Find research datasets worth reusing
Search datasets from major research repositories and use ShareScore to quickly assess how well each record supports discovery, access, and reuse.
132
datasets available to search
ShareScore release 0.9.0
Dataset results
132 results for “CHATGPT”
ChatGPT's performance in dentistry and allergy-immunology assessments: a comparative study
<p>Data on ChatGPT 3's and ChatGPT 4's performance on self-assessment questions for dentistry (SFLEDM) and allergy and clinical immunology (EEAACI), sourced from the University of Bern’s Institute for Medical Education platform.</p>
Anexo del artículo «Desafiando el estado del arte en el uso de ChatGPT en educación en el año 2023»
<p><span>Este documento es el anexo que contiene la lista completa de las referencias revisadas en el artículo citado a continuación.</span></p> <p><span>Saz-Pérez, F., y Pizà-Mir, B. (2024). Desafiando el estado del arte en el uso de ChatGPT en educación en el año 2023. <em>REIRE Revista d’Innovació i Recerca en Educació</em>, <em>17</em>(1), 1-13. <a href="https://doi.org/10.1344/reire.44018">https://doi.org/10.1344/reire.44018</a> </span></p>
Just another copy and paste? Comparing the security vulnerabilities of ChatGPT generated code and StackOverflow answers
<div>Supplemental material for the paper "Just another copy and paste? Comparing the security vulnerabilities of ChatGPT generated code and StackOverflow answers" published in DLSP 2024.</div> <div> <div> </div> </div>
How ChatGPT defines risk - Data
<p>This is the dataset associated to the paper <em>How ChatGPT defines risk?</em> published on Journal of Risk Research. Here is the paper abstract:</p> <p> </p> <p>Large Language Models (LLMs) are prominent AI tools potentially useful in various applications involving natural language interactions and exchanges of information with human users. These models, however, have the potential to spread misinformation and misconceptions, especially when used by individuals lacking the necessary expertise to critically assess their output. ChatGPT, developed by OpenAI, is a public available LLM and one of the most renowned. The paper explores whether the information it provides about the concept of risk and some related basic notions can be considered sufficiently correct. Specifically, ChatGPT was first utilized to build a glossary of basic concepts for the risk analysis field, modeled after that of the Society for Risk Analysis (SRA). The model was then used to assess the quality of the generated entries for clarity, precision, completeness and presence of examples, and to compare such entries to those of the SRA glossary again for quality and for semantic similarity. Independent ChatGPT user sessions were used throughout so as to avoid influencing one output by the previous ones. The results suggest that, while the SRA and ChatGPT entries may differ in focus and scopes, they share a common core and do not conflict in substantial ways. Therefore, we can develop the working hypothesis that ChatGPT does not promote major misconceptions as to the foundational definitions about risk, at least with respect to those provided by the SRA</p>
Using ChatGPT to Generate Gendered Language
<p>Dataset for Gendered Language</p>
Analysis of the impact of ChatGPT utilization on the levels of laziness and productivity
<p><span>This study used a quantitative research design to examine the impact of ChatGPT utilization on student laziness and productivity. Quantitative research allows the collection of numerical data and facilitates statistical analysis to test the proposed hypothesis and draw empirical conclusions. Data was collected through an online survey distributed to undergraduate students from various universities. The survey included questions about ChatGPT usage patterns, perceived laziness (measured through procrastination and task avoidance scales), and self-reported productivity metrics.</span></p>
Overview exoplanets for ChatGPT and Zenodo
<p> <strong>This Excel sheet is produced based on the data from the NASA Exoplanet Archive. It is Reference [14] in Seeking Evidence for the Cosmic Influx Theory (CIT) Collaborating with ChatGPT <span><a href="../records/12683899">https://zenodo.org/records/12683899</a> </span> </strong> <br><br>You find the calculations for the Preferred Distances of star systems in column N from cell 7 down. The largest planets are likely found at that preferred distance, based on NASA data. Open the Excel sheet and scroll down to row 195 to view the original data from NASA.<br><a href="http://exoplanetarchive.ipac.caltech.edu">http://exoplanetarchive.ipac.caltech.edu</a> </p> <p>In <em>'Seeking Evidence for the Cosmic Influx Theory (CIT)</em> <em>Unveiling a Universal Ether-like Energy Field spanning the vast scales of exoplanets down to the minute details of dew and rime)</em>', we find surprising confirmations from ChatGPT that many natural phenomena exemplify an influx of energy, converting the Big Bang (BB) concept into Continuous Creation (CC) This perspective represents a significant shift in how we view familiar phenomena—from rain to stardust, and from volcanoes to the spreading ocean floor.<br>Citation: Loeffen, R. (2024). Seeking Evidence for the Cosmic Influx Theory (CIT) Collaborating with ChatGPT<br><span><a href="../records/12683899">https://zenodo.org/records/12683899</a> </span> </p>
Detecting Code Smells using ChatGPT: Initial Insights
<h2><strong>Replication Package: Detecting Code Smells using ChatGPT: Initial Insights</strong></h2> <h3><br><strong>Replication Package</strong></h3> <p>This repository contains data files to replicate our study presented in the paper <em><strong>Detecting Code Smells using ChatGPT: Initial Insights</strong></em>.</p> <h3>Dataset Structure</h3> <p>We provide raw and preprocessed data in different formats: </p> <ul> <li><code>csv - the database in CSV format with comma separated columns.</code></li> <li><code>csv-semi-comma - the database in CSV format with semi-comma separated columns.</code></li> <li><code>html - the database in HTML format to ease the visualization of the dataset. </code></li> <li><code>sql - the original SQL (schema and data) of the database used in this study.</code></li> </ul> <p>Each compressed directory contains a README file explaining which information each column includes. </p> <h3>Dataset Overview</h3> <p>A. <code>sql</code> directory consists of two tables: </p> <p>1. The table <code>tb_unique_bad_smell</code> contains information we collect from ChatGPT. </p> <p><br><code>CREATE TABLE public.tb_unique_bad_smell (</code><br><code> id_unique_bad_smell integer NOT NULL,</code><br><code> id_bad_smell bigint,</code><br><code> id_source_code bigint,</code><br><code> chat_gpt_response text,</code><br><code> question text,</code><br><code> badsmell_base text,</code><br><code> bad_smell_gpt text,</code><br><code> found_any boolean,</code><br><code> valid_bad_smell boolean,</code><br><code> bad_smell_in_base boolean,</code><br><code> bad_smell_not_in_the_base text,</code><br><code> bad_smell_not_found text,</code><br><code> index integer,</code><br><code> index_base integer,</code><br><code> url_github text,</code><br><code> id_base bigint,</code><br><code> dt_insertion timestamp without time zone,</code><br><code> nr_question smallint</code><br><code>);</code><br><br></p> <p><code>id_source_code</code> - integer identifier when we imported the original dataset to our database.<br><code>chat_gpt_response</code> - contains ChatGPT's response to our prompts. <br><code>question</code> - This column contains all the prompts we submitted to ChatGPT. Each question includes the prompt under evaluation and the source code we were interested in evaluating.<br><code>badsmell_base</code> - code smells assigned in the original dataset. <br><code>bad_smell_gpt</code> - code smells identified by the ChatGPT. We extracted these smells from `chat_gpt_response`.<br><code>found_any</code> - a boolean field indicating if any of the smells found by ChatGPT are in the dataset. In other words, if ChatGPT answered yes, finding any smell (even if it is not in the original dataset).<br><code>valid_bad_smell</code> - text field containing the smells in the original dataset that the ChatGPT identified.<br><code>bad_smell_in_base</code> - a boolean field indicating if the smells found by ChatGPT are in the original dataset.<br><code>bad_smell_not_in_the_base</code> - text field containing the smells that ChatGPT found and they are not in the dataset. <br><code>bad_smell_not_found</code> - text field containing the smells in the dataset that ChatGPT did not detect.<br><code>index</code> and <code>index_base</code> - indexes imported from the original dataset. GitHub provided them.<br><code>url_github</code> - the GitHub URL of the source code extracted from the original dataset. <br><code>id_base</code> - id field in the original dataset.<br><code>nr_question</code> - integer field containing 1 or 2, identifying the prompt we submitted to ChatGPT. </p> <p><br>2. The table `tb_unique_source_code` contains information we imported from the original dataset we used to submit to ChatGPT and information to evaluate ChatGPT's performance. The primary data we relied on are:</p> <p><br><code>smell</code> - presents the smell assigned to the code.<br><code>severity</code> - the severity of the smells, Major, Critical, Minor. <br><code>type</code> - function or class.<br><code>code_name</code> - the full path for the smell, e.g., <code>nm_package.nm_class.nm_method</code><br><code>start_line</code> - the code where the smell starts.<br><code>end_line</code> - line in the code where the smell ends.<br><code>link</code> - the GitHub URL of the source code evaluated.</p> <p> </p>
Code on Demand: A Comparative Analysis of the Efficiency, Understandability, and Self-Correction Capability of Copilot, ChatGPT, and Gemini - Data resulting from the study
<p>Este conjunto de dados foi gerado como parte do estudo "Code on Demand: A Comparative Analysis of the Efficiency, Understandability, and Self-Correction Capability of Copilot, ChatGPT, and Gemini - Data resulting from the study". O estudo focou na avaliação do desempenho das ferramentas Copilot, ChatGPT e Gemini, utilizando problemas do LeetCode em quatro linguagens de programação: Python, Java, JavaScript e C.</p> <p>O conjunto de dados atualizado está organizado nas seguintes pastas:</p> <ol> <li> <p><strong>c_programs</strong>: Esta pasta contém os scripts Python utilizados para calcular a complexidade ciclomática e a complexidade cognitiva do código C gerado pelas ferramentas.</p> <ul> <li><code>calculate_cyclomatic_complexity.py</code>: Script para calcular a complexidade ciclomática.</li> <li><code>calculate_cognitive_complexity.py</code>: Script para calcular a complexidade cognitiva.</li> </ul> </li> <li> <p><strong>codes_suggested_by_the_tools</strong>: Esta pasta contém as sugestões de código geradas pelo Copilot, ChatGPT e Gemini para cada problema do LeetCode.</p> <ul> <li>Subpastas: <code>ChatGPT</code>, <code>Copilot</code>, <code>Gemini</code>, cada uma contendo as sugestões de código correspondentes nos formatos das linguagens.</li> </ul> </li> <li> <p><strong>complexity_of_codes</strong>: Esta pasta contém dois arquivos CSV que fornecem os resultados da análise de complexidade para o código gerado.</p> <ul> <li><code>AI analysis results table - Cognitive.csv</code>: Resultados da complexidade cognitiva do código gerado.</li> <li><code>AI analysis results table - Cyclomatic.csv</code>: Resultados da complexidade ciclomática do código gerado.</li> </ul> </li> </ol> <p>Este conjunto de dados atualizado oferece insights valiosos sobre o desempenho das ferramentas de geração de código com IA e pode ser utilizado para análises futuras ou estudos de replicação.</p>
Online supplement to manuscript: "Ability of ChatGPT to generate competent radiology reports for distal radius fracture by use of RSNA template items and integrated AO classifier." Current problems in diagnostic radiology (2023).
<p>Online supplement to manuscript: </p> <p>Bosbach, Wolfram A., Jan F. Senge, Bence Nemeth, Siti H. Omar, Milena Mitrakovic, Claus Beisbart, András Horváth, Johannes Heverhagen, and Keivan Daneshvar. "Ability of ChatGPT to generate competent radiology reports for distal radius fracture by use of RSNA template items and integrated AO classifier." <em>Current problems in diagnostic radiology</em> (2023). <a href="https://doi.org/10.1067/j.cpradiol.2023.04.001">doi.org/10.1067/j.cpradiol.2023.04.001</a></p>
ChatGPT responses to questions from Vetala_Tales-Arthur_Ryder-Twenty_Two_Goblins
<p><strong>ChatGPT (GPT-4 & GPT-3.5) responses to Vetala-questions from stories (paraphrased) selected from Twenty-Two Goblins by Arthur W. Ryder based on Vetala Tales composed in Sanskrit in the 11th century</strong></p> <p>Data for preprint<strong>: </strong><strong>Goblin's Challenge to ChatGPT: Exploring AI's Dilemma Resolution and Mentalization through Riddle Tales</strong></p> <p><strong>https://ssrn.com/abstract=4476837 </strong>OR<strong> https://dx.doi.org/10.2139/ssrn.4476837</strong></p> <p>Authors: </p> <p>Kunal Bhattacharya (ORCID: 0000-0002-4943-3814)</p> <p>Aalto University School of Science, Finland, </p> <p>kunal.bhattacharya@aalto.fi</p> <p> </p> <p>Anjan K. Nandi (ORCID: 0000-0002-7859-4596)</p> <p>Edukorus Knowledge Services, Kolkata, India</p> <p>anjanandi@gmail.com</p> <p><br> *Ry- denotes the story number in Ryder’s adaptation (J.M. Dent & Sons Ltd., London, 1917 available at https://www.gutenberg.org/ebooks/2290) </p> <p>*Pe- N. M. Penzer. The Ocean of Story, being C. H. Tawney’s translation of Somadeva’s Katha Sarit Sagara, volumes VI-VII. Chas. J. Sawyer, London, 1924-26</p> <p>Stories selected: </p> <p>Ry1, Ry2, Ry3, Ry5, Ry6, Ry7, Ry10, Ry11, Ry12, Ry13, Ry17, & Pe10 </p> <p>ChatGPT sessions conducted in the first two weeks of May, 2023.</p>
Validation of references provided by ChatGPT
<p>Validation of references provided by ChatGPT for the generation of communication strategies for the long-term involvement of participants in Citizen Science projects.<br> <br> Article: <strong>Unlocking long-term engagement with citizen science: communication strategies driven by complex thinking under an AI-assisted approach.</strong></p>
Online appendix for "Leveraging Execution Trace with ChatGPT: A Case Study on Automated Fault Diagnosis" (New Ideas and Emerging Results Track in ICSME 2023)
<p>All the prompts we prepared for ChatGPT and the fault diagnosis results</p> <ul> <li>prompt_*: Prompt for ChatGPT <ul> <li>prompt_ChatGPT_setup_*.txt: Prompts to setup ChatGPT before starting the question to ChatGPT</li> <li>The other prompts: Prompts input to ChatGPT for fault diagnosis</li> </ul> </li> <li>result_*: Response from ChatGPT</li> <li>without_trace: Prompt or result when execution trace is not entered in ChatGPT</li> <li>with_trace: Prompt or result when execution trace is entered in ChatGPT</li> </ul>
Accelerating medical education with ChatGPT: An implementation guide
<p>This dataset is a collection of prompt examples of how ChatGPT has been used to accelerate medical education. </p>
How User Language Affects Conflict Fatality Estimates in ChatGPT, Query script and dataset
<p>*both authors contributed equally</p> <p>Automated query script for automated language bias studies in GPT 3-5</p> <p>Dataset of the paper "How User Language Affects Conflict Fatality Estimates in ChatGPT" preprint available on ArXiv</p>
Supplementary information to "What does ChatGPT know about natural science and engineering?"
<p>This Excel workbook contains the survey data and data analysis from the manuscript "What does ChatGPT know about natural science and engineering?" by Schulze Balhorn et al.</p>
Microservice ChatGPT Reasoning
<p>This dataset contains:</p> <p>- The source code data extraction from TrainTicket testbench V0.1.0.</p> <p>- The PO-CCG (Persistence Operation aware Component Call Graph) data extraction of TrainTicket V0.1.0.</p> <p>- The questions/ answers and the evaluation of ChatGPT answers.</p>
ChatGPT impacts in programming education: A short list of input questions and output answers from ChatGTP
<p><strong>Five questions (in red) were addressed to ChatGTP ver3.5 between the period July 24<sup>th</sup> – July 28<sup>th</sup> 2023, in order to identify the opinion of ChatGTP in the field of computer science programming and programmers training. It is chosen to pose questions related to the main applications, advantages, disadvantages, future implications, and ethical considerations.</strong></p>
Species occurrence test set for "Using ChatGPT with Confidence for Biodiversity-Related Information Tasks" (hash://sha256/fd1c8a5304508981a02481a31dc88ad429356a26dff6f89d24fe47fc60a275a4)
Species occurrence test set described in "Using ChatGPT with Confidence for Biodiversity-Related Information Tasks", presented at TDWG 2023 in SYM05
ChatGPT Helping Advance Training for Medical Students: A Study on Self-Directed Learning Enhancement
ClinicalTrials.gov study NCT06276049. IPD Sharing: UNDECIDED. Countries: 1. Publications: 1.
ScienceDex guides
Understand access before you commit
These curated guides explain access requirements, typical timelines, costs, and reuse considerations for widely used research datasets.
Allen Brain Atlas
Allen Brain Atlas is an Allen Institute collection of brain map atlases, datasets, APIs, and analysis tools covering mouse, human, and non-human primate brain resources.
Annotated Behaviour and Observability Dataset (ABODe)
ABODe is a University of Edinburgh DataShare dataset for behavior classification in group-housed mice using home-cage video, identities, bounding boxes, ground-plate positions, and annotator labels.
DANDI Archive for NWB datasets
DANDI is a BRAIN Initiative archive for publishing and sharing neurophysiology data, including electrophysiology, optophysiology, and behavioral data packaged as NWB and related standards.
International Brain Laboratory public data
The International Brain Laboratory public data releases expose standardized mouse decision-making experiments, including Neuropixels recordings, widefield calcium imaging, behavior, and session metadata accessed through the ONE API.
OpenNeuro
OpenNeuro is a free, open platform for sharing neuroimaging datasets, with public search, dataset pages, and download paths for web, S3, DataLad, and the OpenNeuro CLI.