arXiv subjects

Jiannan Wang

Publications and source records attributed to Jiannan Wang.

At least 19 recordsLinked to original sources

arXiv2026

Throughput Analysis for Near-Field Mobile Communications: Beamfocusing or Caustic Beamforming?

The migration to the Terahertz (THz) band and the deployment of extremely large antenna arrays (ELAAs) are transitioning wireless communications into the radiative near-field regime, fundamentally evolving conventional angular beam steering to beamfocusing (BF). However, the combination of the extremely narrow beamwidth and the mobility of the users necessitates frequent beamfocusing reconfigurations, incurring a significant switching overhead that degrades the system achievable throughput. In this regard, caustic beamforming (CB) is a promising alternative based on the synthesis of a continuous curved beam, which eliminates the need for beam tracking at the expense of a distributed beamforming gain. By leveraging the Airy beam as a canonical model, this paper develops an analytical framework to compare the throughputs achieved by CB and BF. Our main results include closed-form throughput expressions for both beamforming strategies and a performance boundary for paradigm selection. First, we derive the BF throughput by modeling a defocusing penalty induced by continuous user movement. The optimal beam dwell time that maximizes the throughput is analytically determined, and the impact of user speed and switching overhead on the throughput is quantified. For the CB scheme, we demonstrate that its throughput is determined by the signal-to-noise ratio (SNR) and the geometry of the trajectory of the user, yet invariant to the user speed. Finally, we analytically establish a threshold for the switching overhead to define the crossover point of the achievable throughput of both beamformers. Crucially, this threshold asymptotically vanishes at extremely high frequencies, positioning the continuous CB scheme as the preferred beam design paradigm for high-mobility THz communications.

eess.SP↗

arXiv2026

The Case for Text-to-SQL Friendly Logical Database Design

Logical database design has traditionally optimized database schemas, including tables, columns, keys, constraints, and views, for correctness, integrity, and human-written application queries. LLM-based Text-to-SQL changes the consumer: the schema is now often read as text by a language model, so design choices that preserve database semantics can still change SQL-generation accuracy. We argue that this creates a new design objective alongside the classical ones - LLM-friendly logical database design, the property that a schema is easy for a language model to map from natural language to correct SQL - and treat it as the optimization target of this paper. We instantiate this objective with three semantics-preserving schema transformations that re-purpose classical schema-design ideas: schema abstraction (+A: logical views that materialize recurring join paths), schema partitioning (+P: workload-aware logical partitions that prune irrelevant context), and schema renaming (+R: descriptive identifiers that improve downstream column linking and predicate construction). The three operators compose, and each preserves the underlying database semantics. When historical question-SQL pairs are available, they guide both partitioning and abstraction; in zero-shot settings, renaming applies directly, and abstraction falls back to an ad-hoc per-question variant. We evaluate the resulting schemas on BIRD-Union and Spider-Union across multiple Text-to-SQL pipelines and language model backbones, with gains of up to 4.2% in execution accuracy. The best transformation varies modestly across pipelines and models, with the full +A+P+R consistently improving; multiple operator combinations are competitive on each pipeline. These results show that LLM-friendly logical design is a practical and underexplored database-side optimization target, complementary to existing Text-to-SQL pipelines.

cs.DB↗

arXiv2026

PrepBench: How Far Are We from Natural-Language-Driven Data Preparation?

Data preparation is a central and time-consuming stage in data analysis workflows. Traditionally, commercial tools have relied on graphical user interfaces (GUIs) to simplify data preparation, allowing users to define transformations through visual operators and workflows. Recent advances in large language models (LLMs) raise the possibility of a paradigm shift toward natural language (NL)-driven data preparation, in which users can specify preparation intents in NL directly. However, it remains unclear how far current LLM-based agents are from this paradigm shift in practice. Existing code generation benchmarks do not capture key characteristics of data preparation, including ambiguous user intents, imperfect real-world data, and the need to translate code into interpretable workflows for validation. To bridge this gap, we present PrepBench, a benchmark designed to evaluate NL-driven data preparation along three core capabilities: interactive disambiguation, prep-code generation, and code-to-workflow translation. We crawl data from the Preppin' Data Challenges, and then extend it into a systematically designed benchmark. The benchmark covers diverse domains, and each task involves 3 to 18 data preparation steps. Nearly half of the tasks require over 100 lines of Python code, and the longest solutions approach 300 lines. Our evaluation shows that, despite recent progress, realizing this paradigm shift remains challenging for state-of-the-art LLMs. PrepBench provides a principled benchmark for measuring this gap and helps identify key challenges toward realizing NL-driven data preparation.

cs.DB↗

arXiv2026

HFX: Joint Design of Algorithms and Systems for Multi-SLO Serving and Fast Scaling

Large language model (LLM) serving faces the dual challenge of meeting strict user-specific service-level objectives (SLOs) while minimizing computational cost under dynamic, multi-task workloads. Existing approaches either rely on static scheduling policies or focus on single-task settings, limiting their applicability in real-world deployments with heterogeneous requests, variable prompt lengths, and elastic scaling requirements. We present HFX, a production LLM serving system that jointly optimizes request scheduling and elastic scaling across model replicas to satisfy diverse SLOs. HFX introduces a \textbf{scheduler} that performs proactive budget estimation and prioritization to ensure SLO compliance for both new and in-flight requests. HFX also integrates a \textbf{scaler} that supports fast device-to-device (D2D) weight transfer, reducing cold-start latency. Additionally, the system supports both colocated and disaggregated prefill/decode deployments, enabling adaptation to diverse workload patterns and cloud environments. Through extensive experiments on multi-task workloads, we demonstrate consistently higher SLO attainment, lower end-to-end latency, and lower NPU usage cost by up to 4.44$\times$, 65.82\%, and 49.81\%, respectively, compared to state-of-the-art systems. Our results highlight the effectiveness of SLO-aware scheduling and scaling in practical LLM serving, providing a robust framework for cost-efficient and SLO-compliant deployments.

cs.DC↗

arXiv2026

Auto-FP: An Experimental Study of Automated Feature Preprocessing for Tabular Data

Classical machine learning models, such as linear models and tree-based models, are widely used in industry. These models are sensitive to data distribution, thus feature preprocessing, which transforms features from one distribution to another, is a crucial step to ensure good model quality. Manually constructing a feature preprocessing pipeline is challenging because data scientists need to make difficult decisions about which preprocessors to select and in which order to compose them. In this paper, we study how to automate feature preprocessing (Auto-FP) for tabular data. Due to the large search space, a brute-force solution is prohibitively expensive. To address this challenge, we interestingly observe that Auto-FP can be modelled as either a hyperparameter optimization (HPO) or a neural architecture search (NAS) problem. This observation enables us to extend a variety of HPO and NAS algorithms to solve the Auto-FP problem. We conduct a comprehensive evaluation and analysis of 15 algorithms on 45 public ML datasets. Overall, evolution-based algorithms show the leading average ranking. Surprisingly, the random search turns out to be a strong baseline. Many surrogate-model-based and bandit-based search algorithms, which achieve good performance for HPO and NAS, do not outperform random search for Auto-FP. We analyze the reasons for our findings and conduct a bottleneck analysis to identify the opportunities to improve these algorithms. Furthermore, we explore how to extend Auto-FP to support parameter search and compare two ways to achieve this goal. In the end, we evaluate Auto-FP in an AutoML context and discuss the limitations of popular AutoML tools. To the best of our knowledge, this is the first study on automated feature preprocessing. We hope our work can inspire researchers to develop new algorithms tailored for Auto-FP.

cs.LG↗

arXiv2026

Energy-Efficient Velocity Profile Optimization for Movable Antenna-Enabled Sensing Systems

Movable antennas (MAs) enable the reconfiguration of array geometry within a bounded region to exploit sub-wavelength spatial degrees of freedom in wireless communication and sensing systems. However, most prior research has predominantly focused on the communication and sensing performance, overlooking the mechanical power consumption inherent in antenna movement. To bridge this gap, this paper investigates a velocity profile optimization framework for MA-assisted direction-of-arrival (DoA) estimation, explicitly balancing sensing accuracy with mechanical energy consumption of MAs. We first establish a Newtonian-based mechanical energy model, and formulate a functional optimization problem for sensing energy efficiency (EE) maximization. By applying the calculus of variations, this formulation is transformed into an infinite-dimensional problem defined by the Euler-Lagrange equation. To solve it, we propose a spectral discretization framework based on the Galerkin method, which expands the velocity profile over a sinusoidal basis. In the regime where energy consumption is dominated by linear damping, we prove that the optimal velocity profile follows a closed-form sinusoidal shape. For more general scenarios involving strong nonlinear aerodynamic drag, we leverage the Markov-Lukács theorem to transform the kinematic constraints into strictly convex sum-of-squares (SOS) conditions. Consequently, the infinite-dimensional problem is reformulated as a tractable finite-dimensional nonlinear algebraic system, which is solved by a two-layer algorithm combining Dinkelbach's method with successive convex approximation (SCA). Numerical results demonstrate that our optimized velocity profile significantly outperforms baselines in terms of EE across various system configurations. Insights into the optimized velocity profiles and practical design guidelines are also provided.

eess.SP↗

arXiv2026

Boosting Text-to-Chart Retrieval through Training with Synthesized Semantic Insights

Text-to-chart retrieval, enabling users to find relevant charts via natural language queries, has gained significant attention. However, evaluating models in real-world business intelligence (BI) scenarios is challenging, as current benchmarks fail to simulate realistic user queries or test for deep semantic understanding with static chart images.To address this gap, we introduce CRBench, the first real-world BI-sourced benchmark comprising 21,862 charts and 326 queries, utilizing a Target-and-Distractor paradigm to evaluate discriminative retrieval among highly similar candidates. Testing on CRBench reveals that existing methods, which rely primarily on visual features, perform poorly and fail to capture the rich analytical semantics of charts. To address this performance bottleneck, we propose a semantic insights synthesis pipeline that automatically generates three hierarchical levels of insights for charts: visual patterns, statistical properties, and practical applications. Using this pipeline, we produced 207,498 semantic insights for 69,166 charts as training data. By leveraging this data to bridge the gap between natural language query intent and latent visual representations via multi-level semantic supervision, we develop ChartFinder, a specialized model capable of deep cross-model reasoning. Experimental results show ChartFinder significantly outperforms state-of-the-art methods on CRBench, achieving up to 66.9% NDCG@10 for precise queries (an 11.58% improvement) and an average increase of 5% across nearly all metrics for fuzzy queries. This work provides the community with a much-needed benchmark for realistic evaluation and demonstrates a powerful data synthesis paradigm for enhancing a model's semantic understanding of charts.

cs.IR↗

arXiv2026

A Survey of Data Agents: Emerging Paradigm or Overstated Hype?

The rapid advancement of large language models (LLMs) has spurred the emergence of data agents, autonomous systems designed to orchestrate Data + AI ecosystems for tackling complex data-related tasks. However, the term "data agent" currently suffers from terminological ambiguity and inconsistent adoption, conflating simple query responders with sophisticated autonomous architectures. This terminological ambiguity fosters mismatched user expectations, accountability challenges, and barriers to industry growth. Inspired by the SAE J3016 standard for driving automation, this survey introduces the first systematic hierarchical taxonomy for data agents, comprising six levels that delineate and trace progressive shifts in autonomy, from manual operations (L0) to a vision of generative, fully autonomous data agents (L5), thereby clarifying capability boundaries and responsibility allocation. Through this lens, we offer a structured review of existing research arranged by increasing autonomy, encompassing specialized data agents for data management, preparation, and analysis, alongside emerging efforts toward versatile, comprehensive systems with enhanced autonomy. We further analyze critical evolutionary leaps and technical gaps for advancing data agents, especially the ongoing L2-to-L3 transition, where data agents evolve from procedural execution to autonomous orchestration. Finally, we conclude with a forward-looking roadmap, envisioning proactive, generative data agents.

cs.DB↗

arXiv2026

ReviveMoE: Fast Recovery for Hardware Failures in Large-Scale MoE LLM Inference Deployments

As LLM deployments scale over more hardware, the probability of a single failure in a system increases significantly, and cloud operators must consider robust countermeasures to handle these inevitable failures. A common recovery approach is to simply restart the LLM serving instance; however, this is costly in model-as-a-service (MaaS) inference settings, where reloading model weights and recompiling computation graphs can introduce significant delays to incoming requests. We propose ReviveMoE, a method for rapid failure recovery in large-scale LLM deployments without restarting the serving instance. ReviveMoE is designed to support both the traditional LLM architecture, which collocates MoE and attention on the same hardware, and the disaggregated architectures, which separate MoE from attention. Integrated into Huawei Cloud's MaaS, ReviveMoE is built on top of Huawei's xDeepServe serving platform and the XCCL communications library.

cs.DC↗

arXiv2026

ST-Raptor: An Agentic System for Semi-Structured Table QA

Semi-structured table question answering (QA) is a challenging task that requires (1) precise extraction of cell contents and positions and (2) accurate recovery of key implicit logical structures, hierarchical relationships, and semantic associations encoded in table layouts. In practice, such tables are often interpreted manually by human experts, which is labor-intensive and time-consuming. However, automating this process remains difficult. Existing Text-to-SQL methods typically require converting semi-structured tables into structured formats, inevitably leading to information loss, while approaches like Text-to-Code and multimodal LLM-based QA struggle with complex layouts and often yield inaccurate answers. To address these limitations, we present ST-Raptor, an agentic system for semi-structured table QA. ST-Raptor offers an interactive analysis environment that combines visual editing, tree-based structural modeling, and agent-driven query resolution to support accurate and user-friendly table understanding. Experimental results on both benchmark and real-world datasets demonstrate that ST-Raptor outperforms existing methods in both accuracy and usability. The code is available at https://github.com/weAIDB/ST-Raptor, and a demonstration video is available at https://youtu.be/9GDR-94Cau4.

cs.AI↗

arXiv2026

CoRe: Benchmarking LLMs Code Reasoning Capabilities through Static Analysis Tasks

Large language models (LLMs) have been widely adopted across diverse domains of software engineering, such as code generation, program repair, and vulnerability detection. These applications require understanding beyond surface-level code patterns: value propagation, control flow, and interdependence between program elements. However, existing benchmarks primarily evaluate end-to-end outcomes, such as whether code is correctly repaired or generated, leaving the models' ability for program semantic reasoning underexplored. This work presents CORE, a high-quality, human-verified benchmark designed to evaluate LLMs on fundamental static analysis tasks. CORE includes 12,553 task instances spanning data dependency, control dependency, and information flow across programs written in C/C++, Java, and Python. To ensure semantic diversity and reasoning complexity, we propose a semantics-aware diverse sampling strategy that selects targets and task instances based on structural coverage and dependency depth. We evaluate 10 mainstream LLMs and show that, while they perform well at identifying dependencies, models still struggle with tasks that require deeper semantic understanding and multi-step reasoning. We further conduct qualitative analyses to uncover key challenges, such as complex control structures and backward dependency patterns, offering insights into improving LLMs' code reasoning capabilities.

cs.SE↗

arXiv2025

PipeFusion: Patch-level Pipeline Parallelism for Diffusion Transformers Inference

This paper presents PipeFusion, an innovative parallel methodology to tackle the high latency issues associated with generating high-resolution images using diffusion transformers (DiTs) models. PipeFusion partitions images into patches and the model layers across multiple GPUs. It employs a patch-level pipeline parallel strategy to orchestrate communication and computation efficiently. By capitalizing on the high similarity between inputs from successive diffusion steps, PipeFusion reuses one-step stale feature maps to provide context for the current pipeline step. This approach notably reduces communication costs compared to existing DiTs inference parallelism, including tensor parallel, sequence parallel and DistriFusion. PipeFusion enhances memory efficiency through parameter distribution across devices, ideal for large DiTs like Flux.1. Experimental results demonstrate that PipeFusion achieves state-of-the-art performance on 8$\times$L40 PCIe GPUs for Pixart, Stable-Diffusion 3, and Flux.1 models. Our source code is available at https://github.com/xdit-project/xDiT.

cs.CV↗

arXiv2025

ChartCards: A Chart-Metadata Generation Framework for Multi-Task Chart Understanding

The emergence of Multi-modal Large Language Models (MLLMs) presents new opportunities for chart understanding. However, due to the fine-grained nature of these tasks, applying MLLMs typically requires large, high-quality datasets for task-specific fine-tuning, leading to high data collection and training costs. To address this, we propose ChartCards, a unified chart-metadata generation framework for multi-task chart understanding. ChartCards systematically synthesizes various chart information, including data tables, visualization code, visual elements, and multi-dimensional semantic captions. By structuring this information into organized metadata, ChartCards enables a single chart to support multiple downstream tasks, such as text-to-chart retrieval, chart summarization, chart-to-table conversion, chart description, and chart question answering. Using ChartCards, we further construct MetaChart, a large-scale high-quality dataset containing 10,862 data tables, 85K charts, and 170 K high-quality chart captions. We validate the dataset through qualitative crowdsourcing evaluations and quantitative fine-tuning experiments across various chart understanding tasks. Fine-tuning six different models on MetaChart resulted in an average performance improvement of 5% across all tasks. The most notable improvements are seen in text-to-chart retrieval and chart-to-table tasks, with Long-CLIP and Llama 3.2-11B achieving improvements of 17% and 28%, respectively.

cs.CL↗

arXiv2025

ST-Raptor: LLM-Powered Semi-Structured Table Question Answering

Semi-structured tables, widely used in real-world applications (e.g., financial reports, medical records, transactional orders), often involve flexible and complex layouts (e.g., hierarchical headers and merged cells). These tables generally rely on human analysts to interpret table layouts and answer relevant natural language questions, which is costly and inefficient. To automate the procedure, existing methods face significant challenges. First, methods like NL2SQL require converting semi-structured tables into structured ones, which often causes substantial information loss. Second, methods like NL2Code and multi-modal LLM QA struggle to understand the complex layouts of semi-structured tables and cannot accurately answer corresponding questions. To this end, we propose ST-Raptor, a tree-based framework for semi-structured table question answering using large language models. First, we introduce the Hierarchical Orthogonal Tree (HO-Tree), a structural model that captures complex semi-structured table layouts, along with an effective algorithm for constructing the tree. Second, we define a set of basic tree operations to guide LLMs in executing common QA tasks. Given a user question, ST-Raptor decomposes it into simpler sub-questions, generates corresponding tree operation pipelines, and conducts operation-table alignment for accurate pipeline execution. Third, we incorporate a two-stage verification mechanism: forward validation checks the correctness of execution steps, while backward validation evaluates answer reliability by reconstructing queries from predicted answers. To benchmark the performance, we present SSTQA, a dataset of 764 questions over 102 real-world semi-structured tables. Experiments show that ST-Raptor outperforms nine baselines by up to 20% in answer accuracy. The code is available at https://github.com/weAIDB/ST-Raptor.

cs.AI↗

arXiv2025

CleanAgent: Automating Data Standardization with LLM-based Agents

Data standardization is a crucial part of the data science life cycle. While tools like Pandas offer robust functionalities, their complexity and the manual effort required for customizing code to diverse column types pose significant challenges. Although large language models (LLMs) like ChatGPT have shown promise in automating this process through natural language understanding and code generation, it still demands expert-level programming knowledge and continuous interaction for prompt refinement. To solve these challenges, our key idea is to propose a Python library with declarative, unified APIs for standardizing different column types, simplifying the LLM's code generation with concise API calls. We first propose Dataprep.Clean, a component of the Dataprep Python Library, significantly reduces the coding complexity by enabling the standardization of specific column types with a single line of code. Then, we introduce the CleanAgent framework integrating Dataprep.Clean and LLM-based agents to automate the data standardization process. With CleanAgent, data scientists only need to provide their requirements once, allowing for a hands-free process. To demonstrate the practical utility of CleanAgent, we developed a user-friendly web application, allowing users to interact with it using real-world datasets.

cs.LG↗

arXiv2025

LINEAGEX: A Column Lineage Extraction System for SQL

As enterprise data grows in size and complexity, column-level data lineage, which records the creation, transformation, and reference of each column in the warehouse, has been the key to effective data governance that assists tasks like data quality monitoring, storage refactoring, and workflow migration. Unfortunately, existing systems introduce overheads by integration with query execution or fail to achieve satisfying accuracy for column lineage. In this paper, we demonstrate LINEAGEX, a lightweight Python library that infers column level lineage from SQL queries and visualizes it through an interactive interface. LINEAGEX achieves high coverage and accuracy for column lineage extraction by intelligently traversing query parse trees and handling ambiguities. The demonstration walks through use cases of building lineage graphs and troubleshooting data quality issues. LINEAGEX is open sourced at https://github.com/sfu-db/lineagex and our video demonstration is at https://youtu.be/5LaBBDDitlw

cs.DB↗

arXiv2025

A Unified Model for Cardinality Estimation by Learning from Data and Queries via Sum-Product Networks

Cardinality estimation is a fundamental component in database systems, crucial for generating efficient execution plans. Despite advancements in learning-based cardinality estimation, existing methods may struggle to simultaneously optimize the key criteria: estimation accuracy, inference time, and storage overhead, limiting their practical applicability in real-world database environments. This paper introduces QSPN, a unified model that integrates both data distribution and query workload. QSPN achieves high estimation accuracy by modeling data distribution using the simple yet effective Sum-Product Network (SPN) structure. To ensure low inference time and reduce storage overhead, QSPN further partitions columns based on query access patterns. We formalize QSPN as a tree-based structure that extends SPNs by introducing two new node types: QProduct and QSplit. This paper studies the research challenges of developing efficient algorithms for the offline construction and online computation of QSPN. We conduct extensive experiments to evaluate QSPN in both single-table and multi-table cardinality estimation settings. The experimental results have demonstrated that QSPN achieves superior and robust performance on the three key criteria, compared with state-of-the-art approaches.

cs.DB↗

arXiv2025

BQSched: A Non-intrusive Scheduler for Batch Concurrent Queries via Reinforcement Learning

Most large enterprises build predefined data pipelines and execute them periodically to process operational data using SQL queries for various tasks. A key issue in minimizing the overall makespan of these pipelines is the efficient scheduling of concurrent queries within the pipelines. Existing tools mainly rely on simple heuristic rules due to the difficulty of expressing the complex features and mutual influences of queries. The latest reinforcement learning (RL) based methods have the potential to capture these patterns from feedback, but it is non-trivial to apply them directly due to the large scheduling space, high sampling cost, and poor sample utilization. Motivated by these challenges, we propose BQSched, a non-intrusive Scheduler for Batch concurrent Queries via reinforcement learning. Specifically, BQSched designs an attention-based state representation to capture the complex query patterns, and proposes IQ-PPO, an auxiliary task-enhanced proximal policy optimization (PPO) algorithm, to fully exploit the rich signals of Individual Query completion in logs. Based on the RL framework above, BQSched further introduces three optimization strategies, including adaptive masking to prune the action space, scheduling gain-based query clustering to deal with large query sets, and an incremental simulator to reduce sampling cost. To our knowledge, BQSched is the first non-intrusive batch query scheduler via RL. Extensive experiments show that BQSched can significantly improve the efficiency and stability of batch query scheduling, while also achieving remarkable scalability and adaptability in both data and queries. For example, across all DBMSs and scales tested, BQSched reduces the overall makespan of batch queries on TPC-DS benchmark by an average of 34% and 13%, compared with the commonly used heuristic strategy and the adapted RL-based scheduler, respectively.

cs.DB↗