MatMMExtract pipeline creates MatSciFig dataset of 391k annotated materials science figure panels and MaterialScope detection dataset with high accuracy.
hub
OpenAlex: A fully-open index of scholarly works, authors, venues, institutions, and concepts
50 Pith papers cite this work, alongside 273 external citations. Polarity classification is still indexing.
50
Pith papers citing it
273
external citations · Pith
abstract
OpenAlex is a new, fully-open scientific knowledge graph (SKG), launched to replace the discontinued Microsoft Academic Graph (MAG). It contains metadata for 209M works (journal articles, books, etc); 2013M disambiguated authors; 124k venues (places that host works, such as journals and online repositories); 109k institutions; and 65k Wikidata concepts (linked to works via an automated hierarchical multi-tag classifier). The dataset is fully and freely available via a web-based GUI, a full data dump, and high-volume REST API. The resource is under active development and future work will improve accuracy and coverage of citation information and author/institution parsing and deduplication.
hub tools
citation-role summary
background 3
dataset 1
citation-polarity summary
representative citing papers
Established and new NLP authors are migrating from *ACL venues to general ML venues, with ML venues conferring a large content-matched citation premium.
LightGBM models on citation and diversity features predict exogenous diffusion of quantum computing concepts with R² up to 0.78 while endogenous reinforcement remains largely unpredictable after growth controls, with replications in other fields.
Moment-based method identifies regression coefficient beta and LDA concentration alpha_0 from response-weighted corrected word moments and commutativity, without per-document topic estimation.
MemGym unifies agent gyms into a memory benchmark with isolated scoring across tool-use, research, coding, and computer-use regimes plus a lightweight reward model for tractable coding evaluation.
12% of OpenAlex English-language journal abstracts exhibit integrity issues, primarily insufficient content and misplaced metadata, identified through a two-stage human-LLM annotation protocol on a random sample of 10,000 items.
Phantom collaborators—topically similar authors distant in the coauthor graph—become actual coauthors 16-33 times more often than baselines, with a 68-fold similarity gradient.
A new corpus of 301,871 systematic reviews across all sciences is released with extracted method artifacts to support retrieval benchmarking and meta-research.
The innovation annulus is a functional, persistent feature of scholarly metadata production whose width reflects production inefficiency, reshaped by AI and best managed through calibrated governance analogous to optimal patent duration.
LLM agents execute scientific tasks but fail to follow core scientific reasoning norms such as evidence consideration and belief revision based on refutations.
Camyla autonomously generates research proposals, experiments, and manuscripts in medical image segmentation, outperforming baselines on 24 of 31 recent datasets while producing 40 human-reviewed papers.
StructBioReasoner is a scalable multi-agent system that designs IDP-targeting biologics, with over 50% of 787 candidates for Der f 21 showing better binding free energy than human-designed references.
LLM agents surface methodological critiques of computational-physics papers primarily by re-running calculations, not by reading; a deep case revises a Nature Communications L_G=5 nm claim with attacks missing from 21-reviewer peer review.
The paper defines five AI system categories for public administration and reports that 55% of 91 recent papers leave the system type underspecified while 31% study one type but motivate with another.
Lacuna is an LLM-powered research map for ML that outperforms OpenScholar on retrieval benchmarks and GPT-Researcher on multi-stage report generation tasks.
TopVenues materializes a versioned, reproducible corpus of 9,925 cybersecurity papers from 11 venues (2017-2026) with 99.86% abstract coverage and tooling for auditable literature reviews and preprint analysis.
A two-stage LightGBM model on 59 features from concept networks forecasts link formation and intensity with ROC-AUC 0.95-0.967 across domains.
Deep bibliography expansion in literature search achieves high recall while human citations are found to have only 51% moderate relevance compared to 86-88% for AI methods.
Audits of 43 LLMs show that varying persona prompts (language, location, role-and-task) and context affects technical quality and social representativeness of scholar recommendations, with location impacting diversity and factuality.
GSR is a new citation-normalized ranking framework for CS venues and medical journals that combines four open-data metrics with fixed quartile quotas and an approximation for conference impact factors.
Proposes paper.json as a coordination convention supplying stable claim IDs, does-not-claim lists, figure shell commands, and definition IDs so LLM agents can act on papers more reliably.
Faculty mobility follows a persistent institutional prestige hierarchy but yields little evidence of lasting improvements in movers' research productivity or citation impact.
Intern-Atlas constructs a methodological evolution graph with 9.4 million edges from 1.03 million AI papers to capture how methods emerge, adapt, and transition, enabling better idea evaluation and generation for AI-driven research.
CiteRadar is a new open-source pipeline that enriches Google Scholar citations using five external data sources and produces ranked tables plus an offline interactive geographic map from a single command.
citing papers explorer
-
Unlocking the Visual Record of Materials Science: A Large-Scale Multimodal Dataset from Scientific Literature
MatMMExtract pipeline creates MatSciFig dataset of 391k annotated materials science figure panels and MaterialScope detection dataset with high accuracy.
-
The Future of NLP may not be at NLP Conferences: Scholarly Migration Patterns in Natural Language Processing
Established and new NLP authors are migrating from *ACL venues to general ML venues, with ML venues conferring a large content-matched citation premium.
-
Forecasting Conceptual Diffusion in Science: The Case of Quantum Computing
LightGBM models on citation and diversity features predict exogenous diffusion of quantum computing concepts with R² up to 0.78 while endogenous reinforcement remains largely unpredictable after growth controls, with replications in other fields.
-
Moment-Based Inference for Regression with Latent Dirichlet Covariates
Moment-based method identifies regression coefficient beta and LDA concentration alpha_0 from response-weighted corrected word moments and commutativity, without per-document topic estimation.
-
MemGym: a Long-Horizon Memory Environment for LLM Agents
MemGym unifies agent gyms into a memory benchmark with isolated scoring across tool-use, research, coding, and computer-use regimes plus a lightweight reward model for tractable coding evaluation.
-
One in Eight OpenAlex Abstracts Has Integrity Issues
12% of OpenAlex English-language journal abstracts exhibit integrity issues, primarily insufficient content and misplaced metadata, identified through a two-stage human-LLM annotation protocol on a random sample of 10,000 items.
-
Beyond coauthorship: semantic structure and phantom collaborators in transportation research, 1967--2025
Phantom collaborators—topically similar authors distant in the coauthor graph—become actual coauthors 16-33 times more often than baselines, with a 68-fold similarity gradient.
-
A Large-Scale, Cross-Disciplinary Corpus of Systematic Reviews
A new corpus of 301,871 systematic reviews across all sciences is released with extracted method artifacts to support retrieval benchmarking and meta-research.
-
Market Dynamics, Governance and Open Research Metadata in the AI Era
The innovation annulus is a functional, persistent feature of scholarly metadata production whose width reflects production inefficiency, reshaped by AI and best managed through calibrated governance analogous to optimal patent duration.
-
AI scientists produce results without reasoning scientifically
LLM agents execute scientific tasks but fail to follow core scientific reasoning norms such as evidence consideration and belief revision based on refutations.
-
Camyla: Scaling Autonomous Research in Medical Image Segmentation
Camyla autonomously generates research proposals, experiments, and manuscripts in medical image segmentation, outperforming baselines on 24 of 31 recent datasets while producing 40 human-reviewed papers.
-
Scalable Agentic Reasoning for Designing Biologics Targeting Intrinsically Disordered Proteins
StructBioReasoner is a scalable multi-agent system that designs IDP-targeting biologics, with over 50% of 787 candidates for Der f 21 showing better binding free energy than human-designed references.
-
Grounded autonomous scrutiny at scale: emergent critique from reproduction of published computational physics papers
LLM agents surface methodological critiques of computational-physics papers primarily by re-running calculations, not by reading; a deep case revises a Nature Communications L_G=5 nm claim with attacks missing from 21-reviewer peer review.
-
A Technical Typology of AI Systems in Public Administration
The paper defines five AI system categories for public administration and reports that 55% of 91 recent papers leave the system type underspecified while 31% study one type but motivate with another.
-
Lacuna: A Research Map for Machine Learning
Lacuna is an LLM-powered research map for ML that outperforms OpenScholar on retrieval benchmarks and GPT-Researcher on multi-stage report generation tasks.
-
TopVenues: A Reproducible Corpus and Tooling Substrate for Cybersecurity Literature Reviews
TopVenues materializes a versioned, reproducible corpus of 9,925 cybersecurity papers from 11 venues (2017-2026) with 99.86% abstract coverage and tooling for auditable literature reviews and preprint analysis.
-
Explainable Forecasting of Scientific Breakthroughs from Concept Network Dynamics
A two-stage LightGBM model on 59 features from concept networks forecasts link formation and intensity with ROC-AUC 0.95-0.967 across domains.
-
Rethinking Literature Search Evaluation: Deep Research Helps, and Human Citation Lists Are Not a Ground Truth
Deep bibliography expansion in literature search achieves high recall while human citations are found to have only 51% moderate relevance compared to 86-88% for AI methods.
-
Whose Name Comes Up? III: Persona Prompting Effects in LLM-Based Scholar Recommendation
Audits of 43 LLMs show that varying persona prompts (language, location, role-and-task) and context affects technical quality and social representativeness of scholar recommendations, with location impacting diversity and factuality.
-
General Science Ranking (GSR): An Open-Source, Citation-Normalized Journal and Conference Classification System for Computer Science and Medicine
GSR is a new citation-normalized ranking framework for CS venues and medical journals that combines four open-data metrics with fixed quartile quotas and an approximation for conference impact factors.
-
paper.json: A Coordination Convention for LLM-Agent-Actionable Papers
Proposes paper.json as a coordination convention supplying stable claim IDs, does-not-claim lists, figure shell commands, and definition IDs so LLM agents can act on papers more reliably.
-
Faculty mobility reallocates research capacity within persistent institutional hierarchies
Faculty mobility follows a persistent institutional prestige hierarchy but yields little evidence of lasting improvements in movers' research productivity or citation impact.
-
Intern-Atlas: A Methodological Evolution Graph as Research Infrastructure for AI Scientists
Intern-Atlas constructs a methodological evolution graph with 9.4 million edges from 1.03 million AI papers to capture how methods emerge, adapt, and transition, enabling better idea evaluation and generation for AI-driven research.
-
CiteRadar: A Citation Intelligence Platform for Researcher Profiling and Geographic Visualization
CiteRadar is a new open-source pipeline that enriches Google Scholar citations using five external data sources and produces ranked tables plus an offline interactive geographic map from a single command.
-
AI-assisted writing and the reorganization of scientific knowledge
Post-2023, AI-assisted writing intensity positively associates with scientific disruption but shows weakened links to cross-field citation breadth and attenuated negative links to citation concentration.
-
AgentSPEX: An Agent SPecification and EXecution Language
AgentSPEX is a new language and harness for explicitly specifying and running structured LLM-agent workflows with typed steps, control flow, parallel execution, and a visual editor.
-
Structural Diversity Drives Disruptive Scientific Innovation
Structural diversity in a team's prior collaboration network predicts disruptive scientific innovation more strongly than team freshness or edge density and turns large team size from a liability into an advantage via better disciplinary integration.
-
The Tilted Playing Field for Women in Science
Large-scale bibliometric analysis shows gender differences in prestige returns, with men benefiting across the institutional hierarchy and women primarily at the top, linked to differing collaboration networks.
-
DeepSciVerify: Verifying Scientific Claim--Citation Alignment via LLM-Driven Evidence Escalation
DeepSciVerify achieves 86.7 Micro-F1 on SCitance for claim-citation verification via abstract-level reasoning plus selective full-text escalation, beating abstract-only baselines by 4.5 points and skipping full-text retrieval in 67% of cases.
-
Agentic Discovery of Cryomicroneedle Formulations
A closed-loop workflow using Gaussian process surrogate modeling and Bayesian optimization, updated over ten iterations with 106 wet-lab tests, adapted from literature data to identify a cryoprotectant formulation achieving 95.15% post-thaw viability for cryomicroneedles.
-
pAI/MSc: ML Theory Research with Humans on the Loop
pAI/MSc is a customizable multi-agent system that reduces human steering by orders of magnitude when turning a hypothesis into a literature-grounded, mathematically established, experimentally supported manuscript draft in ML theory.
-
Scientific tools and Innovation: Big Science Facilities Yield More Novel and Interdisciplinary Knowledge
Big Science Facilities produce publications with greater recombinant novelty and interdisciplinary integration than matched controls, with stronger effects in fields outside their traditional physical-sciences focus.
-
Polarization and Integration in Global AI Research
Over three decades, global AI research has polarized into US and China poles, with UK/Germany aligning with US, some Europeans with both, and developing countries with China.
-
Can We Still Hear the Accent? Investigating the Resilience of Native Language Signals in the LLM Era
NLI accuracy on research papers declined steadily over time, with Chinese and French showing unexpected resistance while Japanese and Korean declined more sharply in the post-LLM era.
-
Mapping the Landscape of Open Access Dashboards -- A Dataset for Research and Infrastructure Development
A survey identifies nearly 60 open access dashboards and supplies a structured metadata dataset plus community contribution process for open science research.
-
Quantifying Lifetime Productivity Changes: A Longitudinal Study of 320,000 Late-Career Scientists
Large-scale longitudinal tracking shows high immobility in productivity deciles among late-career scientists, with early top performance as the strongest predictor of later top performance.
-
Didact: A Cross-Domain Capability Discovery System for Defence
Didact is a RAG-based prototype with an Evidence Rail for conversational capability discovery from integrated Australian defence documents and research publications.
-
Quantifying the evolving topical structure of science across journals, countries, regions, and research domains
Presents a reproducible framework that combines a unified topic ontology with short time-series estimators to quantify topical prevalence and dynamics in science from OpenAlex metadata.
-
Global training and the collaborative structure of elite U.S. science
Foreign-trained U.S. faculty show higher aggregate elite publications due to institutional placement and team size in mixed domestic-foreign groups, with within-strata differences attenuating sharply and little evidence of niche topic distinctiveness.
-
CitePrism: Human-in-the-Loop AI for Citation Auditing and Editorial Integrity
CitePrism is a human-in-the-loop framework combining LLM contextual analysis, semantic embeddings, and metadata verification for citation auditing, achieving Cohen's kappa of 0.429 against human labels in a single manuscript with 104 references.
-
A Template-Driven Platform for Contextualised Researcher Profiles
BIP! Scholar is a template-driven platform enabling contextualized, multi-dimensional researcher profiles for varied presentation and assessment needs.
-
Construction of a Battery Research Knowledge Graph using a Global Open Catalog
A pipeline builds a battery research knowledge graph from 189k OpenAlex papers using author vectors weighted by OpenAlex concepts, KeyBERT/ChatGPT keyphrases, authorship position, and recency, then serializes it as RDF linked to Wikidata.
-
Auditing automated research assessment: an interpretable machine learning approach to validate funding criteria
ML models show Brazilian PQ grant levels are predicted well by a small set of bibliographic and supervision features but not by the full set of official criteria.
-
Women in Science: Measuring Participation in Europe Across Disciplines, Generations and Over Time
Bibliometric study of 1.74 million scientists shows women at 50%+ in four STEMM disciplines and >50% of young scientists in five, but marginal role in growth of four mathematical disciplines across Europe 1990-2023.
-
Biomedical Open Source Software: Crucial Packages and Hidden Heroes
Network analysis of software mentions in biomedical papers identifies high-centrality foundational packages in three major open-source ecosystems.
-
Mapping bibliographic metadata collections: the case of OpenCitations Meta and OpenAlex
Authors map entities between OpenCitations Meta and OpenAlex to add identifiers and evaluate bibliographic metadata consistency.
-
Beyond Metadata: The Role of DOI, ORCID, and ROR in Shaping Transparent and Interoperable Library Systems
DOI, ORCID, and ROR have evolved from metadata tags into machine-actionable links connecting research entities, with ORCID adoption varying 41-89% across German organizations and ongoing metadata quality issues.
- The Reciprocal Impact of Science and Software: A Cross-Corpus Analysis of How Research Shapes Software and Software Enables Research
- War in the Abstract: The Rise and Consequences of Militarized Language in Scientific Communication
- HALvest-Contrastive: Retrieval-Like Authorship Attribution with Patch-Level Late Interaction