arXiv subjects

Yixuan Cao

Publications and source records attributed to Yixuan Cao.

17 recordsLinked to original sources

arXiv2026

Navigating Large-Scale Document Collections: MuDABench for Multi-Document Analytical QA

This paper introduces the task of analytical question answering over large, semi-structured document collections. We present MuDABench, a benchmark for multi-document analytical QA, where questions require extracting and synthesizing information across numerous documents to perform quantitative analysis. Unlike existing multi-document QA benchmarks that typically require information from only a few documents with limited cross-document reasoning, MuDABench demands extensive inter-document analysis and aggregation. Constructed via distant supervision by leveraging document-level metadata and annotated financial databases, MuDABench comprises over 80,000 pages and 332 analytical QA instances. We also propose an evaluation protocol that measures final answer accuracy and uses intermediate-fact coverage as an auxiliary diagnostic signal for the reasoning process. Experiments reveal that standard RAG systems, which treat all documents as a flat retrieval pool, perform poorly. To address these limitations, we propose a multi-agent workflow that orchestrates planning, extraction, and code generation modules. While this approach substantially improves both process and outcome metrics, a significant gap remains compared to human expert performance. Our analysis identifies two primary bottlenecks: single-document information extraction accuracy and insufficient domain-specific knowledge in current systems. MuDABench is available at https://github.com/Zhanli-Li/MuDABench.

cs.CL↗

arXiv2026

DeepRead: Document Structure-Aware Reasoning to Enhance Agentic Search

With the rapid advancement of tool-use capabilities in Large Language Models (LLMs), Retrieval-Augmented Generation (RAG) is shifting from static, one-shot retrieval toward autonomous, multi-turn evidence acquisition. However, existing agentic search frameworks typically treat long documents as flat collections of unstructured chunks, disregarding the native hierarchical organization and sequential logic essential for human comprehension. To bridge this gap, we introduce \textbf{DeepRead}, a structure-aware document reasoning agent designed to operationalize document-native structural priors into actionable reasoning capabilities. Leveraging the structural fidelity of modern OCR, DeepRead constructs a paragraph-level, coordinate-based navigation system and equips the LLM with two synergistic tools: \textsf{Retrieve} for scanning-aware localization, and \textsf{ReadSection} for contiguous, order-preserving reading within specific hierarchical scopes. This design elicits a human-like ``locate-then-read'' reasoning paradigm, effectively mitigating the context fragmentation inherent in traditional retrieval methods. Extensive evaluations across four benchmarks spanning diverse document types demonstrate that DeepRead outperforms Search-o1-style agentic search baselines by an average of 10.3\%. Fine-grained behavioral analysis further confirms that DeepRead autonomously adopts human-aligned reading strategies, validating the critical role of structural awareness in achieving precise document reasoning. Our code is available at https://github.com/Zhanli-Li/DeepRead.

cs.AI↗

arXiv2025

The Gray Zone of Faithfulness: Taming Ambiguity in Unfaithfulness Detection

Ensuring that Large Language Models (LLMs) generate summaries faithful to a given source document is essential for real-world applications. While prior research has explored LLM faithfulness, existing benchmarks suffer from annotation ambiguity, primarily due to the ill-defined boundary of permissible external knowledge in generated outputs. For instance, common sense is often incorporated into responses and labeled as "faithful", yet the acceptable extent of such knowledge remains unspecified, leading to inconsistent annotations. To address this issue, we propose a novel faithfulness annotation framework, which introduces an intermediate category, Out-Dependent, to classify cases where external knowledge is required for verification. Using this framework, we construct VeriGray (Verification with the Gray Zone) -- a new unfaithfulness detection benchmark in summarization. Statistics reveal that even SOTA LLMs, such as GPT-5, exhibit hallucinations ($\sim 6\%$ of sentences) in summarization tasks. Moreover, a substantial proportion ($\sim 9\%$ on average of models) of generated sentences fall into the Out-Dependent category, underscoring the importance of resolving annotation ambiguity in unfaithfulness detection benchmarks. Experiments demonstrate that our benchmark poses significant challenges to multiple baseline methods, indicating considerable room for future improvement.

cs.CL↗

arXiv2025

Efficient Candidate-Free R-S Set Similarity Joins with Filter-and-Verification Trees on MapReduce

Given two different collections of sets R and S, the exact R-S set similarity join (R-S Join) finds all set pairs with similarity no less than a given threshold, which has widespread applications. Existing algorithms accelerate large-scale R-S Joins using a two-stage filter-and-verification framework along with the parallel and distributed MapReduce framework, however, they suffer from excessive candidate set pairs (candidates), leading to significant I/O and verification overhead. This paper proposes novel candidate-free R-S Join (CF-RS-Join) algorithms that integrate filtering and verification into a single stage through the filter-and-verification tree (FVT) and its linear variant (LFVT). First, CF-RS-Join with FVT (CF-RS-Join/FVT) is proposed to leverage an innovative FVT structure that compresses elements and associated sets in memory, enabling single-stage processing that eliminates candidate generation, enables fast lookups, and reduces database scans. Correctness proofs are provided. Second, CF-RS-Join with LFVT (CF-RS-Join/LFVT) is proposed to exploit a more compact Linear FVT, which compresses non-branching paths into single nodes and stores them in linear arrays for optimized traversal. Third, MR-CF-RS-Join/FVT and MR-CF-RS-Join/LFVT are proposed to extend our approaches using MapReduce for parallel processing. Extensive experiments have been conducted on the proposed algorithms against state-of-the-art (SOTA) baselines in terms of execution time, scalability, memory usage, and disk usage. The results show that MR-CF-RS-Join/LFVT outperforms the runner-up by up to 1.37x-15.78x on 7 real-world datasets.

cs.DC↗

arXiv2025

AutoFSM: A Multi-agent Framework for FSM Code Generation with IR and SystemC-Based Testing

With the rapid advancement of large language models (LLMs) in code generation, their applications in hardware design are receiving growing attention. However, existing LLMs face several challenges when generating Verilog code for finite state machine (FSM) control logic, including frequent syntax errors, low debugging efficiency, and heavy reliance on test benchmarks. To address these challenges, this paper proposes AutoFSM, a multi-agent collaborative framework designed for FSM code generation tasks. AutoFSM introduces a structurally clear intermediate representation (IR) to reduce syntax error rate during code generation and provides a supporting toolchain to enable automatic translation from IR to Verilog. Furthermore, AutoFSM is the first to integrate SystemC-based modeling with automatic testbench generation, thereby improving debugging efficiency and feedback quality. To systematically evaluate the framework's performance, we construct SKT-FSM, the first hierarchical FSM benchmark in the field, comprising 67 FSM samples across different complexity levels. Experimental results show that, under the same base LLM, AutoFSM consistently outperforms the open-source framework MAGE on the SKT-FSM benchmark, achieving up to an 11.94% improvement in pass rate and up to a 17.62% reduction in syntax error rate. These results demonstrate the potential of combining LLMs with structured IR and automated testing to improve the reliability and scalability of register-transfer level (RTL) code generation.

cs.SE↗

arXiv2025

Read the Docs Before Rewriting: Equip Rewriter with Domain Knowledge via Continual Pre-training

A Retrieval-Augmented Generation (RAG)-based question-answering (QA) system enhances a large language model's knowledge by retrieving relevant documents based on user queries. Discrepancies between user queries and document phrasings often necessitate query rewriting. However, in specialized domains, the rewriter model may struggle due to limited domain-specific knowledge. To resolve this, we propose the R\&R (Read the doc before Rewriting) rewriter, which involves continual pre-training on professional documents, akin to how students prepare for open-book exams by reviewing textbooks. Additionally, it can be combined with supervised fine-tuning for improved results. Experiments on multiple datasets demonstrate that R\&R excels in professional QA across multiple domains, effectively bridging the query-document gap, while maintaining good performance in general scenarios, thus advancing the application of RAG-based QA systems in specialized fields.

cs.IR↗

arXiv2025

DETree: DEtecting Human-AI Collaborative Texts via Tree-Structured Hierarchical Representation Learning

Detecting AI-involved text is essential for combating misinformation, plagiarism, and academic misconduct. However, AI text generation includes diverse collaborative processes (AI-written text edited by humans, human-written text edited by AI, and AI-generated text refined by other AI), where various or even new LLMs could be involved. Texts generated through these varied processes exhibit complex characteristics, presenting significant challenges for detection. Current methods model these processes rather crudely, primarily employing binary classification (purely human vs. AI-involved) or multi-classification (treating human-AI collaboration as a new class). We observe that representations of texts generated through different processes exhibit inherent clustering relationships. Therefore, we propose DETree, a novel approach that models the relationships among different processes as a Hierarchical Affinity Tree structure, and introduces a specialized loss function that aligns text representations with this tree. To facilitate this learning, we developed RealBench, a comprehensive benchmark dataset that automatically incorporates a wide spectrum of hybrid texts produced through various human-AI collaboration processes. Our method improves performance in hybrid text detection tasks and significantly enhances robustness and generalization in out-of-distribution scenarios, particularly in few-shot learning conditions, further demonstrating the promise of training-based approaches in OOD settings. Our code and dataset are available at https://github.com/heyongxin233/DETree.

cs.CL↗

arXiv2025

Reasoning Pattern Matters: Learning to Reason without Human Rationales

Large Language Models (LLMs) have demonstrated remarkable reasoning capabilities under the widely adopted SFT+RLVR paradigm, which first performs Supervised Fine-Tuning (SFT) on human-annotated reasoning trajectories (rationales) to establish initial reasoning behaviors, then applies Reinforcement Learning with Verifiable Rewards (RLVR) to optimize the model using verifiable signals without golden rationales. However, annotating high-quality rationales for the SFT stage remains prohibitively expensive. This paper investigates when and how rationale annotation costs can be substantially reduced without compromising reasoning performance. We identify a broad class of problems, termed patterned reasoning tasks, where reasoning follows a fixed, procedural strategy consistent across instances. Although instances vary in content such as domain knowledge, factual information, or numeric values, the solution derives from applying a shared reasoning pattern. We argue that the success of SFT+RLVR on such tasks primarily stems from its ability to enable models to internalize these reasoning patterns. Using numerical semantic matching as a representative task, we provide both causal and behavioral evidence showing that reasoning patterns rather than the quantity or quality of rationales are the key determinant of performance. Building on these insights, we propose Pattern-Aware LLMs as Rationale AnnOtators (PARO), a simple yet effective framework that enables LLMs to generate rationales aligned with task-specific reasoning patterns without requiring human rationale annotations. Experiments show that PARO-generated rationales achieve comparable SFT+RLVR performance to human rationales that are 10 times larger. These results suggest that large-scale human rationale annotations can be replaced with LLM-based automatic annotations requiring only limited human supervision over reasoning patterns.

cs.CL↗

arXiv2025

AttnComp: Attention-Guided Adaptive Context Compression for Retrieval-Augmented Generation

Retrieval-augmented generation improves the factual accuracy of Large Language Models (LLMs) by incorporating external context, but often suffers from irrelevant retrieved content that hinders effectiveness. Context compression addresses this issue by filtering out irrelevant information from context before LLM generation. However, existing methods struggle to adaptively adjust compression rates for different context, maintain low latency and integrate information across multiple documents. To overcome these limitations, We introduce AttnComp, an adaptive, efficient and context-aware compression framework. By leveraging the attention mechanism of LLMs to identify relevant information, AttnComp employs a Top-P compression algorithm to retain the minimal set of documents whose cumulative attention weights exceeds a predefined threshold. In addition to compression, AttnComp estimates response confidence by assessing the overall relevance of the retrieved content, enabling users to gauge response reliability. Experiments demonstrate that AttnComp outperforms existing compression methods and uncompressed baselines, achieving higher accuracy with substantial compression rates and lower latency.

cs.CL↗

arXiv2025

ORCAS: Obfuscation-Resilient Binary Code Similarity Analysis using Dominance Enhanced Semantic Graph

Binary code similarity analysis (BCSA) serves as a foundational technique for binary analysis tasks such as vulnerability detection and malware identification. Existing graph based BCSA approaches capture more binary code semantics and demonstrate remarkable performance. However, when code obfuscation is applied, the unstable control flow structure degrades their performance. To address this issue, we develop ORCAS, an Obfuscation-Resilient BCSA model based on Dominance Enhanced Semantic Graph (DESG). The DESG is an original binary code representation, capturing more binaries' implicit semantics without control flow structure, including inter-instruction relations (e.g., def-use), inter-basic block relations (i.e., dominance and post-dominance), and instruction-basic block relations. ORCAS takes binary functions from different obfuscation options, optimization levels, and instruction set architectures as input and scores their semantic similarity more robustly. Extensive experiments have been conducted on ORCAS against eight baseline approaches over the BinKit dataset. For example, ORCAS achieves an average 12.1% PR-AUC improvement when using combined three obfuscation options compared to the state-of-the-art approaches. In addition, an original obfuscated real-world vulnerability dataset has been constructed and released to facilitate a more comprehensive research on obfuscated binary code analysis. ORCAS outperforms the state-of-the-art approaches over this newly released real-world vulnerability dataset by up to a recall improvement of 43%.

cs.CR↗

arXiv2025

Tech-ASan: Two-stage check for Address Sanitizer

Address Sanitizer (ASan) is a sharp weapon for detecting memory safety violations, including temporal and spatial errors hidden in C/C++ programs during execution. However, ASan incurs significant runtime overhead, which limits its efficiency in testing large software. The overhead mainly comes from sanitizer checks due to the frequent and expensive shadow memory access. Over the past decade, many methods have been developed to speed up ASan by eliminating and accelerating sanitizer checks, however, they either fail to adequately eliminate redundant checks or compromise detection capabilities. To address this issue, this paper presents Tech-ASan, a two-stage check based technique to accelerate ASan with safety assurance. First, we propose a novel two-stage check algorithm for ASan, which leverages magic value comparison to reduce most of the costly shadow memory accesses. Second, we design an efficient optimizer to eliminate redundant checks, which integrates a novel algorithm for removing checks in loops. Third, we implement Tech-ASan as a memory safety tool based on the LLVM compiler infrastructure. Our evaluation using the SPEC CPU2006 benchmark shows that Tech-ASan outperforms the state-of-the-art methods with 33.70% and 17.89% less runtime overhead than ASan and ASan--, respectively. Moreover, Tech-ASan detects 56 fewer false negative cases than ASan and ASan-- when testing on the Juliet Test Suite under the same redzone setting.

cs.SE↗

arXiv2025

Towards Efficient Quantity Retrieval from Text:An Approach via Description Parsing and Weak Supervision

Quantitative facts are continually generated by companies and governments, supporting data-driven decision-making. While common facts are structured, many long-tail quantitative facts remain buried in unstructured documents, making them difficult to access. We propose the task of Quantity Retrieval: given a description of a quantitative fact, the system returns the relevant value and supporting evidence. Understanding quantity semantics in context is essential for this task. We introduce a framework based on description parsing that converts text into structured (description, quantity) pairs for effective retrieval. To improve learning, we construct a large paraphrase dataset using weak supervision based on quantity co-occurrence. We evaluate our approach on a large corpus of financial annual reports and a newly annotated quantity description dataset. Our method significantly improves top-1 retrieval accuracy from 30.98 percent to 64.66 percent.

cs.IR↗

arXiv2025

Document-Level Tabular Numerical Cross-Checking: A Coarse-to-Fine Approach

Numerical consistency across tables in disclosure documents is critical for ensuring accuracy, maintaining credibility, and avoiding reputational and economic risks. Automated tabular numerical cross-checking presents two significant challenges: (C1) managing the combinatorial explosion of candidate instances at the document level and (C2) comprehending multi-faceted numerical semantics. Previous research typically depends on heuristic-based filtering or simplified context extraction, often struggling to balance performance and efficiency. Recently, large language models (LLMs) have demonstrated remarkable contextual understanding capabilities that helps address C2 at the instance level, yet they remain hampered by computational inefficiency (C1) and limited domain expertise. This paper introduces CoFiTCheck, a novel LLM-based coarse-to-fine framework that addresses these challenges through two sequential stages: embedding-based filtering and discriminative classification. The embedding-based filtering stage introduces an instructional parallel encoding method to efficiently represent all numerical mentions in a table with LLMs, as well as a decoupled InfoNCE objective to mitigate the isolated mention problem. The discriminative classification stage employs a specialized LLM for fine-grained analysis of the remaining candidate pairs. This stage is further enhanced by our crosstable numerical alignment pretraining paradigm, which leverages weak supervision from cross-table numerical equality relationships to enrich task-specific priors without requiring manual annotation. Comprehensive evaluation across three types of real-world disclosure documents demonstrates that CoFiTCheck significantly outperforms previous methods while maintaining practical efficiency.

cs.CL↗

arXiv2024

Attention with Dependency Parsing Augmentation for Fine-Grained Attribution

To assist humans in efficiently validating RAG-generated content, developing a fine-grained attribution mechanism that provides supporting evidence from retrieved documents for every answer span is essential. Existing fine-grained attribution methods rely on model-internal similarity metrics between responses and documents, such as saliency scores and hidden state similarity. However, these approaches suffer from either high computational complexity or coarse-grained representations. Additionally, a common problem shared by the previous works is their reliance on decoder-only Transformers, limiting their ability to incorporate contextual information after the target span. To address the above problems, we propose two techniques applicable to all model-internals-based methods. First, we aggregate token-wise evidence through set union operations, preserving the granularity of representations. Second, we enhance the attributor by integrating dependency parsing to enrich the semantic completeness of target spans. For practical implementation, our approach employs attention weights as the similarity metric. Experimental results demonstrate that the proposed method consistently outperforms all prior works.

cs.CL↗

arXiv2024

Uncovering Limitations of Large Language Models in Information Seeking from Tables

Tables are recognized for their high information density and widespread usage, serving as essential sources of information. Seeking information from tables (TIS) is a crucial capability for Large Language Models (LLMs), serving as the foundation of knowledge-based Q&A systems. However, this field presently suffers from an absence of thorough and reliable evaluation. This paper introduces a more reliable benchmark for Table Information Seeking (TabIS). To avoid the unreliable evaluation caused by text similarity-based metrics, TabIS adopts a single-choice question format (with two options per question) instead of a text generation format. We establish an effective pipeline for generating options, ensuring their difficulty and quality. Experiments conducted on 12 LLMs reveal that while the performance of GPT-4-turbo is marginally satisfactory, both other proprietary and open-source models perform inadequately. Further analysis shows that LLMs exhibit a poor understanding of table structures, and struggle to balance between TIS performance and robustness against pseudo-relevant tables (common in retrieval-augmented systems). These findings uncover the limitations and potential challenges of LLMs in seeking information from tables. We release our data and code to facilitate further research in this field.

cs.CL↗

arXiv2023

Guideline Learning for In-context Information Extraction

Large language models (LLMs) can perform a new task by merely conditioning on task instructions and a few input-output examples, without optimizing any parameters. This is called In-Context Learning (ICL). In-context Information Extraction (IE) has recently garnered attention in the research community. However, the performance of In-context IE generally lags behind the state-of-the-art supervised expert models. We highlight a key reason for this shortfall: underspecified task description. The limited-length context struggles to thoroughly express the intricate IE task instructions and various edge cases, leading to misalignment in task comprehension with humans. In this paper, we propose a Guideline Learning (GL) framework for In-context IE which reflectively learns and follows guidelines. During the learning phrase, GL automatically synthesizes a set of guidelines based on a few error cases, and during inference, GL retrieves helpful guidelines for better ICL. Moreover, we propose a self-consistency-based active learning method to enhance the efficiency of GL. Experiments on event extraction and relation extraction show that GL can significantly improve the performance of in-context IE.

cs.CL↗

arXiv2021

Extracting Variable-Depth Logical Document Hierarchy from Long Documents: Method, Evaluation, and Application

In this paper, we study the problem of extracting variable-depth "logical document hierarchy" from long documents, namely organizing the recognized "physical document objects" into hierarchical structures. The discovery of logical document hierarchy is the vital step to support many downstream applications. However, long documents, containing hundreds or even thousands of pages and variable-depth hierarchy, challenge the existing methods. To address these challenges, we develop a framework, namely Hierarchy Extraction from Long Document (HELD), where we "sequentially" insert each physical object at the proper on of the current tree. Determining whether each possible position is proper or not can be formulated as a binary classification problem. To further improve its effectiveness and efficiency, we study the design variants in HELD, including traversal orders of the insertion positions, heading extraction explicitly or implicitly, tolerance to insertion errors in predecessor steps, and so on. The empirical experiments based on thousands of long documents from Chinese, English financial market and English scientific publication show that the HELD model with the "root-to-leaf" traversal order and explicit heading extraction is the best choice to achieve the tradeoff between effectiveness and efficiency with the accuracy of 0.9726, 0.7291 and 0.9578 in Chinese financial, English financial and arXiv datasets, respectively. Finally, we show that logical document hierarchy can be employed to significantly improve the performance of the downstream passage retrieval task. In summary, we conduct a systematic study on this task in terms of methods, evaluations, and applications.

cs.IR↗