1167 research outputs found
Sort by
Automatic Reviewers Fail to Detect Faulty Reasoning in Research Papers: A New Counterfactual Evaluation Framework
No full textLarge Language Models (LLMs) have great potential to accelerate and support scholarly peer review and are increasingly used as fully automatic review generators (ARGs). However, potential biases and systematic errors may pose significant risks to scientific integrity; understanding the specific capabilities and limitations of state-of-the-art ARGs is essential. We focus on a core reviewing skill that underpins high-quality peer review: detecting faulty research logic. This involves evaluating the internal consistency between a paper’s results, interpretations, and claims. We present a fully automated counterfactual evaluation framework that isolates and tests this skill under controlled conditions. Testing a range of ARG approaches, we find that, contrary to expectation, flaws in research logic have no significant effect on their output reviews. Based on our findings, we derive three actionable recommendations for future work and release our counterfactual dataset and evaluation framework publicly.
This dataset includes the papers, the extracted research logic for each paper, and all predictions and evaluations per ARG and review difference dimensions. Use it together with the code provided here: https://github.com/UKPLab/counter-review-logic0.
Code and Supplementary Data Publication to the paper "Optimal Ventilation Systems via MILP: Duct Sizing, Fan Placement, Control Strategies"
No full textThis dataset is the code and supplementary data publication for the paper "Optimal Ventilation Systems via MILP: Duct Sizing, Fan Placement, Control Strategies".
It contains all results with full details in the form of HDF5 files. Further, it contains three zipped github repos that were used within the creation of the paper.
1. The entire code base to perform preprocessing, optimisation and postprocessing the optimisations. It contains the optimisation problem as well as all the input data needed to rerun the existing or to perform own optimisation.
2. The code to cluster the load cases
3. The code to underestimate using hyperplanes, i.e. the relaxation performed in the paper.
The repositories are available under "Related Resources".
The latter two additionally contain the preprocessing steps for the paper as examples. More information about the repository can be found under "description_this_repository.txt".
The presented results were obtained within the research project “Algorithmic System Planning of Air Handling Units”, Project No. 22289 N/1, funded by the program for promoting the Industrial Collective Research (IGF) of the German Ministry of Economic Affairs and Climate Action (BMWK), approved by the Deutsches Zentrum für Luft- und Raumfahrt (DLR). We want to thank all the participants of the working group for the constructive collaboration.
Hall effect data of doped indium oxide thin films
No full textThe file contains electrical conductivity and Hall effect data of more than 600 thin films prepared by magnetron sputtering with different dopants and deposition parameters. The data are collected in the group of Andreas Klein (formerly Surface Science, now Electronic Structure of Materials) in the Materials Science Department at Technical University of Darmstadt. The data are summarized in a single file of the software "Igor Pro" from the company "Wavemetrics". A demo version of the program can be downloaded at wavemetrics.com. The demo version allows to open the file and see all data, tables and graphs. It is available for Apple and Windows.The current version contains data collected until October 202
NeoQA: Evidence-based Question Answering with Generated News Events
No full textPublic key ("23456") encrypted dataset for the publication "NeoQA: Evidence-based Question Answering with Generated News Events" (ACL Findings 2025
Simulation Dataset of Utilizing a Design Optimization Platform for Personalized Exosuit Design through Predictive Simulations
No full textThis file contains the dataset of the implementation of the Design Optimization Platform for the design of the passive BATEX device. The dataset for the passive BATEX experiments used in this study can be found in the following directory:
Ahmad Sharbafi, M., Ahmadi, A., Firouzi, V., Davoodi, A., Haufe, D., & Seyfarth, A. (2024). Biomechanical Dataset for Walking with Passive Biarticular Thigh Exosuit (BATEX).
The data for each participant of the experiment is given in separate folders (i.e., S1 to S8). In each folder, the optimization platform files, the SCONE scenarios of baseline and assisted simulations, and the results of the predictive simulations are given
ARR Data Collection Initiative 2025
No full textDataset of peer review reports, meta-reviews, reviewer-author discussions, and paper drafts collected from ACL Rolling Review within the context of the new data collection initiative (https://arr-data.aclweb.org/protocol/). All included data is explicitly licensed by the authors and reviewers for publication. This dataset is not meant for commercial purposes. This dataset should not be used for pre-training of neural models such as large language models.
V1 contains accepted paper data from COLING 2025 and NAACL 2025.
V1.1 and V1.1.1 contain accepted paper data from ACL 2025 (V1.1.1 contatins data from ARR 2024 December and ARR 2025 February, and V1.1 contains only data from ARR 2025 February).
V1.2 contains ARR 2024 April and June submissions that did not appear in EMNLP 2024, released after a one-year grace period with explicit author consent.
V1.3 contains accepted paper data from EMNLP 2025
Business Model Canvas für zirkuläre Geschäftsmodelle - Ableitung der Sub-Dimensionen
No full textZur Erweiterung des Business Model Canvas auf den Bereich zirkulärer Geschäftsmodelle wurde eine qualitative Interviewstudie durchgeführt. In dieser werden auf Basis der Interviewtranskripte Kategorien und Sub-Dimensionen abgeleitet, die wiederum den neun etablierten Dimensionen des Business Model Canvas zugeordnet werden. Eine Veröffentlichung der Transkripte war seitens der Teilnehmer:innen nicht gewünscht. Die veröffentlichte Excel-Datei stellt den Zusammenhang zwischen Dimensionen des Business Model Canvas sowie den abgeleiteten Kategorien und Sub-Dimensionen her
Exposure of cells to near-infrared irradiation relaxes chromatin compaction and facilitates recognition of cyclo-butane pyrimidine dimers
No full textUltraviolet A and B (UVA 320-400 nm and UVB 280-320 nm) induced cyclobutane-pyrimidine dimers (CPDs) are the most critical lesions caused by environmental sun exposure. Here we show that CPD removal is accelerated when, in addition to UV, cells are simultaneously exposed to water-filtered near-infrared (nIR, 750-1600 nm). The described effect is dose-dependent on the nIR-dose and is found in skin keratinocytes and fibroblasts. Accelerated removal of CPDs, which coincides with chromatin relaxation and faster CPD recognition, occurs after nIR exposure. While nIR alone does not affect cellular survival, co-exposure to UVB leads to reduced cellular survival and an increased number of mutations. Increasing single strand break levels (SSB) occur transiently after nIR exposure and independent of reactive oxygen species (ROS) formation. These data suggest that the rate-limiting step in the NER repair process – damage recognition – is facilitated by nIR-induced chromatin relaxation, causing the accumulation of unnatural high levels of SSBs and single stranded DNA, unfavourable for the cell fate resulting in reduced survival and increased mutation rates. Since nIR modulates the UV-dependent damage response, risk estimation of solar radiation-induced DNA damage should not only consider the UV components but also include the nIR fraction of the solar spectrum.submissio
Supplementary movies: Cloud cavitation dynamics with varying incidence
No full textThe high-speed movies serve as supplementary material for the PhD thesis in preparation titled "Cloud cavitation dynamics with varying incidence" by Grigorios Hatzissawidis (Technische Universität Darmstadt). The data were collected in the cavitation tunnel at the Cavitation Research Laboratory (CRL), Australian Maritime College, University of Tasmania. High-speed photography and simultaneous force measurements using a six-component force balance were conducted. Cloud cavitation was observed about a NACA 0015 hydrofoil. From the full dataset, selected subsets were used to create movies that support and visualise the findings. For each movie, a corresponding JSON file is provided with a description and other metadata. For movies 7 and 8, the continuous wavelet transform (CWT) of the lift coefficient is shown alongside the high-speed movies