The Webis-Dataset-Reviews-21 corpus comprises the curated list of 13,372 NLP-related datasets and their 539,411 mentions extracted from all the publications available in ACL Anthology corpus.

The dataset contains argument pairs which are sampled from args.me dataset and cover two topics: abortion and gay marriage. The dataset is used in the same side stance classification challenge which consists of two experiments (cross-topics and within topics)

The Webis-SameSide-21 dataset is a resampled dataset based on the Same Side Stance Classification shared task dataset.

The Webis Bias Flipper 2018 (Webis-Bias-Flipper-2018) comprises 2781 events from allsides.com as of June 1st, 2012 till February 10, 2018. For each event, the title, the summary, all news portals belonging to the event, and the links to the news portals with respective bias were recorded. After that, we crawled the news portals with the given links to retrieve their headlines and the content of all articles, because the content is not provided on allsides.com. For each event we collected the corresponding news articles. A total of 6458 news articles are collected.

This corpus is outdated. Please use its successors PAN-WVC-10 and PAN-WVC-11.

This corpus is an extension of the Ambient data set created by Carpineto and Romano. For each subtopic, the websites of the given URLs were downloaded (if accessible). Those documents are named as the original documents, for example, 1/1.4/1.3.html. Each subtopic was then manually enriched to ten documents with websites retrieved by Google (for example, 1/1.1/g00.html - 'g' for Google, 00 for the first Google result). Some subtopics could not be sufficently enriched and were discarded. Moreover, some subtopics were duplicates or not interpretable and were also discarded.

The Webis-Editorials-16 corpus is a novel corpus with 300 news editorials evenly selected from three diverse online news portals: Al Jazeera, Fox News, and The Guardian. The aim of the corpus is to study (1) the mining and classification of fine-grained types of argumentative discourse units and (2) the analysis of argumentation strategies pursued in editorials to achieve persuasion. To this end, each editorial contains manual type annotations of all units that capture the role that a unit plays in the argumentative discourse, such as assumption or statistics. The corpus consists of 14,313 units of six different types, each annotated by three professional annotators from the crowdsourcing platform upwork.com.

The Webis-WebSeg-20 dataset comprises 42,450 crowdsourced segmentations for 8,490 web pages from the Webis-Web-Archive-17. Segmentations were fused from the segmentations of five crowd workers each.

The web genre corpus 2004 (Genre-KI-04) is designed for the evaluation of techniques for genre classification.It consists of 1239 web documents classified into 8 genres and basic meta data for each of the files.

The Webis-SameSentiment-21 dataset is a collection of sentiment review pairs for Same Sentiment Classification. The dataset only contains the pair ids (business and review id) to allow recreation of the dataset. The actual review text has to be downloaded from Yelp.

Extension of the Spoken Conversational Search dataset by Trippas et al. with annotations for types of meta-information mentioned in each turn. For each of the 1044~dialog turns in the dataset, the turn was examined to check if it communicates meta-information on the (desired) results. No distinction was made as to whether the seeker or the provider communicates meta-information, and the annotation was not restricted to instances in which meta-information has a noticeable effect on the search process.

Dataset, a taxonomy description, and the annotation interface for the paper "Identifying the Human Values behind Arguments".

The corpus contains 136,996 (argumentative text, conclusion) pairs for the task of informative conclusion generation. For each argument in the corpus, argumentative knowledge such as discussion topic, conclusion targets and argument aspects are provided.

The Webis-Web-Archive-17 comprises a total of 10,000 web page archives from mid-2017 that were carefully sampled from the Common Crawl to involve a mixture of high-ranking and low-ranking web pages. The dataset contains the web archive files, HTML DOM, and screenshots of each web page, as well as per-page annotations of visual web archive quality.

TexBiG (from the German Text-Bild-Gefüge, meaning Text-Image-Structure) is a document layout analysis dataset for historical documents in the late 19th and early 20th century. The dataset provides instance segmentation (bounding boxes and polygons/masks) annotations for 19 different classes with more then 52.000 instances. Annotations are manually annotated by experts and evaluated with Krippendorff's Alpha, for each document image are least two different annotators have labeled the document. The dataset uses the common COCO-JSON format.

The Webis Clickbait Corpus 2017 (Webis-Clickbait-17) comprises a total of 38,517 Twitter posts from 27 major US news publishers. In addition to the posts, information about the articles linked in the posts are included. The posts had been published between November 2016 and June 2017. To avoid publisher and topical biases, a maximum of ten posts per day and publisher were sampled. All posts were annotated on a 4-point scale [not click baiting (0.0), slightly click baiting (0.33), considerably click baiting (0.66), heavily click baiting (1.0)] by five annotators from Amazon Mechanical Turk. A total of 9,276 posts are considered clickbait by the majority of annotators. In terms of its size, this corpus outranges the Webis Clickbait Corpus 2016 by one order of magnitude. The corpus is divided into two logical parts, a training and a test dataset. The training dataset has been released in the course of the Clickbait Challenge and a download link is provided below. To allow for an objective evaulatuion of clickbait detection systems, the test dataset is available only through the Evaluation-as-a-Service platform TIRA at the moment. On TIRA, developers can deploy clickbait detection systems and execute them against the test dataset. The performance of the submitted systems can be viewed on the TIRA page of the Clickbait Challenge.

The Webis Search Mission Corpus 2012 (Webis-SMC-12) contains 8840 search engine interactions of 127 users. Two human annotators divided these interactions into 2881 logical sessions and 1378 missions. Cases where the annotators did not agree initially were discussed to reach a consensus.

The Webis-Context-SciSumm-2023 is a large scale dataset suitable for studying contextualized summarization of scientific papers. The corpus contains approximately 540K computer science papers encompassing 4.6M citation texts and relevant information for these citations from the cited papers.

The Webis Query-Task-Mapping Corpus 2019 (Webis-QTM-19) comprises three benchmark datasets on the query-task-mapping problem, which consists of finding the correct task for a new query in a given task-split background query log.

The Webis Argument Quality Corpus 2020 (Webis-ArgQuality-20) contains 1271 arguments spanning 20 topics with scores for rhetorical, logical, dialectical, and overall quality as well as topical relevance. Quality scores are inferred from a total of 42k pairwise judgments. Arguments are sourced from the args.me corpus.