treats skills as an explicit distillation medium between agents of different capabilities. Given a student failure and the teacher trajectory on the same task, SKILL-KD distills their actionable discrepancy into a textual skill patch, evaluates the patch by re-running the student, and iteratively refines the patch when the student still fails. To prevent repeated local updates from causing skill drift, SKILL-KD further maintains trace-linked edit histories and performs Drift-Aware Skill Consolidation, deciding whether each patch should add a new rule, delete or modify an existing rule, or be skipped. Across five agent benchmarks and two student settings, SKILL-KD consistently improves frozen student agents over fixed-model adaptation baselines.

Large language model agents are increasingly evaluated as autonomous tool users, yet most benchmarks focus on bounded tasks with immediate success criteria. Real-world deployments often require Long-Term Coherence, the capacity to preserve purposeful behavior across extended horizons while adapting decisions to accumulated evidence. Evaluating this capacity requires a persistent environment in which actions constrain future choices, feedback arrives at heterogeneous delays, and incoherent behavior produces measurable cumulative effects. Seller-side e-commerce provides a suitable setting for this evaluation through recurrent and interdependent decisions over Product Sourcing, Listing and Pricing Control, Cash-Flow Management, and Mixed-Latency Feedback Adaptation. We introduce MerchantBench, a 365-day order-level simulation grounded in 98,843 real e-commerce product records and equipped with 26 tools for agent interaction. MerchantBench couples promptly observable Upstream Supplier Events with delayed Downstream Order Outcomes, requiring agents to follow individual order lifecycles and revisit earlier decisions. We evaluate eight LLMs under two agent frameworks in 48 runs, each spanning 365 simulated days. Our results reveal a substantial gap between even the latest LLMs and human participants, with the best LLM configuration attaining only 27.3% of the mean final net assets achieved by human participants.

Trajectory visualizations play a crucial role in urban computing, supporting tasks from analyzing road networks and vehicle movements to informing transport policy. However, creating such visualizations faces usability and flexibility limitations in trajectory visualization scenarios: limited usability due to insufficient trajectory-specific specialization, and limited flexibility due to tightly-coupled interactions that are not treated as first-class components. To address these challenges, we first conduct a systematic review of prior work and define a comprehensive design space for interactive trajectory visualization across three key dimensions: transformation, display, and interaction. We then present TrajGram, a declarative grammar built upon this design space that streamlines the creation of diverse interactive trajectory visualizations. TrajGram enables rapid prototyping with rich interactions grounded in real-world traffic analysis needs, while remaining extensible for complex scenarios. We demonstrate the effectiveness, usability, and learning curve of TrajGram through two usage scenarios and a user study.

Data profiling plays a critical role in understanding the structure of complex datasets and supporting numerous downstream tasks, such as social media analytics and financial fraud detection. While existing research predominantly focuses on structured data formats, a substantial portion of semi-structured textual data still requires ad-hoc and arduous manual profiling to extract and comprehend its internal structures. In this work, we propose StructVizor, an interactive profiling system that facilitates sensemaking and transformation of semi-structured textual data. Our tool mainly addresses two challenges: a) extracting and visualizing the diverse structural patterns within data, such as how information is organized or related, and b) enabling users to efficiently perform various wrangling operations on textual data. Through automatic data parsing and structure mining, StructVizor enables visual analytics of structural patterns, while incorporating novel interactions to enable profile-based data wrangling. A comparative user study involving 12 participants demonstrates the system's usability and its effectiveness in supporting exploratory data analysis and transformation tasks.

Recent advances in video generation have enabled the synthesis of videos with strong temporal consistency and impressive visual quality, marking a crucial step toward vision foundation models. To evaluate these video generation models, existing benchmarks primarily focus on factors related to visual perception and understanding, like visual aesthetics, instruction adherence, and temporal coherence. However, the rule-based reasoning capabilities of video generation models remain largely unexplored. Although recent studies have carried out preliminary explorations into whether video models can serve as zero-shot learners, they still lack a fine-grained decomposition of reasoning capabilities and a comprehensive evaluation protocol. To address this gap, we introduce RULER-Bench, a benchmark designed to evaluate the reasoning ability of video generation models from the perspective of cognitive rules. Built upon two fundamental paradigms: text-to-video and image-to-video, RULER-Bench covers 40 representative tasks spanning six rule categories with 622 high-quality annotated instances. For the evaluation of each generated video, we construct a checklist covering four metrics and leverage GPT-o3 to assign scores to each question, achieving 85

Unregistered illegal facilities that do not qualify for chemical production pose substantial threats to human lives and the environment. For human safety and environmental protection, the government needs to figure out the illegal facilities and shut them down. A new, convenient, and affordable approach to detect such facilities is to analyze the trajectories of hazardous chemicals transportation (HCT) trucks. The existing study leverages a machine learning model to predict how likely a place is illegal. However, such a model lacks interpretability and cannot provide actionable justifications required for decision-making. In this study, we collaborate with HCT experts and propose an interactive visual analytics approach to explore the suspicious stay points, analyze abnormal HCT truck behaviors, and figure out unregistered illegal chemical facilities. First, experts receive an initial result from the detection model for reference. Then, they are supported to check the detailed information of the suspicious places with three coordinated views. We apply a visualization that tightly encodes the geo-referred movement activities along the timeline to present the HCT truck behaviors, which can help experts finally verify their conclusions. We demonstrate the effectiveness of the system with two case studies on real-world data. We also received experts’ positive feedback from an expert interview.

Efficient and accurate detection of anomalies in photovoltaic (PV) strings is essential for ensuring the normal operation of PV power stations. Most existing studies focus on developing automated anomaly detection models based on temporal abnormalities in PV strings. However, since analyzing anomalies often requires domain knowledge, existing automated methods have significant limitations in assisting experts to understand the causes and impact of these anomalies. In close collaboration with domain experts, this work has summarized the specific user requirements for PV string anomaly detection and designed PVeSight, an interactive visual analysis system to help experts discover and analyze anomalies in PV strings. We use dimensionality reduction techniques to generate string pattern map. These maps are used for anomaly detection, classifying anomalies, comparative analysis between strings, and hierarchical analysis under inverters and combiner boxes. This helps experts trace the causes of anomalies and acquire valuable insights into anomalous PV strings. Through case studies and expert evaluation, we verified the usability and effectiveness of PVeSight for PV string anomaly detection.