Reading Comprehension

Contains two different types: cloze-style reading comprehension and user query reading comprehension, associated with large-scale training data as well as human-annotated validation and hidden test set. Source: Dataset for the First Evaluation on Chinese Machine Reading Comprehension

Tasks Our shared task has three subtasks. Subtask 1 and 2 focus on evaluating machine learning models' performance with regard to two definitions of abstractness (Spreen and Schulz, 1966; Changizi, 2008), which we call imperceptibility and nonspecificity, respectively. Subtask 3 aims to provide some insights to their relationships. • Subtask 1: ReCAM-Imperceptibility Concrete words refer to things, events, and properties that we can perceive directly with our senses (Spreen and Schulz, 1966; Coltheart 1981; Turney et al., 2011), e.g., donut, trees, and red. In contrast, abstract words refer to ideas and concepts that are distant from immediate perception. Examples include objective, culture, and economy. In subtask 1, the participanting systems are required to perform reading comprehension of abstract meaning for imperceptible concepts. Below is an example. Given a passage and a question, your model needs to choose from the five candidates the best one for replacing @placeholder. • Subtask 2: ReCAM-Nonspecificity Subtask 2 focuses on a different type of definition. Compared to concrete concepts like groundhog and whale, hypernyms such as vertebrate are regarded as more abstract (Changizi, 2008). • Subtask 3: ReCAM-Intersection Subtask 3 aims to provide more insights to the relationship of the two views on abstractness, In this subtask, we test the performance of a system that is trained on one definition and evaluted on the other.

A large scale analogue of Stanford SQuAD in the Russian language - is a valuable resource that has not been properly presented to the scientific community. Source: SberQuAD -- Russian Reading Comprehension Dataset: Description and Analysis See DeepPavlov link ### Model results | Model config | EM (dev) | F-1 (dev) | |------------------------------|-------------|-------------| |DeepPavlov RuBERT | 66.30+-0.24 | 84.60+-0.11 | | DeepPavlov multilingual BERT | 64.35+-0.39 | 83.39+-0.08 | | DeepPavlov R-Net | 60.62 | 80.04 |

RadQA is a radiology question answering dataset with 3074 questions posed against radiology reports and annotated with their corresponding answer spans (resulting in a total of 6148 question-answer evidence pairs) by physicians. The questions are manually created using the clinical referral section of the reports that take into account the actual information needs of ordering physicians and eliminate bias from seeing the answer context (and, further, organically create unanswerable questions). The answer spans are marked within the Findings and Impressions sections of a report. The dataset aims to satisfy the complex clinical requirements by including complete (yet concise) answer phrases (which are not just entities) that can span multiple lines.

A human-curated ChineseReading Comprehension dataset on Opinion. The questions in ReCO are opinion based queries issued to the commercial search engine. The passages are provided by the crowdworkers who extract the support snippet from the retrieved documents. Source: ReCO: A Large Scale Chinese Reading Comprehension Dataset on Opinion

CMRC 2019 is a Chinese Machine Reading Comprehension dataset that was used in The Third Evaluation Workshop on Chinese Machine Reading Comprehension. Specifically, CMRC 2019 is a sentence cloze-style machine reading comprehension dataset that aims to evaluate the sentence-level inference ability. Source: http://ymcui.com/cmrc2019/

QUASAR-S is a large-scale dataset aimed at evaluating systems designed to comprehend a natural language query and extract its answer from a large corpus of text. It consists of 37,362 cloze-style (fill-in-the-gap) queries constructed from definitions of software entity tags on the popular website Stack Overflow. The posts and comments on the website serve as the background corpus for answering the cloze questions. The answer to each question is restricted to be another software entity, from an output vocabulary of 4874 entities. Source: Quasar: Datasets for Question Answering by Search and Reading Image Source: Quasar: Datasets for Question Answering by Search and Reading

WiC: The Word-in-Context Dataset A reliable benchmark for the evaluation of context-sensitive word embeddings. Depending on its context, an ambiguous word can refer to multiple, potentially unrelated, meanings. Mainstream static word embeddings, such as Word2vec and GloVe, are unable to reflect this dynamic semantic nature. Contextualised word embeddings are an attempt at addressing this limitation by computing dynamic representations for words which can adapt based on context. Russian SuperGLUE task borrows original data from the Russe project, Word Sense Induction and Disambiguation shared task (2018) ### Task Type Reading Comprehension. Binary Classification: true/false ### Example { "idx" : 8, "word" : "дорожка", "sentence1" : "Бурые ковровые дорожки заглушали шаги", "sentence2" : "Приятели решили выпить на дорожку в местном баре", "start1" : 15, "end1" : 23, "start2" : 26, "end2" : 34, "label" : false, "gold_sense1" : 1, "gold_sense2" : 2 } ### How did we collect data? All text examples were collected from Russe original dataset, already collected by Russian Semantic Evaluation at ACL SIGSLAV. Human assessment was carried out on Yandex.Toloka. In version 2, we have manually collected in the same format testset.

The Video-based Multimodal Summarization with Multimodal Output (VMSMO) corpus consists of 184,920 document-summary pairs, with 180,000 training pairs, 2,460 validation and test pairs. The task for this dataset is generating and appropriate textual summary of an article and choosing a proper cover frame from a video accompanying the article. Source: https://github.com/yingtaomj/VMSMO

A Winograd schema is a pair of sentences that differ in only one or two words and that contain an ambiguity that is resolved in opposite ways in the two sentences and requires the use of world knowledge and reasoning for its resolution. The schema takes its name from a well-known example by Terry Winograd. The set would then be presented as a challenge for AI programs, along the lines of the Turing test. The strengths of the challenge are that it is clear-cut, in that the answer to each schema is a binary choice; vivid, in that it is obvious to non-experts that a program that fails to get the right answers clearly has serious gaps in its understanding; and difficult, in that it is far beyond the current state of the art. ### Task Type Logic and Reasoning, World knowledge. Binary Classification: true/false ### Example { "text": "Кубок не помещается в коричневый чемодан, потому что он слишком большой." "label": false, "idx": 5, "target": { "span1_text": "чемодан", "span2_text": "он слишком большой", "span1_index": 5, "span2_index": 8 }, } ### How did we collect data? All text examples were collected manually translating and adapting original Winograd dataset for Russian. Human assessment was carried out on Yandex.Toloka.

The VNHSGE (VietNamese High School Graduation Examination) dataset, developed exclusively for evaluating large language models (LLMs), is introduced in this article. The dataset, which covers nine subjects, was generated from the Vietnamese National High School Graduation Examination and comparable tests. 300 literary essays have been included, and there are over 19,000 multiple-choice questions on a range of topics. The dataset assesses LLMs in multitasking situations such as question answering, text generation, reading comprehension, visual question answering, and more by including both textual data and accompanying images. Using ChatGPT and BingChat, we evaluated LLMs on the VNHSGE dataset and contrasted their performance with that of Vietnamese students to see how well they performed. The results show that ChatGPT and BingChat both perform at a human level in a number of areas, including literature, English, history, geography, and civics education. They still have space to grow, though, especially in the areas of mathematics, physics, chemistry, and biology. The VNHSGE dataset seeks to provide an adequate benchmark for assessing the abilities of LLMs with its wide-ranging coverage and variety of activities. We intend to promote future developments in the creation of LLMs by making this dataset available to the scientific community, especially in resolving LLMs' limits in disciplines involving mathematics and the natural sciences.

BiPaR is a manually annotated bilingual parallel novel-style machine reading comprehension (MRC) dataset, developed to support monolingual, multilingual and cross-lingual reading comprehension on novels. The biggest difference between BiPaR and existing reading comprehension datasets is that each triple (Passage, Question, Answer) in BiPaR is written in parallel in two languages. BiPaR is diverse in prefixes of questions, answer types and relationships between questions and passages. Answering the questions requires reading comprehension skills of coreference resolution, multi-sentence reasoning, and understanding of implicit causality. Source: BiPaR

We present a reading comprehension challenge in which questions can only be answered by taking into account information from multiple sentences. The dataset is the first to study multi-sentence inference at scale, with an open-ended set of question types that requires reasoning skills. ### Task Type Binary classification by each answer. True/False ### Example { "id": 397, "text": "(1) Мужская сборная команда Норвегии по биатлону в рамках этапа Кубка мира в немецком Оберхофе выиграла эстафетную гонку. (2) Вторыми стали французы, а бронзу получила немецкая команда. (3) Российские биатлонисты не смогли побороться даже за четвертое место, отстав от норвежцев более чем на две минуты. (4) Это худший результат сборной России в текущем сезоне. (5) Четвёртыми в Оберхофе стали австрийцы. (6) В составе сборной Норвегии на четвёртый этап вышел легендарный Уле-Эйнар Бьорндален. (7) Впрочем, Норвегия с самого начала гонки была в числе лидеров, успешно проведя все четыре этапа. (8) За сборную России в Оберхофе выступали Иван Черезов, Антон Шипулин, Евгений Устюгов и Максим Чудов. (9) Гонка не задалась уже с самого начала: если на стрельбе из положения лежа Черезов был точен, то из положения стоя он допустил несколько промахов, в результате чего ему пришлось бежать один дополнительный круг. (10) После этого отставание российской команды от соперников только увеличивалось. (11) Напомним, что днем ранее российские биатлонистки выиграли свою эстафету. (12) В составе сборной России выступали Анна Богалий-Титовец, Анна Булыгина, Ольга Медведцева и Светлана Слепцова. (13) Они опередили своих основных соперниц - немок - всего на 0,3 секунды.", "questions": [ { "question": "На сколько секунд женская команда опередила своих соперниц?", "answers": [ { "text": "Всего на 0,3 секунды.", "label": 1 }, { "text": "На 0,3 секунды.", "label": 1 }, { "text": "На секунду.", "label": 0 }, { "text": "На 0.5 секунд.", "label": 0 } ], "idx": 0 }] } ### How did we collect data? Our challenge dataset contains ∼6k questions for +800 paragraphs across 5 different domains: * elementary school texts * news * fiction stories * fairy tales * summary of series First, we have collected all data from open sources and automatically preprocessed them, filtered only those paragraphs that corresponding to the following parameters: 1) paragraph length 2) number of NER entities 3) number of coreference relations. Afterwords we have check the correct splitting on sentences and numerate each of them. Next, in Yandex.Toloka we have generated the crowdsource task to get from tolkers information: 1) generate questions 2) generate answers 3) check that to solve every question man need more than one sentence in the text. ### Principles * We exclude any question that can be answered based on a single sentence from a paragraph. * Answers are not written in the full match form in the text. * Answers to the questions are independent from each other. Their number can distinguish.

XQA is a data which consists of a total amount of 90k question-answer pairs in nine languages for cross-lingual open-domain question answering. Source: XQA: A Cross-lingual Open-domain Question Answering Dataset

Contains 40K human judgement scores on model outputs from 6 diverse question answering datasets and an additional set of minimal pairs for evaluation. Source: MOCHA: A Dataset for Training and Evaluating Generative Reading Comprehension Metrics

OneStopQA provides an alternative test set for reading comprehension which alleviates these shortcomings and has a substantially higher human ceiling performance. Source: STARC: Structured Annotations for Reading Comprehension

Russian reading comprehension with Commonsense reasoning (RuCoS) is a large-scale reading comprehension dataset that requires commonsense reasoning. RuCoS consists of queries automatically generated from CNN/Daily Mail news articles; the answer to each query is a text span from a summarizing passage of the corresponding news. The goal of RuCoS is to evaluate a machine`s ability of commonsense reasoning in reading comprehension. Example {'source': 'Lenta', 'passage': { 'text': 'Мать двух мальчиков, брошенных отцом в московском аэропорту Шереметьево, забрала их. Об этом сообщили ТАСС в пресс-службе министерства образования и науки Хабаровского края. Сейчас младший ребенок посещает детский сад, а старший ходит в школу. В учебных заведениях с ними по необходимости работают штатные психологи. Также министерство социальной защиты населения рассматривает вопрос о бесплатном оздоровлении детей в летнее время. Через несколько дней после того, как Виктор Гаврилов бросил своих детей в аэропорту, он явился с повинной к следователям в городе Батайске Ростовской области.\n@context\nБросившего детей в Шереметьево отца задержали за насилие над женой\n@context\nРоссиянина заподозрили в истязании брошенных в Шереметьево детей\n@context\nОставивший двоих детей в Шереметьево россиянин сам пришел к следователям', 'entities': [ {'start': 60, 'end': 71, 'text': 'Шереметьево'}, {'start': 102, 'end': 106, 'text': 'ТАСС'}, {'start': 155, 'end': 172, 'text': 'Хабаровского края'}, {'start': 470, 'end': 485, 'text': 'Виктор Гаврилов'}, {'start': 563, 'end': 571, 'text': 'Батайске'}, {'start': 572, 'end': 590, 'text': 'Ростовской области'}, {'start': 620, 'end': 631, 'text': 'Шереметьево'}, {'start': 725, 'end': 736, 'text': 'Шереметьево'}, {'start': 778, 'end': 789, 'text': 'Шереметьево'} ] }, 'qas': [ { 'query': '26 января @placeholder бросил сыновей в возрасте пяти и семи лет в Шереметьево.', 'answers': [ {'start': 470, 'end': 485, 'text': 'Виктор Гаврилов'} ], 'idx': 0 } ], 'idx': 0 } ### How did we collect data? All text examples were collected from open news sources, then automatically filtered with QA systems to prevent obvious questions to infiltrate the dataset. The texts were then filtered by IPM frequency of the contained words and, finally, manually reviewed.

Taiga is a corpus, where text sources and their meta-information are collected according to popular ML tasks. Each text in corpus is represented in plain text and with morphological and syntactic annotation (UDPipe, homonymy resolved automatically) + has metainformation - date, theme, authorship, text difficulcy…etc (depending on source) By now, about 5 billions of words are 77% literary texts (33 literary magazines), 19% of naive poetry, 2% of news (4 popular sites) and 2% of other (popular science, culture mags, social networks, amateur poems and prose), with documentation available. Segments Info | Genres | Tokens, millions | % | |------------------|------------------|-----| | News | 92 | 1.5 | | Literary Texts | 4605 | 76 | | Special datasets | 2.5 | 0.5 | | Social media | 80 | 1.5 | | Subtitles | 101 | 1.5 | | Poems | 1130 | 19 | **Annotation Example ** (CONLL-u): # newdoc # newpar # sent_id = 1 # 2003Armeniya.xml 1 # text = В советский период времени число ИТ- специалистов в Армении составляло около десяти тысяч. # sent_id = 1 1 В в ADP _ _ 3 case 3:case _ 2 советский советский ADJ _ Animacy=Inan|Case=Acc|Degree=Pos|Gender=Masc|Number=Sing 3 amod 3:amod _ 3 период период NOUN _ Animacy=Inan|Case=Acc|Gender=Masc|Number=Sing 11 obl 11:obl _ 4 времени время NOUN _ Animacy=Inan|Case=Gen|Gender=Neut|Number=Sing 3 nmod 3:nmod _ 5 число число NOUN _ Animacy=Inan|Case=Acc|Gender=Neut|Number=Sing 11 obj 11:obj _ 6 ИТ ит PROPN _ Animacy=Inan|Case=Nom|Gender=Neut|Number=Sing 8 compound 8:compound SpaceAfter=No 7 - - PUNCT _ _ 6 punct 6:punct _ 8 специалистов специалист NOUN _ Animacy=Anim|Case=Gen|Gender=Masc|Number=Plur 5 nmod 5:nmod _ 9 в в ADP _ _ 10 case 10:case _ 10 Армении армения PROPN _ Animacy=Inan|Case=Loc|Gender=Fem|Number=Sing 5 nmod 5:nmod _ 11 составляло составлять VERB _ Aspect=Imp|Gender=Neut|Mood=Ind|Number=Sing|Tense=Past|VerbForm=Fin|Voice=Act 0 root 0:root _ 12 около около ADP _ _ 14 case 14:case _ 13 десяти десять NUM _ Case=Gen 14 nummod 14:nummod _ 14 тысяч тысяча NOUN _ Animacy=Inan|Case=Gen|Gender=Fem|Number=Plur 11 nsubj 11:nsubj SpaceAfter=No 15 . . PUNCT _ _ 14 punct 14:punct _

UIT-ViNewsQA is a new corpus for the Vietnamese language to evaluate healthcare reading comprehension models. The corpus comprises 22,057 human-generated question-answer pairs. Crowd-workers create the questions and their answers based on a collection of over 4,416 online Vietnamese healthcare news articles, where the answers comprise spans extracted from the corresponding articles. Source: New Vietnamese Corpus for Machine ReadingComprehension of Health News Articles

A challenging machine comprehension corpus with multiple-choice questions, intended for research on the machine comprehension of Vietnamese text. This corpus includes 2,783 multiple-choice questions and answers based on a set of 417 Vietnamese texts used for teaching reading comprehension for 1st to 5th graders. Answers may be extracted from the contents of single or multiple sentences in the corresponding reading text. Source: ViMMRC

ConcurrentQA is a textual multi-hop QA benchmark to require concurrent retrieval over multiple data-distributions (i.e. Wikipedia and email data). The dataset follow the exact same schema and design as HotpotQA. The data set is downloadable here: https://github.com/facebookresearch/concurrentqa. It also contains model and result analysis code. This benchmark can also be used to study privacy when reasoning over data distributed in multiple privacy scopes --- i.e. Wikipedia in the public domain and emails in the private domain. The following is a blog post about the benchmark: https://ai.facebook.com/blog/building-systems-to-reason-securely-over-private-data/

This dataset is constructed and based on the online free-access fictions that are tagged with sci-fi, urban novel, love story, youth, etc. It is used for Writing Polishment with Smile (WPS) a task that aims to polish plain text with similes. All similes are extracted by rich regular expression, and the extraction precision is estimated as 92% by labelling 500 random extracted samples. It contains 5M samples for training and 2.5k for validation and test respectively. Source: https://github.com/mrzjy/writing-polishment-with-simile

The DeepMind Q&A Dataset consists of two datasets for Question Answering, CNN and DailyMail. Each dataset contains many documents (90k and 197k each), and each document companies on average 4 questions approximately. Each question is a sentence with one missing word/phrase which can be found from the accompanying document/context.

Shmoop Corpus is a dataset of 231 stories that are paired with detailed multi-paragraph summaries for each individual chapter (7,234 chapters), where the summary is chronologically aligned with respect to the story chapter. From the corpus, a set of common NLP tasks are constructed, including Cloze-form question answering and a simplified form of abstractive summarization, as benchmarks for reading comprehension on stories. Source: Shmoop Corpus

The UIT-ViWikiQA is a dataset for evaluating sentence extraction-based machine reading comprehension in the Vietnamese language. The UIT-ViWikiQA dataset is converted from the UIT-ViQuAD dataset, consisting of 23,074 question-answers based on 5,109 passages of 174 Vietnamese articles from Wikipedia.

A dataset containing 2,221 questions from matriculation exams for twelfth grade in various subjects -history, biology, geography and philosophy-, and 412 additional questions from online quizzes in history. Source: Beyond English-Only Reading Comprehension: Experiments in Zero-Shot Multilingual Transfer for Bulgarian

CodeQueries Benchmark dataset consists of instances of semantic queries, code context and code spans in the context corresponding to the semantic queries. The dataset can be used in experiments involving semantic query comprehension with an extractive question-answering methodology over code. More details can be found in the paper.

ExpMRC is a benchmark for the Explainability evaluation of Machine Reading Comprehension. ExpMRC contains four subsets of popular MRC datasets with additionally annotated evidences, including SQuAD, CMRC 2018, RACE+ (similar to RACE), and C3, covering span-extraction and multiple-choice questions MRC tasks in both English and Chinese.

A dataset of around 2 million examples for machine reading-comprehension. Source: Building Large Machine Reading-Comprehension Datasets using Paragraph Vectors

MAUD is an expert-annotated merger agreement reading comprehension dataset based on the American Bar Association's 2021 Public Target Deal Points study, where lawyers and law students answered 92 questions about 152 merger agreements. With over 39,000 examples and 47,000 total annotations, it is the largest expert-annotated legal reading comprehension dataset in the English language, as well as the first expert-annotated merger agreement dataset.

OIR is a financial-domain dataset of the outbound intent recognition task. It aims to identify the intent of customer response in the outbound call scenario.

Large scale machine reading comprehension dataset in Urdu language.

A newly developed public dataset and the task of multiple property extraction. It uses the same data as WikiReading but does not inherit its predecessor's identified disadvantages. Source: From Dataset Recycling to Multi-Property Extraction and Beyond

X-WikiRE is a new, large-scale multilingual relation extraction dataset in which relation extraction is framed as a problem of reading comprehension to allow for generalization to unseen relations. Source: X-WikiRE: A Large, Multilingual Resource for Relation Extraction as Machine Comprehension

Dataset for our paper Disambiguation-Centric Finetuning Makes Enterprise Tool-Calling LLMs More Realistic and Less Risky which includes 5000 enterprise tools and the corresponding dialogues generated using DiaFORGE UTC data engine. The dataset is generated with the data generation engine. The engine simulates a user agent and an assistant agent in a dialogue, where the user agent has a persona and the assistant agent has access to a set of tools. Detailed information about the data generation process can be found in the paper. Each entry in the dataset contains: - seed: Gold tool for the generated dialogue. - user_persona: Persona of the simulated user agent. - messages: List of messages in the dialogue, where each message is a dictionary containing: - distractor_tools: List of distractor tools that are not the gold tool but are relevant to the dialogue. These tools are used by user agent to generate utterances that are hard to disambiguate from the gold tool. - retrieved_tools: List of tools retrieved by the assistant agent and used by the assistant agent in order to ask clarifying questions to the user agent.

IDK-MRC is an Indonesian Machine Reading Comprehension (MRC) dataset consists of more than 10K questions in total with over 5K unanswerable questions with diverse question types. Source: IDK-MRC: Unanswerable Questions for Indonesian Machine Reading Comprehension Image Source: https://arxiv.org/pdf/2210.13778v1.pdf