Mendeley Data Showcase

Disturbance-Impact-Domain-Driven Feature-Adaptive Transfer Reinforcement Learning for Reactive Robust Scheduling in Ship-Block Mixed-Flow Construction

The Darbest dataset is a Universal Dependencies (UD) treebank dataset for Standard Sorani Kurdish written in the Perso-Arabic script. It contains 69,000 annotated sentences and 1,205,855 tokens collected from nine textual domains. The corpus was collected from seven Kurdish online news websites and supplemented with texts from published books. Before preprocessing, the collected corpus contained 1,250,275 words from 5,627 web pages together with book-based texts and was preprocessed using a Python-based pipeline involving text cleaning, Unicode and punctuation normalization, sentence segmentation, and tokenization. The dataset was developed to provide a large-scale syntactically and morphologically annotated resource for Standard Sorani Kurdish. A separate 100-sentence gold-standard set was manually annotated according to the Universal Dependencies v2 guidelines. Sorani Kurdish linguistic experts supported the selection of sentences representing diverse and linguistically complex structures and reviewed LLM-generated annotations for errors. The gold-standard set was used to construct few-shot prompts for annotating the remaining corpus. The resulting annotations were represented in the standard CoNLL-U format and validated using the official Universal Dependencies validation tool, followed by manual correction and quality review. The released treebank is divided into training, development, and test sets and includes lemmas, Universal Part-of-Speech (UPOS) tags, morphological features, syntactic heads, and dependency relations. The dataset can be used to train, evaluate, and benchmark NLP models for part-of-speech tagging, lemmatization, morphological analysis, dependency parsing, and related computational linguistics tasks. The accompanying repository also contains the separate 100-sentence gold-standard set, plain-text corpus splits, README documentation, and a dataset statistics spreadsheet.

This dataset consists of 1159 entries scraped from Reddit and X/Twitter between February 10 and March 20, 2026, including at least 100 posts from each platform representing each of four different political/cultural/historical/technological moments in the online cultural struggle between the interests of LGBTQIA+ youth and those of conservative Christian ideologies and institutions. Posts were collected from before the Trump era, during the Trump Era but before the COVID-19 lockdowns, during the COVID Era, and after the COVID Era. Each post has been assigned three sentiment codes: One by a machine learning approach created for general language, one by a machine learning approach created specifically for X/Twitter, and one by a social scientist who researches LGBTQIA+ youths’ issues with conservative Christianity, who has previously published research that involved coding short-format text data. To keep work on this dataset within the realm of activity that is not human subjects research, the dataset does not include the body text of posts marked as deleted or removed in their respective databases. For the sake of technical completeness, the small percentage of posts found by our algorithms that were not really part of the discourse (e.g., product advertisements, “AI slop”) were left in, along with the sentiment categories our machine learning approaches assigned to them. This dataset may be useful to investigators of LGBTQIA+ youths’ real-lived experiences with online discourse involving conservative Christian ideologies and institutions, and to computer scientists interested in training sentiment analysis models to address politically charged issues.

This dataset contains pseudonymized dynamic Automatic Identification System (AIS) vessel trajectory data from the Port of Antwerp, collected via the AISStream.io API between 31 December 2023 and 24 January 2024. Collection was not continuous: no or fewer observations exist for 4, 6, 7, 8, or 11 January 2024. Following spatial restriction to the Port of Antwerp and removal of invalid identifiers, 5,462,834 messages from 2,553 vessels were retained. Trips were separated using a six-hour inter-message threshold, stopped observations were defined as speed over ground below 1 knot, and moving subtrajectories shorter than three points were discarded (see our published work for further details), yielding 956,995 observations from 2,377 vessels across 5,488 trips and 36,748 moving subtrajectories. Three files are provided. training_trajectories.csv (884,044 observations) is the trip-level 90% partition used for graph construction. test_trajectories_complete.csv (72,951 observations) contains the complete held-out moving subtrajectories used as ground truth. test_trajectories_sparse.csv (15,189 observations) is the same held-out data after synthetic gaps were introduced, generated within each subtrajectory using normally distributed temporal thresholds with mean 6 minutes and standard deviation 1.5 minutes. The train/test split was performed at trip level; 355 vessels contribute trips to both partitions. Each observation provides position, speed over ground, course over ground, reported true heading, and navigational status. These data support research on vessel trajectory reconstruction, maritime traffic modelling, and movement pattern analysis in constrained port environments, and were used for our work titled "Data-Driven Graph-Based Methods for In-Port AIS Vessel Trajectory Reconstruction." The data are pseudonymized rather than anonymous: original MMSIs, vessel names, and source identifiers have been removed and replaced with stable pseudonymous identifiers, but exact timestamps and positions describe public vessel movements and may permit linkage to external AIS sources. Users must not attempt to re-identify vessels or individuals.

Reproducibility package for the manuscript “Speed-coupled fuel storage and the probabilistic life-cycle cost of a storage-neutral nuclear boundary against capture and alternative-fuel pathways in deep-sea container shipping.” The package contains source code, declared model inputs, parameter-provenance information, expected-results checks, environment specifications, and scripts for regenerating all reported figures, machine-readable tables, and probabilistic results.

This dataset provides the reproducibility package for the manuscript “PayoffAlpha: A Knowledge-Based Evidence-Governance Architecture for Adaptive Discovery Systems.” It contains deterministic Python scripts, locked configurations, derived non-price ledgers, boundary-state decisions, mutation and threshold tests, fair MCDM comparisons, performance benchmarks, and robustness evidence. Licensed Tushare source market data are not redistributed; retrieval and reconstruction instructions are included. The validation scripts require Python 3.11 or later and no third-party packages.

This dataset contains data from a population-based cross-sectional study investigating the associations between bullying and cyberbullying victimization and individual and contextual factors among male adolescents. The data were obtained from the 2019 National School Health Survey (Pesquisa Nacional de Saúde do Escolar – PeNSE), conducted by the Brazilian Institute of Geography and Statistics (IBGE). The study population comprises male students aged 13 to 17 years enrolled in Brazilian schools. The dataset includes sociodemographic characteristics, mental health indicators, alcohol and other substance use, experiences related to bullying and cyberbullying, and use of health services. Statistical analyses accounted for the complex sampling design of the PeNSE, including weighted descriptive analyses, tests of association, and logistic regression models. Odds ratios (ORs) and 95% confidence intervals (95% CIs) were estimated. The data are secondary, publicly available, and anonymized. The original data were obtained from the IBGE and contain no directly identifiable information about the participants. The dataset provided here contains the variables and analytical information used in the study and is intended to support transparency, reproducibility, and further research.

This dataset is collected from Bilibili, a mainstream UGC video platform in China, using Python crawlers and official open APIs. The final dataset consists of 353,501 video records created by 25,147 unique content creators, with multi-layered observational indicators covering both creator-level and video-level attributes, designed to empirically explore the correlations between creator-initiated interactive widgets and two core user engagement outcomes: online tipping (video coins) and comment-based content co-creation. The data collection process starts with generating billions of random user IDs to verify valid platform accounts. We only retain creators who have published at least one video, and remove invalid samples including removed videos, paywall-locked content and records with missing features. The core independent variables are two binary dummy variables representing affective interaction and cognitive interaction, operationalized by Bilibili’s built-in triple-interaction pop-up and voting pop-up respectively. The dependent variables are the number of coins and comments each video receives, log-transformed to mitigate skewness and heteroskedasticity. Multi-dimensional control variables are included: video-level covariates cover video resolution, publication duration, video length and total views; creator-level covariates include total upload quantity, gender dummy variables, follower count, account level, VIP membership and official verification tier. We also label videos as entertainment or knowledge categories to test content-type moderation, and add a grouping indicator to identify creators who deployed interactive widgets for cross-content spillover analysis.

1

Supplementary dataset 1 contains raw numerical data for all main and supplementary figures, plus information for how many mice were used per cage, sex of the mice, and how many independent experiments were conducted. Raw_image_files_Figure_1.zip contains photographs used in Figure 1. Raw reads for all strains sequenced in this study can also be found here.