Loading...
Loading...
News corpora, social media analysis, movie/music metadata, sports data, cultural datasets, misinformation
11,261 datasets
Movie Reviews is a synthetic text dataset generated using the DataDreamer tool. The dataset card indicates it was created for attributed prompts and was last updated on February 1, 2024. Its specific content and scale are not detailed in the provided metadata.
PMData combines traditional lifelogging with sports activity logging. The dataset was created by aai530-group6 and published in a 2019 ACM paper. It enables analysis of how sports data relates to everyday health metrics like weight and sleep.
708,241 English language news articles collected from global news sites between January 2017 and December 2019. The collection is derived from the Common Crawl CC-NEWS bucket and processed using the news-please extraction tool to provide structured text and metadata.
Bulgarian news articles from 377 distinct sources covering domains like politics, tips&tricks, and general facts. It was curated for the 'Hack the Fake News' hackathon to facilitate research into clickbait and questionable factuality within the Bulgarian media landscape.
1,800 annotated tweets categorized into positive and negative sentiment classes. The collection features text written in both Modern Standard Arabic (MSA) and the specific Jordanian dialect.
1.3 million news articles and summaries across 38 major publications. Metadata includes publication date and URL, alongside quantitative metrics like extractive density and compression ratios.
12,800 short statements labeled for truthfulness, sourced from PolitiFact.com's API. Each statement was evaluated by a PolitiFact editor, and the label distribution is relatively balanced, with instances for most labels ranging from 2,063 to 2,638 cases. The dataset was uploaded by 'ucsbai' and last updated on Hugging Face in January 2024.
5,000,000 questions and 3,000,000 answers collected from Google search autocomplete and answer boxes. The dataset features naturalistic, short-form queries paired with diverse response formats including short text, long text, and structured collections.
News articles and human-written summaries from newser.com categorized for multi-document summarization tasks. Each record pairs a professionally edited `summary` with a `document` field containing multiple source articles concatenated using a '|||||' delimiter.
280,000 user reviews and code metrics for 395 Android applications across 23 categories from the F-Droid repository. The collection spans approximately 600 app versions, linking user feedback to specific software iterations and evolution patterns.
A collection of French news articles across 5 main categories and 8 subcategories scraped from the Orange Actu website between 2011 and 2020. It features professionally written single-sentence titles and brief abstracts for each article, supporting two distinct summarization tasks.
A text dataset prepared for siamese network fine-tuning, intended to distinguish between legal contract text and general text from books, news, and reviews. The dataset was created by polestarllp and was last updated on Hugging Face on February 8, 2024. Its specific size and row count are not provided in the metadata.
The Stanford Sentiment Treebank contains 11,855 single sentences extracted from movie reviews, each annotated for sentiment. It includes 215,154 unique phrases from parse trees, enabling compositional sentiment analysis. The dataset was created by StanfordNLP based on earlier work by Pang and Lee.
67,889 Thai news articles from Prachathai.com categorized into 12 curated tags for multi-label classification. The collection spans 14 years of reporting from 2004 to 2018 and filters out entries with fewer than 500 characters to ensure text quality.
200,000 French-language movie reviews from Allociné.fr categorized into 100,000 positive and 100,000 negative sentiment labels. The collection spans user reviews from 2006 to 2020 and is structured into training, validation, and test splits.
12,800 human-labeled short statements from politifact.com categorized into six truthfulness levels. Each record pairs a political statement with a professional editor's verdict and a detailed analysis report justifying the label.
Holding between 100,000 and 1,000,000 English business reviews extracted from the 2015 Yelp Dataset Challenge. Created by Yelp, the data pairs review text with sentiment labels to support research in text classification.
8,364 Arabic restaurant reviews from qaym.com labeled for binary sentiment analysis. Each entry includes the raw review text and a corresponding polarity label of 0 or 1.
100,000 news items categorized into four topics—economy, microsoft, obama, and palestine—collected over an eight-month period from November 2015 to July 2016. The data includes social feedback metrics from Facebook, Google+, and LinkedIn to support predictive modeling and sentiment analysis.
Filtered data from 14 subreddits including 'AskAcademia', 'askscience', 'explainlikeimfive', and 'changemyview', uploaded by user euclaise. The dataset was last updated on 2024-01-19 20:59:05. It likely contains question-and-answer style text from communities focused on academic, technical, and philosophical discussion.