Index for the question “In your last contact with the municipality’s employees by telephony, how satisfied were you with the response?”. In the survey, residents have been given their answers and satisfaction on a scale from 1-5 where 1 is the worst and 5 is the best. In the analysis of the result, the grades are converted to an index from 0-100. The index is calculated by producing an average based on the different values of the grading scale. Grade 5 gives 100 points, grade 4 gives 75 points, grade 3 gives 50 points, grade 2 gives 25 points and grade 1 gives 0 points. In order to get a value, the municipality must have received at least 30 responses.

The Customer Service Response Index tracks utility performance and identifies the current level of customer service and responsiveness delivered by each utility service provider under the Commission’s jurisdiction with respect to consumer complaints filed with the Commission.

Azure Q&A Index with FAISS

This dataset contains:

Preprocessed Azure Q&A pairs (question → accepted answer) FAISS index built from SentenceTransformer all-mpnet-base-v2 embeddings Pickle file mapping questions to their corresponding answers

  🧠 Usage

You can use this dataset to build a similarity-based question-answering system.

  Load the FAISS index and associated data

from huggingface_hub import hf_hub_download import faiss import pickle

Download files… See the full description on the dataset page: https://huggingface.co/datasets/LudivineRB/azure-qa-index.

The National Drought Mitigation Center produces VegDRI in collaboration with the US Geological Survey's (USGS) Center for Earth Resources Observation and Science (EROS), and the High Plains Regional Climate Center (HPRCC), with sponsorship from the US Department of Agriculture's (USDA) Risk Management Agency (RMA). Main researchers working on VegDRI are Dr. Brian Wardlow and Dr. Tsegaye Tadesse at the NDMC, and Jesslyn Brown with the USGS, and Dr. Yingxin Gu with ASRC Research and Technology Solutions, contractor for the USGS at EROS. VegDRI maps are produced every two weeks and provide regional to sub-county scale information about drought's effects on vegetation. In 2006, VegDRI covered seven states in the Northern Great Plains (CO, KS, MT, NE, ND, SD, and WY). It expanded across eight more states in 2007 to cover the rest of the Great Plains (NM, OK, MO, and TX) and parts of the Upper Midwest (IA, IL, MN, and WI). VegDRI expanded to include the western U.S. in 2008 (WA, ID, OR, UT, CA, AZ, NV). In May 2009 VegDRI began depicting the eastern states as well, covering the entire conterminous 48-state area.

With a Global Health security (GHS) Index score of 64.8 (out of 100) in the category rapid response, Mexico is the Latin American country with the highest ability to react and mitigate the spread of an epidemic or pandemic. In comparison, Finland, the best-rated country worldwide, had a score of 70.7. The Global Health Security Index measures a country's readiness to prevent, detect and respond to biological threats.

Number of answers for the index questions “Have you experienced the answer from the municipality’s employees as clearly via email?”, “In your last contact with the municipality’s employees by telephony, how satisfied were you with the reception?” and “How easy was it to get help from the municipality with your question via telephony?”. In the survey, residents have been given their answers and satisfaction on a scale from 1-5 where 1 is the worst and 5 is the best. In the analysis of the result, the grades are converted to an index from 0-100. The index is calculated by producing an average based on the different values of the grading scale. Grade 5 gives 100 points, grade 4 gives 75 points, grade 3 gives 50 points, grade 2 gives 25 points and grade 1 gives 0 points. In order to get a value, the municipality must have received at least 30 responses.

Twelve Data is a technology-driven company that provides financial market data, financial tools, and dedicated solutions. Large audiences - from individuals to financial institutions - use our products to stay ahead of the competition and success.

At Twelve Data we feel responsible for where the markets are going and how people are able to explore them. Coming from different technological backgrounds, we see how the world is lacking the unique and simple place where financial data can be accessed by anyone, at any time. This is what distinguishes us from others, we do not only supply the financial data but instead, we want you to benefit from it, by using the convenient format, tools, and special solutions.

We believe that the human factor is still a very important aspect of our work and therefore our ethics guides us on how to treat people, with convenient and understandable resources. This includes world-class documentation, human support, and dedicated solutions.

The Yahoo! Answers topic classification dataset is constructed using 10 largest main categories. Each class contains 140,000 training samples and 6,000 testing samples. Therefore, the total number of training samples is 1,400,000 and testing samples 60,000 in this dataset. From all the answers and other meta-information, we only used the best answer content and the main category information.

The file classes.txt contains a list of classes corresponding to each label.

The files train.csv and test.csv contain all the training samples as comma-sparated values. There are 4 columns in them, corresponding to class index (1 to 10), question title, question content and best answer. The text fields are escaped using double quotes ("), and any internal double quote is escaped by 2 double quotes (""). New lines are escaped by a backslash followed with an "n" character, that is " ".

What is the Survey of Economic Conditions? Contacts located in the Seventh Federal Reserve District are asked to rate various aspects of economic conditions along a seven-point scale ranging from "large increase" to "large decrease." A series of diffusion indexes summarizing the distribution of responses is then calculated.

How are the indexes constructed? Respondents' answers on the seven-point scale are assigned a numeric value ranging from +3 to –3. Each diffusion index is calculated as the difference between the number of respondents with answers above their respective average responses and the number of respondents with answers below their respective average responses, divided by the total number of respondents. The index is then multiplied by 100 so that it ranges from +100 to −100 and will be +100 if every respondent provides an above-average answer and –100 if every respondent provides a below-average answer. Respondents with no prior history of responses are excluded from the calculation.

What do the numbers mean? Respondents' respective average answers to a question can be interpreted as representing their historical trends, or long-run averages. Thus, zero index values indicate, on balance, average growth (or a neutral outlook) for activity, hiring, capital spending, and cost pressures. Positive index values indicate above-average growth (or an optimistic outlook) on balance, and negative values indicate below-average growth (or a pessimistic outlook) on balance.

Beginning with the May 12, 2020 release, the CFSEC moved to a monthly release schedule. This release, with data for April 2020, now contains estimated monthly historical values for the CFSEC indexes, as will all future releases. For additional information on how the survey and indexes changed, see the CFSEC FAQs available here (https://www.chicagofed.org/research/data/cfsec/current-data).

Prior to April 2022, the Chicago Fed Survey of Economic Conditions was named the Chicago Fed Survey of Business Conditions (CFSBC). The name change was made to better represent the survey’s aim and base of respondents. The goal of the survey is to assess the state of the economy in the Seventh Federal Reserve District. Moreover, since the beginning of the survey, it was been filled out by both business and nonbusiness contacts.

This dataset provides a unique opportunity for NLP researchers to develop models capable of answering multiple-choice questions based on a given context paragraph. It is particularly well-suited for the development and testing of question-answering systems that can handle real-world, noisy data. Originating from grade school science content, this dataset can be utilised to create interactive tools such as a question-answering chatbot, a multiple-choice quiz game, or systems that generate multiple-choice questions for students.

Columns

The dataset is primarily composed of three files: validation.csv, train.csv, and test.csv. Each file contains the following columns:

• id: A unique identifier for each question record.
• question: The text of the question (String).
• choices: A list of multiple-choice answers for the question (List of Strings).
• answerKey: The integer index corresponding to the correct answer within the choices list.
• fact1: The first piece of supporting information (String).
• fact2: The second piece of supporting information (String).
• combinedfact: A combined piece of supporting information (String).
• formatted_question: The question text with the multiple-choice answers inserted into it (String).

Distribution

The data files are typically provided in CSV format. For the test.csv file, there are 920 unique records for the id, question, choices, answerKey, and formatted_question columns. The fact1, fact2, and combinedfact columns are noted as having 100% null values in some distributions. This is a free dataset, listed on a data marketplace with a quality rating of 5 out of 5 and is available globally. The current version is 1.0.

Usage

This dataset is ideal for: * Developing and evaluating Natural Language Processing (NLP) models for question answering. * Creating question-answering chatbots that can respond to science-based queries. * Designing multiple-choice quiz games for educational purposes. * Generating multiple-choice questions to aid student learning and assessment. * Research into handling noisy, real-world data in Q&A systems.

Coverage

The dataset's scope is global in terms of availability. Its content focuses on grade school science, making it relevant for primary and secondary education contexts. While a specific time range for data collection is not provided, the dataset was listed on 16/06/2025.

License

CC0

Who Can Use It

• NLP Researchers and Data Scientists focusing on question answering, text classification, and natural language understanding.
• Educators and Content Developers looking to create educational tools, quizzes, or automated question generation systems.
• Game Developers interested in building educational quiz games.
• Anyone working on AI and Machine Learning models that require structured question-answer pairs for training and testing.

Dataset Name Suggestions

• Grade School Science Q&A
• Educational NLP Challenge Data
• Multi-Choice Science Questions
• Woodchuck Science Quiz Dataset
• Primary/Secondary Science QA

Attributes

Original Data Source: Woodchuck (Grade School Science Multi-Choice Q&A)

After the global COVID-19 outbreak, the University of Oxford set up a Government Response Tracker, which analyzes the stringency to which governments around the world have responded to the sanitary crisis. According to the index, the French Government was easy on policies at the beginning of the year 2020, reaching a high value starting in March, just as the lockdown measures were implemented. The index value decreased in June before reaching a second high in November 2020. Vaccination programs were implemented from December 2020, and the French government introduced separate restrictions for vaccinated and non-vaccinated persons from June 2021. These separate restrictions stayed in place for just over one year, until August 2022.

This dataset is part of the bachelor thesis "Evaluating SQuAD-based Question Answering for the Open Research Knowledge Graph Completion". It was created for the finetuning of Bert Based models pre-trained on the SQUaD dataset. The Dataset was created using semi-automatic approach on the ORKG data. The dataset.csv file contains the entire data (all properties) in a tabular for and is unsplit. The json files contain only the necessary fields for training and evaluation, with additional fields (index of start and end of the answers in the abstracts). The data in the json files is split (training data) and evaluation data. We create 4 variants of the training and evaluation sets for each one of the question labels ("no label", "how", "what", "which") For detailed information on each of the fields in the dataset, refer to section 4.2 (Corpus) of the Thesis document that can be found in https://www.repo.uni-hannover.de/handle/123456789/12958. The script used to generate the dataset can be found in the public repository https://github.com/as18cia/thesis_work and https://gitlab.com/TIBHannover/orkg/nlp/experiments/orkg-fine-tuning-squad-based-models

United States SBOI: sa: CO: AE: Number Answer data was reported at 1.000 % in Mar 2025. This records an increase from the previous number of 0.000 % for Feb 2025. United States SBOI: sa: CO: AE: Number Answer data is updated monthly, averaging 1.000 % from Jan 2014 (Median) to Mar 2025, with 131 observations. The data reached an all-time high of 10.000 % in Jul 2016 and a record low of 0.000 % in Feb 2025. United States SBOI: sa: CO: AE: Number Answer data remains active status in CEIC and is reported by National Federation of Independent Business. The data is categorized under Global Database’s United States – Table US.S042: NFIB Index of Small Business Optimism. [COVID-19-IMPACT]

The Oxford COVID-19 Government Response Tracker (OxCGRT) systematically collects information on several different common policy responses that governments have taken to respond to the pandemic on 18 indicators such as school closures and travel restrictions. It now has data from more than 180 countries. The data is also used to inform a Risk of Openness Index which aims to help countries understand if it is safe to ‘open up’ or whether they should ‘close down’ in their fight to tackle the coronavirus.

United States SBOI: sa: CO: AE: A Year Ago: Number Answer data was reported at 0.000 % in Mar 2025. This records a decrease from the previous number of 1.000 % for Feb 2025. United States SBOI: sa: CO: AE: A Year Ago: Number Answer data is updated monthly, averaging 1.000 % from Jan 2014 (Median) to Mar 2025, with 131 observations. The data reached an all-time high of 10.000 % in Jul 2017 and a record low of 0.000 % in Mar 2025. United States SBOI: sa: CO: AE: A Year Ago: Number Answer data remains active status in CEIC and is reported by National Federation of Independent Business. The data is categorized under Global Database’s United States – Table US.S042: NFIB Index of Small Business Optimism. [COVID-19-IMPACT]

This dataset is part of the bachelor thesis "Evaluating SQuAD-based Question Answering for the Open Research Knowledge Graph Completion". It was created for the finetuning of Bert Based models pre-trained on the SQUaD dataset. The Dataset was created using semi-automatic approach on the ORKG data.

The dataset.csv file contains the entire data (all properties) in a tabular for and is unsplit. The json files contain only the necessary fields for training and evaluation, with additional fields (index of start and end of the answers in the abstracts). The data in the json files is split (training data) and evaluation data. We create 4 variants of the training and evaluation sets for each one of the question labels ("no label", "how", "what", "which")

For detailed information on each of the fields in the dataset, refer to section 4.2 (Corpus) of the Thesis document that can be found in https://www.repo.uni-hannover.de/handle/123456789/12958.

The script used to generate the dataset can be found in the public repository https://github.com/as18cia/thesis_work and https://gitlab.com/TIBHannover/orkg/nlp/experiments/orkg-fine-tuning-squad-based-models

This dataset is a reformatted version of the original GPQA dataset from Idavidrein/gpqa. It includes only the main question, four shuffled answer choices, the correct answer index, subdomain, and a unique id for each entry.Please cite the GPQA paper if you use this data: GPQA: A Graduate-Level Google-Proof Q&A Benchmark.

Our recollections tend to become more similar to the correct information when we recollect an initial response using the correct information, known as the hindsight bias. This study investigated the effect of memory load of information encoded on the hindsight bias’s magnitude. We assigned participants (N = 63) to either LOW or HIGH conditions, in which they answered 20 or 50 questions, which were their initial responses. Then, they memorized and remembered the correct information. They finally recollected the initial responses. Twenty of the fifty questions in the HIGH condition were identical to those in the LOW condition. We used the answers to these 20 common questions in LOW and HIGH conditions to examine the effect of the memory load of information encoded, defined as the number of correct answers to remember (i.e., 20 or 50) on the hindsight bias. Results indicated that the magnitude of the hindsight bias was more prominent in the HIGH than the LOW condition, suggesting that the memory load amplifies the hindsight bias’s magnitude. This finding also implies that controlling the memory load of information encoded when learning correct information could mitigate the hindsight bias. We expect these findings to have practical implications in occupational settings where hindsight bias could lead to critical errors such as financial losses or medical problems.

elg_berta_QA_catalan is an implementation of the ROBERTA model, fined tuned for question-answering tasks, trained on a Catalan dataset. Its objective is, given a question and a context, i.e. a snippet of text that contains the answer to the given question, output the start and end token index that spans the answer.

Monthly response rates for the UK Monthly Business Survey (production) by turnover and questionnaire.