Datasets for Question Answering by Search and Reading
Alternatives To Quasar
Project NameStarsDownloadsRepos Using ThisPackages Using ThisMost Recent CommitTotal ReleasesLatest ReleaseOpen IssuesLicenseLanguage
2 days ago2mitPython
Handwritten Text Recognition (HTR) system implemented with TensorFlow.
2 months ago2
Maintained collection of OSINT related resources. (All Free & Actionable)
5 months ago27otherPython
Joint Detection and Identification Feature Learning for Person Search
a year ago23apache-2.0Python
NASBench: A Neural Architecture Search Dataset and Benchmark
Showcase Recipe Search400
2 years ago2apache-2.0JavaScript
Instantly search 2M cooking recipes using Typesense Search (an open source alternative to Algolia / ElasticSearch) ⚡ 🥘 🔍
Jstarcraft Rns359
8 months agoapache-2.0Java
专注于解决推荐领域与搜索领域的两个核心问题:排序预测(Ranking)和评分预测(Rating). 为相关领域的研发人员提供完整的通用设计与参考实现. 涵盖了70多种排序预测与评分预测算法,是最快最全的Java推荐与搜索引擎.
2 years ago3October 19, 20202mitPython
Image Dataset Tool (idt) is a cli tool designed to make the otherwise repetitive and slow task of creating image datasets into a fast and intuitive process.
4 years ago3Jupyter Notebook
Visual Search using Apache MXNet and gluon
Uci Ml Api189
2 years ago3mitPython
Simple API for UCI Machine Learning Dataset Repository (search, download, analyze)
Semantic Search145
3 years ago6Python
Semantic search for images and words using neural networks.
Alternatives To Quasar
Select To Compare

Alternative Project Comparisons


Datasets accompanying the paper Quasar: Datasets for QA by Search and Reading.


Both Quasar-S and Quasar-T are available for download here. See the accompanying readme.txt for a description of the included files. The release includes:

  1. Question sets for train / test / dev splits.
  2. Short- and Long-pseudodocuments retrieved by the system described in the paper.
  3. For Quasar-S, an overall list of candidates to which all answers belong.
  4. For Quasar-T, separate lists of noun phrases extracted from the contexts of each question.
  5. Annotations of subset of dev set questions, as described in the paper.

The background corpus for Quasar-S can be downloaded from here (via the SO Scrape link at the bottom). The background corpus for Quasar-T is the ClueWeb09 dataset which you can access here.

An implementation of the F1 and EM evaluation metrics for Quasar-T is included in metric.py.

Explore the Datasets

Below you can find links to the web interface used for evaluating human performance and collecting annotations. This includes an interactive quiz in which the questions are presented one by one, and a search engine for querying the background corpus which you can use in a separate window. Please note that any answers and annotations you enter will be recorded!

Quasar-S: Quiz | Search Engine

Quasar-T: Quiz | Search Engine

The search engine used here is Solr, which uses a Lucene backend similar to the search phase described in the paper. The corpus in each case is the set of short pseudodocuments (sentences) pooled across all queries for that corpus, with no query identifiers or tag filtering. This is just a bare-bones search engine guaranteed not to yield the exact source sentence for the question (as you'd get by just using Google).


If you use these datasets please cite the following:

  title={Quasar: Datasets for Question Answering by Search and Reading},
  author={Dhingra, Bhuwan and Mazaitis, Kathryn and Cohen, William W},
  journal={arXiv preprint arXiv:1707.03904},
Popular Search Projects
Popular Dataset Projects
Popular Computer Science Categories

Get A Weekly Email With Trending Projects For These Categories
No Spam. Unsubscribe easily at any time.
Questions And Answers