{% extends 'base.html' %} {% load crispy_forms_tags %} {% block main %}

Tutorial

Overview

The objective of this task is to generate both true and mutated claims from information extracted from Wikipedia.

This task is separated into two screens that will be given back-to-back. The first screen (WF1a) is the claim generation part of the task. The second screen (WF1b) is the mutations to the generated claims.

WF1a

The objective of this task is to generate true claims

The claims you generate will be based from a source sentence that was extracted from Wikipedia.

WF1b

The objective of this task is to generate modifications to the claims you have just created.

There are six types of modification. For each claim you write, you will have to generate all six modifications.

WF1a Claim Generation

What is a Claim?

A claim is a single sentence expressing information (true or mutated) about a single aspect of one target entity.

In WF1a, the claims you generate will be true claims based on a sentence given to you from Wikipedia

Requirements/Conventions:

  • Claims must reference the target entity directly and avoid use of pronouns/nominals (e.g. he, she, it, the country).
  • Claims must not use speculative/cautious/vague language (e.g. may be, might be, it is reported that).
  • True claims should only be facts that can be deduced by information given in the source sentence and dictionary.
  • Minor variations over the entity name are acceptable: (e.g. Amazon River vs River Amazon), (JFK/John F Kennedy).
  • Correct capitalisation of entity names should be followed (India, not india).
  • Sentences should end with a period.
  • Numbers can be formatted in any appropriate English format (including as words for smaller quantities).

How complex should I make claims?

The claims you generate in WF1a may vary in complexity depending on the information available.

You could use just the source sentence to generate claims. But this will result in simple claims that are not challenging.

We introduce a dictionary of terms containing additional knowledge that can be incorporated into your claims and make them more complex.

How many claims should I generate?

The number of claims you generate will depend on the quality and information content of the sentence given to you that was extracted from Wikipedia.

As a guide, you should aim to generate between 2-5 claims per sentence. But, if the sentence contains lots of information, this could be higher

If the sentence is not informative, you can skip claim generation for this sentence.

Example


Generating Claims About

India

Source Sentence

This is the sentence that is used to substantiate your claims about India

It shares land borders with Pakistan to the west; China, Nepal, and Bhutan to the northeast; and Myanmar (Burma) and Bangladesh to the east.
It is bounded by the Indian Ocean on the south, the Arabian Sea on the southwest, and the Bay of Bengal on the southeast. It shares land borders with Pakistan to the west; China, Nepal, and Bhutan to the northeast; and Myanmar (Burma) and Bangladesh to the east. In the Indian Ocean, India is in the vicinity of Sri Lanka and the Maldives.
Show Context Hide Context

Dictionary

Click the word for a definition. These definitions can be used to support the claims you write

Bhutan

Bhutan , officially the Kingdom of Bhutan , is a landlocked country in Asia, and it is the smallest state located entirely within the Himalaya mountain range.

True Claims (one per line)

Example
The Amazon River, usually abbreviated to Amazon (US: /ˈæməzɒn/ or UK: /ˈæməzən/; Spanish and Portuguese: Amazonas), in South America is the largest river by discharge volume of water in the world and according to some authors, the longest in length.

Good

  • The Amazon River is located in South America.
  • The River Amazon is a river in the southern hemisphere.
  • Amazonas is another name for the Amazon River.
  • The Amazon is the longest river in the world.
  • The River Trent is shorter than the Amazon.

Bad

  • The Amazon is might be the longest river ('might be' is cautious/vague language)
  • The Amazon River is home to river dolphins (not explicitly mentioned in text).
  • One of the land borders that India shares is with the world's most populous country.
  • India borders 6 countries.
  • The Republic of India is situated between Pakistan and Burma.

Why did we chose these examples?

  • The first claim uses information from the dictionary entry for China which states that it is the world's most populous country.
  • The second claim summarises the list of countries in the source sentence.
  • The third claim is deduced by Pakistan being West of India, and Burma being to the East.
{% endblock %}