Extract Entities from Text

This Grounding Plugin prompt template extracts entities related to a user-specified topic from supplied input text. It returns both the extracted entities and the original source text, enabling structured information retrieval from unstructured content.

Use this template when you need to identify and extract specific types of entities (people, places, products, concepts) from text based on a defined topic, especially when you want to maintain traceability to the source material. It is particularly useful when processing documents, articles, or conversations where you need to pull out domain-specific entities while preserving the original context.

Do not use this when you need complex relationship mapping between entities or when the source text is already structured data. Avoid using it when you need real-time streaming extraction, as it processes complete input text.

The template requires three input variables:

• input: The text from which entities will be extracted
• topic: The specified topic that defines what types of entities to look for
• example_entities: Examples that guide the extraction process

The template returns the extracted entities along with the source text, providing both the structured output and the original context.

This prompt template is designed for developers and data engineers working with the Semantic Kernel framework who need to implement entity extraction workflows. It serves teams building knowledge management systems, content analysis pipelines, or applications that require structured data extraction from unstructured text while maintaining source attribution.