What does the Grounding Plugin — Excise Entities prompt do?

It removes a list of ungrounded entities from input text in a coherent manner, returning the original text without those ungrounded elements while preserving readability and flow.

When should I use this prompt template?

Use it when you need to clean text by removing specific entities you've already identified as ungrounded or unsupported, particularly for post-processing AI-generated content where certain claims have been flagged as problematic.

What shouldn't I use this template for?

Do not use it if you need to identify which entities are ungrounded in the first place, or if you need to replace ungrounded content with alternatives rather than simply removing it.

Prompt

Remove Unwanted Entities from Text

Name: Grounding Plugin - Excise Entities
Availability: OnlineOnly
Author: Semantic Kernel

A prompt template that removes ungrounded entities from text while maintaining coherence, returning clean output without specified unsupported claims or facts.

Copy prompt

Works with github

Semantic Kernel

Maintainer?

Spark score

out of 100

Updated last month

Version python-1.43.1

Add to Favorites

Why it matters

Cleanse your text by removing specific, ungrounded entities. This asset ensures your content is focused and relevant by excising extraneous information.

Outcomes

What it gets done

Identify and remove a list of specified entities from input text.

Maintain text coherence after entity removal.

Process text to exclude unwanted data points.

Install

Add it to your toolbox

Run in your project directory:

curl -fsSL https://spark.entire.vc/get/sk-groundingplugin-exciseentities | bash

Overview

Grounding Plugin - Excise Entities

This prompt template removes a list of ungrounded entities from provided text while maintaining coherence. It takes two inputs - the text to process and a list of entities to remove - and returns the original text with those entities excised in a way that preserves readability. Use this when you have identified specific entities in text that lack factual grounding and need to remove them cleanly. It is designed for scenarios where you already know which entities are problematic and want to excise them while keeping the remaining text coherent.

What it does

This prompt template removes a list of ungrounded entities from input text in a coherent manner. It processes text to excise specific entities that lack factual support or grounding, returning the original input without those ungrounded elements while preserving readability and flow.

When to use - and when NOT to

Use this template when you need to clean text by removing specific entities identified as ungrounded, unsupported, or lacking factual basis. It is ideal for post-processing AI-generated content where certain claims or entities have been flagged as problematic. Use it when you have already identified which entities need removal and want to maintain text coherence during the excision process.

Do not use this template if you need to identify which entities are ungrounded - it only removes entities you explicitly provide in the list. Do not use it if you need to replace ungrounded content with alternatives rather than simply removing it.

Inputs and outputs

The template requires two input variables: input (the text to be processed) and ungrounded_entities (the list of entities to remove). It returns the input text with all specified ungrounded entities removed, maintaining coherent sentence structure and flow.

Source README

Remove a list of ungrounded entities from a given text in a coherent manner. Returns the input text without the ungrounded entities in the list

Input variables: input, ungrounded_entities

FAQ

Common questions

Discussion