arXiv subjects

Xiang Qi

Publications and source records attributed to Xiang Qi.

16 recordsLinked to original sources

arXiv2026

LongDocBench: Benchmarking TOC Hierarchy and Contextual Relationship Recovery in Long Documents

Parsing visual documents into machine-readable representations is fundamental to document intelligence. Existing benchmarks focus on page-level element recognition, reading order, formula recognition, and table structure. Long documents, however, also require document-level structure recovery. This includes reconstructing cross-page table-of-contents (TOC) hierarchies and identifying typed links from tables and figures to their captions, notes, and sources, often in one-to-many form. Because these structures are covered only partially or subsumed within broader parsing protocols, existing benchmarks cannot directly evaluate two key document-level tasks: \emph{Table-of-Contents Hierarchy Recovery} and \emph{Contextual Relationship Recovery}. To benchmark these two tasks, we introduce \textsc{LongDocBench}, comprising 85 real-world financial reports, textbooks, and academic papers spanning 2,582 pages, with up to 105 pages per document. It provides human-verified annotations for 3,937 heading nodes (mean node depth 3.55; maximum depth 9) and 3,258 contextual relationships annotated across 2,680 table and figure objects. We further evaluate both the downstream utility and recoverability of these structures. Long-document question-answering experiments show that human-verified TOC hierarchies and contextual relationships improve reasoning, with their combination providing complementary benefits. Meanwhile, representative document parsers remain limited on both recovery tasks despite strong page-level performance. To support further progress, we publicly release \textsc{LongDocBench} and its evaluation protocol and reproducible testbed for advancing document-level structure recovery in long documents.

cs.AI↗

arXiv2026

Curvature of Physical Space and Geometric Projection: A Possible Geometric Understanding of Quantum Mechanics and Relativity

Starting from a formal comparison between the Schr\"{o}dinger equation and the Jacobi field equation, this paper proposes the hypothesis that, within the framework of non-relativistic quantum mechanics, the sectional curvature of microscopic physical space can be expressed as $K(x) = 2m(E-V(x))/\hbar^2$, and attempts to offer a possible unified geometric understanding of quantum mechanics and relativity. The paper is divided into three parts. The first part identifies the Schr\"{o}dinger equation, in its mathematical form, as a Jacobi field equation. The wave function is made to correspond to a Jacobi field, giving it the geometric meaning of the deviation of geodesics in physical space. Phenomena such as quantum tunneling, energy quantization, and the path integral are then interpreted, from this perspective, as geometric manifestations of different spatial curvatures. The second part proposes a geometric projection scheme based on semi-geodesic coordinates. The transverse metric factor of a curved space, which satisfies a Jacobi field equation, is taken as a unified geometric framework. It is demonstrated that the metric correction terms in classical mechanics, special relativity, general relativity, and quantum mechanics can all be derived from the same equation, with the curvature $K$ arising from different physical sources. The third part extends this framework to statistical mechanics.

physics.gen-ph↗

arXiv2026

AgenticDataBench: A Comprehensive Benchmark for Data Agents

Data science aims to derive actionable insights from heterogeneous raw data, unlocking the value of the massive amounts of data generated in modern society. Automating this process is essential to reducing labor-intensive efforts for data scientists and enabling scalable data-driven applications. Recently, large language model (LLM)-based data agents have emerged as a promising solution to automate data science workflows. However, the field lacks comprehensive benchmarks to rigorously evaluate these agents across diverse scenarios with fine-grained granularity. To address this gap, we propose AgenticDataBench, a comprehensive benchmark featuring realistic tasks spanning diverse domains with fine-grained ground-truth labels. This enables evaluations to capture the diversity and complexity of data science workflows and the detailed performance of agents. First, to cover diverse domains, we collect real datasets and tasks from 15 vertical domains, including 5 real-world B2B use cases from a leading fintech company. Second, to remove redundancy in real-world tasks and generate high-quality tasks for domains lacking real data, we introduce data science skills, recurring data-centric operational patterns, and quantify benchmark coverage by the number of skills included. Representative skills are extracted from large-scale task solutions on Stack Overflow using skill-aligned hierarchical clustering. Third, for real-world business tasks, we select task-solution pairs that maximize diversity in skill composition, ensuring broad coverage of practical scenarios. Fourth, to generate realistic tasks for devise domains without real tasks, we propose a systematic LLM-based task generation approach to create workflows and tasks based on these skills. Finally, we evaluate state-of-the-art data agents using our annotated benchmark and open-sourced testbed, providing detailed skill-level insights.

cs.DB↗

arXiv2026

SkillAdaptor: Self-Adapting Skills for LLM Agents from Trajectories

Large language model (LLM) agents increasingly rely on reusable external skills to solve long-horizon interactive tasks. Existing training-free skill adaptation pipelines usually update skills from full trajectories or session-level feedback, which makes failure attribution coarse and often produces unstable or overly broad revisions. We propose SkillAdaptor, a training-free step-level skill adaptation framework with explicit failure attribution, and it can plug into OpenClaw-class agent harnesses. Given a failed trajectory, SkillAdaptor identifies a first actionable fault step, links responsibility to candidate skills, and applies targeted updates under explicit acceptance checks while keeping the backbone frozen. We evaluate on WebShop, PinchBench, and Claw-Eval with Kimi-K2.5, GLM-5, and GPT-5.2. SkillAdaptor improves over no-skill and skill-adaptation baselines on all three suites, with the largest single-metric improvements of +1.5 points on PinchBench Avg Score%, +1.8 on Claw-Eval Avg Score, and +1.7 on WebShop success rate. These results indicate that step-level attribution supports more stable and auditable training-free skill maintenance\footnote{The code will be released at https://github.com/zjunlp/SkillAdaptor.}.

cs.CL↗

arXiv2026

Every Step Counts: Step-Level Credit Assignment for Tool-Integrated Text-to-SQL

Tool-integrated Text-to-SQL parsing has emerged as a promising paradigm, framing SQL generation as a sequential decision-making process interleaved with tool execution. However, existing reinforcement learning approaches mainly rely on coarse-grained outcome supervision, resulting in a fundamental credit assignment problem: models receive the same reward for any trajectory that yields the correct answer, even when intermediate steps are redundant, inefficient, or erroneous. Consequently, models are encouraged to explore suboptimal reasoning spaces, limiting both efficiency and generalization. To address this problem, we propose FineStep, a novel framework for step-level credit assignment in tool-augmented Text-to-SQL. First, we introduce a reward design with independent process rewards to alleviate the signal sparsity of outcome supervision. Next, we present a step-level credit assignment mechanism to precisely quantify the value of each reasoning step. Finally, we develop a policy optimization method based on step-level advantages for efficient updates. Extensive experiments on BIRD benchmarks show that FineStep achieves state-of-the-art performance and reduces redundant tool interactions, with a 3.25% average EX gain over GRPO at the 4B scale.

cs.CL↗

arXiv2026

SGA-MCTS: Decoupling Planning from Execution via Training-Free Atomic Experience Retrieval

LLM-powered systems require complex multi-step decision-making abilities to solve real-world tasks, yet current planning approaches face a trade-off between the high latency of inference-time search and the limited generalization of supervised fine-tuning. To address this limitation, we introduce \textbf{SGA-MCTS}, a framework that casts LLM planning as non-parametric retrieval. Offline, we leverage Monte Carlo Tree Search (MCTS) to explore the solution space and distill high-fidelity trajectories into State-Goal-Action (SGA) atoms. These atoms are de-lexicalized primitives that abstract concrete entities into symbolic slots, preserving reusable causal logic while discarding domain-specific noise. Online, a retrieval-augmented agent employs a hybrid symbolic-semantic mechanism to fetch relevant SGAs and re-ground them into the current context as soft reasoning hints. Empirical results on complex benchmarks demonstrate that this paradigm enables frozen, open-weights models to match the performance of SOTA systems (e.g., GPT-5) without task-specific fine-tuning. By effectively amortizing the heavy computational cost of search, SGA-MCTS achieves System 2 reasoning depth at System 1 inference speeds, rendering autonomous planning both scalable and real-time feasible.

cs.AI↗

arXiv2026

SkillX: Automatically Constructing Skill Knowledge Bases for Agents

Learning from experience is critical for building capable large language model (LLM) agents, yet prevailing self-evolving paradigms remain inefficient: agents learn in isolation, repeatedly rediscover similar behaviors from limited experience, resulting in redundant exploration and poor generalization. To address this problem, we propose SkillX, a fully automated framework for constructing a \textbf{plug-and-play skill knowledge base} that can be reused across agents and environments. SkillX operates through a fully automated pipeline built on three synergistic innovations: \textit{(i) Multi-Level Skills Design}, which distills raw trajectories into three-tiered hierarchy of strategic plans, functional skills, and atomic skills; \textit{(ii) Iterative Skills Refinement}, which automatically revises skills based on execution feedback to continuously improve library quality; and \textit{(iii) Exploratory Skills Expansion}, which proactively generates and validates novel skills to expand coverage beyond seed training data. Using a strong backbone agent (GLM-4.6), we automatically build a reusable skill library and evaluate its transferability on challenging long-horizon, user-interactive benchmarks, including AppWorld, BFCL-v3, and $\tau^2$-Bench. Experiments show that SkillKB consistently improves task success and execution efficiency when plugged into weaker base agents, highlighting the importance of structured, hierarchical experience representations for generalizable agent learning. Our code will be publicly available soon at https://github.com/zjunlp/SkillX.

cs.CL↗

arXiv2026

Elder-Sim: A Psychometrically Validated Platform for Personality-Stable Elderly Digital Twins

Background: LLMs enable patient-facing conversational agents, creating a pathway toward digital twins that capture older adults' lived experiences and behavioral responses across time. A central barrier is personality drift -- inconsistent trait expression across repeated interactions -- which undermines reliability of generated trajectories and intervention-response simulation in geriatric care. Objective: To develop ELDER-SIM, a multi-role elderly-care conversational platform for building personality-stable digital twin agents, and to propose a psychometric validation framework for quantifying personality consistency in LLM-based agents. Methods: ELDER-SIM was implemented via n8n workflow orchestration with local LLM inference (Ollama/vLLM), integrating (1) Big Five (OCEAN) trait specifications, (2) a Cognitive Conceptualization Diagram (CCD) grounded in Beck's CBT framework, and (3) a MySQL-based long-term memory module. Ablation studies across four conditions -- Baseline, +Memory, +CCD, and +LoRA (fine-tuned on 19,717 instruction pairs from CHARLS) -- were evaluated via Cronbach's $\alpha$, ICC, and role discrimination accuracy. Results: Reliability was acceptable to excellent across conditions (Cronbach's $\alpha$: 0.70--0.94; ICC: 0.85--0.96). Role discrimination improved from 83.3% (Baseline) to 88.9% (+Memory), 94.4% (+CCD), and 97.2% (+LoRA). CCD produced the largest consistency gain (mean $\alpha$ 0.702$\to$0.892), while LoRA achieved the highest overall consistency ($\alpha$ 0.940; ICC 0.958). Conclusions: ELDER-SIM provides a psychometrically validated approach for constructing personality-consistent elderly digital twin agents. Structured cognitive modeling and domain adaptation reduce personality drift, supporting reliable longitudinal simulation for elderly mental health care and reproducible in silico evaluation before clinical deployment.

cs.HC↗

arXiv2025

Agentar-Scale-SQL: Advancing Text-to-SQL through Orchestrated Test-Time Scaling

State-of-the-art (SOTA) Text-to-SQL methods still lag significantly behind human experts on challenging benchmarks like BIRD. Current approaches that explore test-time scaling lack an orchestrated strategy and neglect the model's internal reasoning process. To bridge this gap, we introduce Agentar-Scale-SQL, a novel framework leveraging scalable computation to improve performance. Agentar-Scale-SQL implements an Orchestrated Test-Time Scaling strategy that synergistically combines three distinct perspectives: i) Internal Scaling via RL-enhanced Intrinsic Reasoning, ii) Sequential Scaling through Iterative Refinement, and iii) Parallel Scaling using Diverse Synthesis and Tournament Selection. Agentar-Scale-SQL is a general-purpose framework designed for easy adaptation to new databases and more powerful language models. Extensive experiments show that Agentar-Scale-SQL achieves SOTA performance on the BIRD benchmark, reaching 81.67% execution accuracy on the test set and ranking first on the official leaderboard, demonstrating an effective path toward human-level performance.

cs.CL↗

arXiv2025

Agentar-Fin-R1: Enhancing Financial Intelligence through Domain Expertise, Training Efficiency, and Advanced Reasoning

Large Language Models (LLMs) exhibit considerable promise in financial applications; however, prevailing models frequently demonstrate limitations when confronted with scenarios that necessitate sophisticated reasoning capabilities, stringent trustworthiness criteria, and efficient adaptation to domain-specific requirements. We introduce the Agentar-Fin-R1 series of financial large language models (8B and 32B parameters), specifically engineered based on the Qwen3 foundation model to enhance reasoning capabilities, reliability, and domain specialization for financial applications. Our optimization approach integrates a high-quality, systematic financial task label system with a comprehensive multi-layered trustworthiness assurance framework. This framework encompasses high-quality trustworthy knowledge engineering, multi-agent trustworthy data synthesis, and rigorous data validation governance. Through label-guided automated difficulty-aware optimization, tow-stage training pipeline, and dynamic attribution systems, we achieve substantial improvements in training efficiency. Our models undergo comprehensive evaluation on mainstream financial benchmarks including Fineva, FinEval, and FinanceIQ, as well as general reasoning datasets such as MATH-500 and GPQA-diamond. To thoroughly assess real-world deployment capabilities, we innovatively propose the Finova evaluation benchmark, which focuses on agent-level financial reasoning and compliance verification. Experimental results demonstrate that Agentar-Fin-R1 not only achieves state-of-the-art performance on financial tasks but also exhibits exceptional general reasoning capabilities, validating its effectiveness as a trustworthy solution for high-stakes financial applications. The Finova bench is available at https://github.com/antgroup/Finova.

cs.CL↗

arXiv2025

Agentar-DeepFinance-100K: A Large-Scale Financial Dataset via Systematic Chain-of-Thought Synthesis Optimization

Recent advancements in large language models (LLMs) have demonstrated remarkable general reasoning capabilities, holding significant potential for applications in the financial domain, a field that requires robust and reliable reasoning. It has been demonstrated that distilling high-quality chain-of-thought (CoT) rationales from advanced general reasoning models offers a promising and efficient path to the financial reasoning model. However, existing CoT synthesis methods suffer from shallow CoT sampling, leaving the question of how to construct a well-designed knowledge space for finance reasoning unexplored. In this paper, we present Agentar-DeepFinance-100K, a large-scale financial reasoning dataset characterized by its systematic CoT synthesis optimization. We first introduce a comprehensive CoT synthesis pipeline featuring Multi-perspective Knowledge Extraction (MKE) and Self-Corrective Rewriting (SCR) to generate exhaustive and deep financial reasoning trajectories. Furthermore, a systematic investigation, termed CoT Cube, is conducted to analyze critical factors that influence CoT effectiveness, such as necessity, length and synthesizer, yielding valuable insights for high-quality financial CoT construction. Experiments demonstrate that models trained on our Agentar-DeepFinance-100K achieve significant improvements on financial benchmarks. We publicly release Agentar-DeepFinance-100K , hoping to advance the research in financial reasoning models.

cs.CE↗

arXiv2025

Natural van der Waals canalization lens for non-destructive nanoelectronic circuit imaging and inspection

Optical inspection has long served as a cornerstone non-destructive method in semiconductor wafer manufacturing, particularly for surface and defect analysis. However, conventional techniques such as bright-field and dark-field scattering optics face significant limitations, including insufficient resolution and the inability to penetrate and detect buried structures. Atomic force microscopy (AFM), while offering higher resolution and precise surface characterization, is constrained by slow speed, limited to surface-level imaging, and incapable of resolving subsurface features. Here, we propose an approach that integrates the strengths of dark-field scattering optics and AFM by leveraging a van der Waals (vdW) canalization lens based on natural biaxial {\alpha}-MoO3 crystals. This method enables ultrahigh-resolution subwavelength imaging with the ability to visualize both surface and buried structures, achieving a spatial resolution of 15 nm and grating pitch detection down to 100 nm. The underlying mechanism relies on the unique anisotropic properties of {\alpha}-MoO3, where its atomic-scale unit cells and biaxial symmetry facilitate the diffraction-free propagation of both evanescent and propagating waves via a flat-band canalization regime. Unlike metamaterial-based superlenses and hyperlenses, which suffer from high plasmonic losses, fabrication imperfections, and uniaxial constraints, {\alpha}-MoO3 provides robust and aberration-free imaging in multiple directions. We successfully applied this approach to high-resolution inspection of buried nanoscale electronic circuits, offering unprecedented capabilities essential for next-generation semiconductor manufacturing.

physics.optics↗

arXiv2024

BoolQuestions: Does Dense Retrieval Understand Boolean Logic in Language?

Dense retrieval, which aims to encode the semantic information of arbitrary text into dense vector representations or embeddings, has emerged as an effective and efficient paradigm for text retrieval, consequently becoming an essential component in various natural language processing systems. These systems typically focus on optimizing the embedding space by attending to the relevance of text pairs, while overlooking the Boolean logic inherent in language, which may not be captured by current training objectives. In this work, we first investigate whether current retrieval systems can comprehend the Boolean logic implied in language. To answer this question, we formulate the task of Boolean Dense Retrieval and collect a benchmark dataset, BoolQuestions, which covers complex queries containing basic Boolean logic and corresponding annotated passages. Through extensive experimental results on the proposed task and benchmark dataset, we draw the conclusion that current dense retrieval systems do not fully understand Boolean logic in language, and there is a long way to go to improve our dense retrieval systems. Furthermore, to promote further research on enhancing the understanding of Boolean logic for language models, we explore Boolean operation on decomposed query and propose a contrastive continual training method that serves as a strong baseline for the research community.

cs.IR↗

arXiv2024

Trustworthy Alignment of Retrieval-Augmented Large Language Models via Reinforcement Learning

Trustworthiness is an essential prerequisite for the real-world application of large language models. In this paper, we focus on the trustworthiness of language models with respect to retrieval augmentation. Despite being supported with external evidence, retrieval-augmented generation still suffers from hallucinations, one primary cause of which is the conflict between contextual and parametric knowledge. We deem that retrieval-augmented language models have the inherent capabilities of supplying response according to both contextual and parametric knowledge. Inspired by aligning language models with human preference, we take the first step towards aligning retrieval-augmented language models to a status where it responds relying merely on the external evidence and disregards the interference of parametric knowledge. Specifically, we propose a reinforcement learning based algorithm Trustworthy-Alignment, theoretically and experimentally demonstrating large language models' capability of reaching a trustworthy status without explicit supervision on how to respond. Our work highlights the potential of large language models on exploring its intrinsic abilities by its own and expands the application scenarios of alignment from fulfilling human preference to creating trustworthy agents.

cs.CL↗

arXiv2023

Awayvirus: A Playful and Tangible Approach to Improve Children's Hygiene Habits in Family Education

Despite various playful and educational tools have been developed to support children's learning abilities, limited work focuses on tangible toys designed to improve and maintain children's hygiene perception, habits and awareness, as well as fostering their collaboration and social abilities in home education contexts. We developed \textbf{Awayvirus} to address this research and design gap, aiming to help children gain hygiene habits knowledge through tangible blocks. Our findings indicate that a playful tangible interaction method can effectively increase children's interest in learning and encourage parents to become actively involved in their children's hygiene and health education. Additionally, Awayvirus seeks to build a collaborative bridge between children and parents, promoting communication strategies while mitigating the adverse effects of the challenging the post-pandemic period.

cs.HC↗

arXiv2020

Revisiting Regex Generation for Modeling Industrial Applications by Incorporating Byte Pair Encoder

Regular expression is important for many natural language processing tasks especially when used to deal with unstructured and semi-structured data. This work focuses on automatically generating regular expressions and proposes a novel genetic algorithm to deal with this problem. Different from the methods which generate regular expressions from character level, we first utilize byte pair encoder (BPE) to extract some frequent items, which are then used to construct regular expressions. The fitness function of our genetic algorithm contains multi objectives and is solved based on evolutionary procedure including crossover and mutation operation. In the fitness function, we take the length of generated regular expression, the maximum matching characters and samples for positive training samples, and the minimum matching characters and samples for negative training samples into consideration. In addition, to accelerate the training process, we do exponential decay on the population size of the genetic algorithm. Our method together with a strong baseline is tested on 13 kinds of challenging datasets. The results demonstrate the effectiveness of our method, which outperforms the baseline on 10 kinds of data and achieves nearly 50 percent improvement on average. By doing exponential decay, the training speed is approximately 100 times faster than the methods without using exponential decay. In summary, our method possesses both effectiveness and efficiency, and can be implemented for the industry application.

cs.NE↗