entity resolution

Low accuracy in institutional normalization of author affiliation strings—characterized by nested multi-institutional structures and noise—hampers bibliometric analysis and cross-knowledge-base interoperability. Method: We propose AffRo, an end-to-end framework that jointly addresses affiliation parsing and coreference resolution, integrating rule-enhanced named entity recognition (NER), hierarchical organizational coreference resolution, and context-aware matching ranking. Contribution/Results: We introduce AffRoDB, the first expert-annotated benchmark dataset for affiliation normalization, filling a critical gap in systematic evaluation. On diverse, real-world affiliation strings from multiple sources, AffRo achieves a 12.6% absolute F1-score improvement over state-of-the-art methods, significantly enhancing scholarly metadata quality and enabling robust interoperation of organizational identifiers across knowledge bases.

Medical text sharing poses re-identification risks for patients and healthcare providers due to indirect identifiers—attributes that, when combined with auxiliary information, enable identity inference. Method: We propose the first threat-model-aware, structured taxonomy of nine categories of indirect identifiers, covering diverse attacker scenarios (e.g., acquaintances, family members, clinicians). Leveraging 100 MIMIC-III discharge summaries, we conduct fine-grained, human-in-the-loop annotation (6,199 labeled spans with document IDs), augmented by rule-based heuristics and a BiLSTM-CRF sequence labeling model. Contribution/Results: We release the first publicly available medical text dataset explicitly annotated for indirect identifiers. Our best-performing baseline achieves an F1-score of 78.3%, significantly improving detection of context-sensitive privacy-sensitive information. This advances beyond conventional de-identification methods—which focus solely on direct identifiers—and enables granular, risk-adaptive privacy assessment in clinical NLP applications.

Large language models are prone to generating hallucinations or dubious citations in academic writing, undermining research credibility. This study presents the first systematic evaluation and comparison of mainstream citation verification tools—CheckIfExist, HalluCiteChecker, Hallucinator, HalRef, and RefChecker—on real-world academic documents. The analysis reveals significant limitations in current approaches, particularly concerning citation extraction accuracy, breadth of database coverage, and consistency in verification. While these tools can offer preliminary alerts for potentially fabricated references, their overall effectiveness remains constrained. This work provides an empirical foundation and clear directions for improving the verification of citation authenticity in scholarly communication.

This work addresses the growing threat of hallucinated citations—fabricated references generated by large language models—that have infiltrated top-tier academic venues such as ICLR, ICML, NeurIPS, and USENIX Security, undermining scholarly credibility. To combat this, the authors introduce RefChecker, the first scalable citation verification pipeline tailored for large-scale analysis of conference papers. RefChecker integrates multi-source academic database matching with web-based re-verification to efficiently assess citation authenticity under conservative criteria. The study presents the first systematic quantification of citation hallucinations under a rigorous definition, revealing that approximately 5% of NeurIPS and USENIX Security papers—including some award-winning works—contain at least two hallucinated references. Notably, the approach achieves audit costs as low as $0.04 per paper, demonstrating that automated, low-cost, and reproducible large-scale citation auditing is both feasible and practical.

This work addresses data contamination caused by irreversible entity merging and ontology misclassification based on name fragments in knowledge graph construction. The authors propose a “review-before-linking” mechanism featuring an identity-ladder strategy—leveraging identifiers, names, and type scopes—to enable controlled deduplication, alongside anchor-evidence constraints that govern multi-class ontology label assignment. This approach corrects the evidential asymmetry arising when names are treated as instance labels rather than type assertions. Integrated into a system combining automated merging, evidence validation, and a human review queue, the method was evaluated on a knowledge graph comprising 537,157 entities and 2,198,567 relations. It reduced role assignment errors from 36 to zero, requiring only 775 manual decisions to resolve 48,403 merge proposals, thereby significantly mitigating risks of over-merging and misclassification.

Large language models (LLMs) frequently generate fabricated references in scholarly writing—such as fictitious authors or incorrect DOIs—a problem that has already infiltrated top-tier conferences and journals and often evades detection by conventional peer review. This work proposes the first automated detection tool that integrates LLM-based field extraction with structured querying of academic databases. Specifically, the method employs an LLM to parse citation fields and then leverages Semantic Scholar to perform semantic matching and similarity scoring based on title, author names, and publication venue, yielding a tiered credibility assessment (credible, partially supported, or likely fabricated). Evaluated on a manually annotated test set of papers accepted at NeurIPS 2025, the approach efficiently identifies the vast majority of hallucinated citations, offering a scalable technical safeguard for research integrity.

This work addresses the challenge of detecting hallucinated citations generated by large language models in academic writing, a problem exacerbated by existing methods that rely on fragile parsing or incomplete retrieval and thus lack fine-grained discriminative capability. The authors propose CiteTracer, the first multi-agent cascaded detection framework capable of field-level citation verification, reframing hallucination detection as a 12-class truthfulness classification task. Integrating structured parsing (from PDFs and BibTeX), multi-source evidence retrieval (via academic search engines, web search, and URL scraping), and an expert routing mechanism, CiteTracer achieves 97.1% overall accuracy on a benchmark comprising 2,450 synthetic and 957 real-world hallucinated citations, with F1 scores of 97.0, 95.8, and 98.5 across three critical categories—significantly advancing the detection of ambiguous and fabricated references.

This work addresses the degradation in retrieval performance and lack of decision credibility observed when frozen multimodal encoders are deployed compositionally due to varying connection paths. To tackle this, the authors propose CertBind, a novel framework that extends multimodal compositionality from representation learning to certifiable task-level decisions. CertBind introduces a four-tier certification mechanism—spanning nodes, edges, paths, and queries—integrated with anchored boundary modeling, contract-aware conformal ranking, overlap-aware budget allocation, and clean calibration to construct a certified retrieval system with a finite-sample recovery radius. Experiments demonstrate that under shared-path C-MCR settings, CertBind recovers 96.3% of the original retrieval performance while achieving perfect branch accuracy (1.000), effectively balancing compositional extensibility with decision reliability.

This study addresses the lack of authenticity verification in Git commit authorship, where identity resolution relies solely on unverified claims. Leveraging the World of Code V2604 dataset, this work presents the first large-scale analysis of cryptographic signatures (PGP, SSH, and X.509) across 5.8 billion commits, introducing A2trust—a novel four-tier identity trust labeling framework that distinguishes declared identities from cryptographically bound ones. The contributions include the release of c2sigFull, a large-scale signed-commit dataset; the construction of an author-key graph that disambiguates organizational and personal keys; and the generation of a high-precision alias gold standard for 17.59% of signed commits. These resources establish cryptographic anchors for software identities, enabling trustworthy identity linkage and reproducible research in software provenance.

This study addresses the challenge of identity linkage among hundreds of thousands of Islamic hadith transmitters across heterogeneous Arabic biographical databases, where the absence of unified identifiers impedes cross-resource integration. To resolve this, the work proposes the first cross-database, multi-signal entity resolution framework tailored for Arabic hadith transmitters. The approach employs a two-stage pipeline: first linking transmitters from the Sanadset corpus to the HadithTransmitters database via name similarity, then aligning with the MuslimScholars database through a weighted fusion of multiple signals—namely name, death year, and reliability rating—augmented by a transitive linking strategy. Operating without metadata, the framework achieves high-coverage identity resolution, constructing a directed transmission graph comprising 185,216 nodes and 814,093 edges. This effort yields the first structured integration of these three major resources, accompanied by the public release of high-quality linked corpora and a cross-source biographical knowledge graph.