---
id: problem-solution-en-gilardi2023
title: Problem/ Solution Frames
author: Gilardi et al. 2023
date: 11.4.2023
language: en
task: frames
version: 1.0
addedby: chkla
keywords: frames, content moderation
---

## Prompt Description

[Briefly describe the purpose of the prompt and the context in which it is intended to be used, especially in the context of artificial annotation with generative models.]

## Prompt Text

Content moderation can be seen from two different perspectives:<br>
• Content moderation can be seen as a PROBLEM; for example, as a restriction of free speech<br>
• Content moderation can be seen as a SOLUTION; for example, as a protection from harmful speech<br>
<br>
For each tweet in the sample, follow these instructions:<br>
1. Carefully read the text of the tweet, paying close attention to details.<br>
2. Classify the tweet as describing content moderation as a problem, as a solution, or neither.<br>
<br>
Tweets should be classified as describing content moderation as a PROBLEM if they emphasize negative effects of content moderation, such as restrictions to free speech, or the biases that can emerge from decisions regarding what users are allowed to post.<br>
<br>
Tweets should be classified as describing content moderation as a SOLUTION if they emphasize positive effects of content moderation, such as protecting users from various kinds of harmful content, including hate speech, misinformation, illegal adult content, or spam.<br>
<br>
Tweets should be classified as describing content moderation as NEUTRAL if they do not emphasize possible negative or positive effects of content moderation, for example if they simply report on the content moderation activity of social media platforms without linking them to potential advantages or disadvantages for users or stakeholders.

## Language

- Prompt Language: [Specify the language of the prompt, e.g., English]
- Dataset Language: [Specify the language of the dataset to which the prompt is applied, e.g., English]

## NLP Task

- Task: [Specify the NLP task in more detail, e.g., sentiment analysis, named entity recognition, summarization]
- Subtask: [If applicable, provide any subtask or variation related to the main NLP task, e.g., binary sentiment classification, multi-class sentiment classification]

## Example Input and Output

- Example 1
  - Input: [Provide an example input for the prompt]
  - Output: [Provide an example output for the given input]
- Example 2
  - Input: [Provide another example input for the prompt]
  - Output: [Provide another example output for the given input]

## Parameters and Constraints

- Parameter 1: [Specify any parameters, such as temperature or token count]
- Parameter 2: [Specify additional parameters or constraints if applicable]

## Evaluation Metrics

[List the evaluation metrics used to assess the quality of the generated artificial annotations, such as accuracy, F1 score, or BLEU score.]

## Use Cases

[List any specific use cases or applications for the prompt in artificial annotation, such as data annotation, semi-supervised learning, or active learning.]

## Limitations and Potential Biases

[Briefly discuss any limitations or potential biases associated with the prompt, as well as any steps taken to mitigate them, in the context of artificial annotation with generative models.]

## Related Research and References

[List any relevant research papers, articles, or resources that informed the creation of the prompt or are closely related to it, especially in the area of artificial annotation with generative models. Include proper citations where applicable.]