Papers where the original authors have engaged with our plain-language explanation.
For every paper on this page, at least one of the original authors has seen our plain-language explanation and engaged with it — either confirming it reads accurately or requesting corrections that we then applied. An endorsement does not mean the authors formally approve every sentence, but it does mean the explanation has passed the eyes of the people who wrote the paper.
948 papers reviewed by authors · 11–20 / 948
This paper introduces ScienceMeter, a framework for evaluating scientific knowledge updates in Large Language Models through three metrics—preservation, acquisition, and projection—revealing that current update methods struggle to maintain existing knowledge while effectively learning new facts and anticipating future scientific advancements.
This bibliometric review of 437 publications from 1996 to 2026 analyzes the thematic evolution, collaboration networks, and emerging gaps in wetland and waterway mitigation research, revealing a shift toward restoration effectiveness and ecosystem-based management while highlighting the need for standardized frameworks and advanced geospatial technologies.
This paper demonstrates that diversity-generating retroelements (DGRs) can be engineered in *Escherichia coli* and coupled with horizontal gene transfer to create a simple, programmable system for the continuous, iterative, and position-specific mutagenesis of target genes, enabling the rapid evolution of proteins such as pyrrolysyl-tRNA synthetase for biotechnological applications.
This survey reviews over two decades of research on border basis schemes, highlighting their defining property of commuting multiplication matrices, the challenges and solutions regarding their high-dimensional coordinate rings through re-embedding techniques, and their connections to positive -algebras and unimodular matrix problems, while concluding with a selection of open problems that suggest the field's ongoing evolution.
This paper develops a canonical Post-Newtonian Hamiltonian framework for radiating charges by integrating the Landau-Lifshitz reduced radiation reaction into the Darwin system to derive inspiral laws, and extends this formalism to charged compact binaries in Einstein-Maxwell theory to characterize the transition between electromagnetic and gravitational flux-dominated inspirals.
This study utilizes a network-based metric to demonstrate that elite AI chess players sustain significantly higher and longer-lasting strategic tension than human grandmasters, revealing that AI systems tolerate densely interconnected positions while humans systematically limit complexity, with human tension levels peaking at ply 40 and scaling linearly with skill and time controls.
This paper establishes a quantitative framework for interpreting quantum Hall scanning-probe images by deriving a transport operator functional that distinguishes between thermal, electrical, and viscous contrasts, revealing that thermoelectric signals are pinned by energy weighting rather than defect geometry in strong fields and that Hall-odd images in hydrodynamic regimes require careful disentanglement of mixed transport channels.
This systematic benchmarking study demonstrates that frontier large multimodal models, particularly Gemini 2.5 Pro and GPT-4o, significantly outperform both time-pressured human experts and traditional NLP baselines in extracting critical perineural invasion status from unstructured gastrointestinal pathology reports, while a novel failure mode taxonomy provides actionable guidance for model selection based on specific error signatures.
This paper introduces ffconnect, an open-source Python interface for the mid-fidelity FAST.Farm simulator that streamlines wind farm control integration with modern machine learning ecosystems through a restructured API while maintaining negligible runtime overhead.
This study benchmarks various classical machine learning models on TCGA-BRCA multi-omics data for predicting breast cancer Estrogen Receptor status, finding that Random Forest achieved the highest performance (90.3% balanced accuracy) by effectively integrating transcriptomic, genomic, and proteomic features while identifying biologically relevant genes.