Retrieval

The Natural Questions corpus is a question answering dataset containing 307,373 training examples, 7,830 development examples, and 7,842 test examples. Each example is comprised of a google.com query and a corresponding Wikipedia page. Each Wikipedia page has a passage (or long answer) annotated on the page that answers the question and one or more short spans from the annotated passage containing the actual answer. The long and the short answer annotations can however be empty. If they are both empty, then there is no answer on the page at all. If the long answer annotation is non-empty, but the short answer annotation is empty, then the annotated passage answers the question but no explicit short answer could be found. Finally 1% of the documents have a passage annotated with a short answer that is “yes” or “no”, instead of a list of short spans. Source: A BERT Baseline for the Natural Questions Image Source: https://paperswithcode.com/paper/natural-questions-a-benchmark-for-question/

The MS MARCO (Microsoft MAchine Reading Comprehension) is a collection of datasets focused on deep learning in search. The first dataset was a question answering dataset featuring 100,000 real Bing questions and a human generated answer. Over time the collection was extended with a 1,000,000 question dataset, a natural language generation dataset, a passage ranking dataset, keyphrase extraction dataset, crawling dataset, and a conversational search. Source: https://microsoft.github.io/msmarco/ Image Source: https://arxiv.org/pdf/1809.08267.pdf

HotpotQA is a question answering dataset collected on the English Wikipedia, containing about 113K crowd-sourced questions that are constructed to require the introduction paragraphs of two Wikipedia articles to answer. Each question in the dataset comes with the two gold paragraphs, as well as a list of sentences in these paragraphs that crowdworkers identify as supporting facts necessary to answer the question. A diverse range of reasoning strategies are featured in HotpotQA, including questions involving missing entities in the question, intersection questions (What satisfies property A and property B?), and comparison questions, where two entities are compared by a common attribute, among others. In the few-document distractor setting, the QA models are given ten paragraphs in which the gold paragraphs are guaranteed to be found; in the open-domain fullwiki setting, the models are only given the question and the entire Wikipedia. Models are evaluated on their answer accuracy and explainability, where the former is measured as overlap between the predicted and gold answers with exact match (EM) and unigram F1, and the latter concerns how well the predicted supporting fact sentences match human annotation (Supporting Fact EM/F1). A joint metric is also reported on this dataset, which encourages systems to perform well on both tasks simultaneously. Source: Answering Complex Open-domain Questions Through Iterative Query Generation Image Source: Yang et al

DBpedia (from "DB" for "database") is a project aiming to extract structured content from the information created in the Wikipedia project. DBpedia allows users to semantically query relationships and properties of Wikipedia resources, including links to other related datasets. Source: https://en.wikipedia.org/wiki/DBpedia

FEVER is a publicly available dataset for fact extraction and verification against textual sources. It consists of 185,445 claims manually verified against the introductory sections of Wikipedia pages and classified as SUPPORTED, REFUTED or NOTENOUGHINFO. For the first two classes, systems and annotators need to also return the combination of sentences forming the necessary evidence supporting or refuting the claim. The claims were generated by human annotators extracting claims from Wikipedia and mutating them in a variety of ways, some of which were meaning-altering. The verification of each claim was conducted in a separate annotation process by annotators who were aware of the page but not the sentence from which original claim was extracted and thus in 31.75% of the claims more than one sentence was considered appropriate evidence. Claims require composition of evidence from multiple sentences in 16.82% of cases. Furthermore, in 12.15% of the claims, this evidence was taken from multiple pages. Source: FEVER: a large-scale dataset for Fact Extraction and VERification

Outside Knowledge Visual Question Answering (OK-VQA) includes more than 14,000 questions that require external knowledge to answer. Source: OK-VQA: A Visual Question Answering Benchmark Requiring External Knowledge Image Source: https://okvqa.allenai.org/

The task of PubMedQA is to answer research questions with yes/no/maybe (e.g.: Do preoperative statins reduce atrial fibrillation after coronary artery bypass grafting?) using the corresponding abstracts. PubMedQA has 1k expert labeled, 61.2k unlabeled and 211.3k artificially generated QA instances. Source: PubMedQA Image Source: https://arxiv.org/pdf/1909.06146v1.pdf

MTEB is a benchmark that spans 8 embedding tasks covering a total of 56 datasets and 112 languages. The 8 task types are Bitext mining, Classification, Clustering, Pair Classification, Reranking, Retrieval, Semantic Textual Similarity and Summarisation. The 56 datasets contain varying text lengths and they are grouped into three categories: Sentence to sentence, Paragraph to paragraph, and Sentence to paragraph. Check the latest leaderboards at HuggingFace.

SciFact is a dataset of 1.4K expert-written claims, paired with evidence-containing abstracts annotated with veracity labels and rationales.

PopQA is an open-domain QA dataset with 14k QA pairs with fine-grained Wikidata entity ID, Wikipedia page views, and relationship type information. Source: https://paperswithcode.com/paper/when-not-to-trust-language-models Image Source: https://arxiv.org/pdf/2212.10511v1.pdf

TREC-COVID is a community evaluation designed to build a test collection that captures the information needs of biomedical researchers using the scientific literature during a pandemic. One of the key characteristics of pandemic search is the accelerated rate of change: the topics of interest evolve as the pandemic progresses and the scientific literature in the area explodes. The COVID-19 pandemic provides an opportunity to capture this progression as it happens. TREC-COVID, in creating a test collection around COVID-19 literature, is building infrastructure to support new research and technologies in pandemic search.

The Remote Sensing Image Captioning Dataset (RSICD) is a dataset for remote sensing image captioning task. It contains more than ten thousands remote sensing images which are collected from Google Earth, Baidu Map, MapABC and Tianditu. The images are fixed to 224X224 pixels with various resolutions. The total number of remote sensing images is 10921, with five sentences descriptions per image. Source: https://github.com/201528014227051/RSICD_optimal Image Source: https://github.com/201528014227051/RSICD_optimal

The Reuters-21578 dataset is a collection of documents with news articles. The original corpus has 10,369 documents and a vocabulary of 29,930 words. Source: Topic Model Based Multi-Label Classification from the Crowd

This dataset contains 21,889 outfits from polyvore.com, in which 17,316 are for training, 1,497 for validation and 3,076 for testing. Source: GitHub Image Source: https://arxiv.org/pdf/1707.05691.pdf

SciDocs evaluation framework consists of a suite of evaluation tasks designed for document-level tasks. Source: Allen Institute for AI

Quora Question Pairs (QQP) dataset consists of over 400,000 question pairs, and each question pair is annotated with a binary value indicating whether the two questions are paraphrase of each other. Source: Bilateral Multi-Perspective Matching for Natural Language Sentences

ALCE is a benchmark for Automatic LLMs' Citation Evaluation. ALCE collects a diverse set of questions and retrieval corpora and requires building end-to-end systems to retrieve supporting evidence and generate answers with citations. Source: Enabling Large Language Models to Generate Text with Citations Image Source: Enabling Large Language Models to Generate Text with Citations

A new publicly available dataset for verification of climate change-related claims. Source: CLIMATE-FEVER: A Dataset for Verification of Real-World Climate Claims

In this project, we introduce InfoSeek, a visual question answering dataset tailored for information-seeking questions that cannot be answered with only common sense knowledge. Using InfoSeek, we analyze various pre-trained visual question answering models and gain insights into their characteristics. Our findings reveal that state-of-the-art pre-trained multi-modal models (e.g., PaLI-X, BLIP2, etc.) face challenges in answering visual information-seeking questions, but fine-tuning on the InfoSeek dataset elicits models to use fine-grained knowledge that was learned during their pre-training.

Attribution, Relation, and Order (ARO) benchmark to systematically evaluate the ability of VLMs to understand different types of relationships, attributes, and order information. ARO consists of Visual Genome Attribution, to test the understanding of objects' properties; Visual Genome Relation, to test for relational understanding; and COCO-Order & Flickr30k-Order, to test for order sensitivity in VLMs. ARO is orders of magnitude larger than previous benchmarks of compositionality, with more than 50,000 test cases.

The IMAGE-CHAT dataset is a large collection of (image, style trait for speaker A, style trait for speaker B, dialogue between A & B) tuples that we collected using crowd-workers, Each dialogue consists of consecutive turns by speaker A and B. No particular constraints are placed on the kinds of utterance, only that we ask the speakers to both use the provided style trait, and to respond to the given image and dialogue history in an engaging way. The goal is not just to build a diagnostic dataset but a basis for training models that humans actually want to engage with.

The 20 Newsgroups data set is a collection of approximately 20,000 newsgroup documents, partitioned (nearly) evenly across 20 different newsgroups.

ProofNet is a benchmark for autoformalization and formal proving of undergraduate-level mathematics. The ProofNet benchmarks consists of 371 examples, each consisting of a formal theorem statement in Lean 3, a natural language theorem statement, and natural language proof. The problems are primarily drawn from popular undergraduate pure mathematics textbooks and cover topics such as real and complex analysis, linear algebra, abstract algebra, and topology. Source: ProofNet: Autoformalizing and Formally Proving Undergraduate-Level Mathematics Image Source: https://arxiv.org/pdf/2302.12433v1.pdf

DIOR-RSVG is a large-scale benchmark dataset of remote sensing data (RSVG). It aims to localize the referred objects in remote sensing (RS) images with the guidance of natural language. This new dataset includes image/expression/box triplets for training and evaluating visual grounding models. Source: RSVG: Exploring Data and Models for Visual Grounding on Remote Sensing Data Image Source: https://arxiv.org/pdf/2210.12634v1.pdf

QAMPARI is an ODQA benchmark, where question answers are lists of entities, spread across many paragraphs. It was created by (a) generating questions with multiple answers from Wikipedia's knowledge graph and tables, (b) automatically pairing answers with supporting evidence in Wikipedia paragraphs, and (c) manually paraphrasing questions and validating each answer. Source: QAMPARI: : An Open-domain Question Answering Benchmark for Questions with Many Answers from Multiple Paragraphs Image Source: https://arxiv.org/pdf/2205.12665v2.pdf

TripClick is a large-scale dataset of click logs in the health domain, obtained from user interactions of the Trip Database health web search engine. Provide: * Approximately 5.2 million user interactions * IR evaluation benchmark * Trainin data for deep learning IR models

A large-scale video dataset, featuring clips from movies with detailed captions.

xCodeEval is one of the largest executable multilingual multitask benchmarks consisting of 17 programming languages with execution-level parallelism. It features a total of seven tasks involving code understanding, generation, translation, and retrieval, and it employs an execution-based evaluation instead of traditional lexical approaches. It also provides a test-case-based multilingual code execution engine, ExecEval that supports all the programming languages in xCodeEval.

It is collection regrouping all datasets constituting the ViDoRe benchmark. It includes the test sets from different academic datasets (ArXiVQA, DocVQA, InfoVQA, TATDQA, TabFQuAD) and from datasets synthetically generated spanning various themes and industrial applications: (Artificial Intelligence, Government Reports, Healthcare Industry, Energy and Shift Project). Further details can be found on the corresponding dataset cards.

We present PeerQA, a real-world, scientific, document-level Question Answering (QA) dataset. PeerQA questions have been sourced from peer reviews, which contain questions that reviewers raised while thoroughly examining the scientific article. Answers have been annotated by the original authors of each paper. The dataset contains 579 QA pairs from 208 academic articles, with a majority from ML and NLP, as well as a subset of other scientific communities like Geoscience and Public Health. PeerQA supports three critical tasks for developing practical QA systems: Evidence retrieval, unanswerable question classification, and answer generation. We provide a detailed analysis of the collected dataset and conduct experiments establishing baseline systems for all three tasks. Our experiments and analyses reveal the need for decontextualization in document-level retrieval, where we find that even simple decontextualization approaches consistently improve retrieval performance across architectures. On answer generation, PeerQA serves as a challenging benchmark for long-context modeling, as the papers have an average size of 12k tokens.

A large-scale curated dataset of over 152 million tweets, growing daily, related to COVID-19 chatter generated from January 1st to April 4th at the time of writing. Source: A large-scale COVID-19 Twitter chatter dataset for open scientific research -- an international collaboration

The dataset contains single-shot videos taken from moving cameras in underwater environments. The first shard of a new Marine Video Kit dataset is presented to serve for video retrieval and other computer vision challenges. In addition to basic meta-data statistics, we present several insights based on low-level features as well as semantic annotations of selected keyframes. 1379 videos with a length from 2 s to 4.95 min, with the mean and median duration of each video is 29.9 s, and 25.4 s, respectively. We capture data from 11 different regions and countries during the time from 2011 to 2022.

The Belgian Statutory Article Retrieval Dataset (BSARD) is a French native corpus for studying statutory article retrieval. BSARD consists of more than 22,600 statutory articles from Belgian law and about 1,100 legal questions posed by Belgian citizens and labeled by experienced jurists with relevant articles from the corpus.

ShapeTalk contains over half a million discriminative utterances produced by contrasting the shapes of common 3D objects for a variety of object classes and degrees of similarity. The dataset provides discriminative utterances for a total of 36,391 shapes, across 30 object classes. Overall, ShapeTalk contains 73,799 distinct contexts, and a total of 536,596 utterances

ComFact is a benchmark for commonsense fact linking, where models are given contexts and trained to identify situationally-relevant commonsense knowledge from KGs. The novel benchmark, C-om-Fact, contains ∼293k in-context relevance annotations for common-sense triplets across four stylistically diverse dialogue and storytelling datasets. Source: ComFact: A Benchmark for Linking Contextual Commonsense Knowledge Image Source: https://arxiv.org/pdf/2210.12678v1.pdf

NFCorpus is a full-text English retrieval data set for Medical Information Retrieval. It contains a total of 3,244 natural language queries (written in non-technical English, harvested from the NutritionFacts.org site) with 169,756 automatically extracted relevance judgments for 9,964 medical documents (written in a complex terminology-heavy language), mostly from PubMed.