Skip to content

Dataset · stanfordnlp

Web Fact Question Answering Dataset

stanfordnlp/web_questions

6,642 English question-answer pairs focused on specific Freebase knowledge base entities.

What's inside

It contains 6,642 English text question-answer pairs. The questions were collected from popular web search queries (relevant as of 2013) and mostly concern a single specific entity whose answer can be found in the Freebase knowledge graph.

What you can do with it

  • Training and testing question answering (QA) models
  • Testing knowledge extraction systems from graphs and databases
  • Evaluating language model factual knowledge without internet access
Tasks
Question answering
Languages
English
Size
1K–10K rows
Downloads this month
16.1K
Likes
45

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("stanfordnlp/web_questions")