Prompt Examples

Prompts examples for row labeling and span labeling projects.

Row labeling

In row labeling projects, you can use the following attributes to define the input and expected output:

• {targetText} : Input text from the selected column in your data.

• {targetOptions} : Answer options from the selected target question.

The extension also supports one-shot or few-shot prompting. Below are prompt examples based on different question types.

Single choice

System prompt	User prompt
You are an expert data labeler for category classification	Article: {targetText} What is the topic for the article above? Choose one from the options below {targetOptions} Answer:

Dropdown

System prompt	User prompt
You are an expert data labeler for category classification.	Article: {targetText} What is the topic for the article above? Choose one from the options below {targetOptions} Answer:

Hierarchical dropdown

System prompt	User prompt
You are an expert data labeler for category classification.	Article: {targetText} What is the topic for the article above? Choose one from the options below {targetOptions} Answer:

Text

System prompt	User prompt
You are an expert data labeler for category classification.	Article: {targetText} What is the topic of the article above? Choose one topic. Do not provide any explanation. Answer:

Date

System prompt	User prompt
You are an expert data labeler for extracting dates from an article.	Article: {targetText} Extract the date and format it as YYYY-MM-DD. Do not provide any explanation. Answer:

Time

System prompt	User prompt
You are an expert data labeler for extracting timestamps from an article.	Article: {targetText} Extract the timestamp and format it as HH🇲🇲ss.SSS. Do not provide any explanation. Answer:

Checkbox

System prompt	User prompt
You are an expert data labeler for category classification.	Statement: {targetText} {targetName}: Choose either true or false. Do not provide any explanation. Answer:

Slider

System prompt	User prompt
You are an expert data labeler for evaluating articles.	Article: {targetText} What is the factual accuracy score of the article above? Choose a number between {targetMin} and {targetMax}. Do not provide any explanation. Answer:

URL

System prompt	User prompt
You are an expert data labeler for extracting URLs from an article.	Article: {targetText} Extract the URL. Do not provide any explanation. Answer:

Span labeling

For span labeling, you can switch the label set by selecting the active label set in the upper-left corner.

You can use the following attributes to define the input and expected output:

• {targetText} represents input text from your selected column in the uploaded data.

• {targetLabelSet} represents label set options from your active label set selection in the upper left corner.

Below is an example user prompt for NER labeling:

Notes:
Return the labeled entity tokens in object format, where each label set is the key and the value is an array of tokens. Do not provide any explanation. Do not provide any explanation.

LabelSet:
{targetLabelSet}

Sentence:
{targetText}