Using A Large Language Model (LLM)

Using a Large Language Models (LLMs), we can generate new titles, descriptions and keywords for documents for a Collection within a SearchBlox Enterprise Search. All the documents in the collection can be updated with LLMs.

🚧

Prerequisites:

📘

Technical Guide:

  • Leveraging LLMs: Using Llama-2-based model, which is known for its superior performance in providing best results.
  • Retrieving and Processing: The model retrieves all indexed documents directly from OpenSearch, the underlying search engine within SearchBlox.
  • Generating Enhanced Metadata:
    • Collecting all the content from each document from a Collection and pass it through the LLM with a prompt.
    • The LLM processes all the collected data, by generating relevant titles, informative descriptions, and a curated list of up to 20 keywords.
  • Seamless Integration: The generated metadata is sent back to Opensearch in JSON format and updated within the document's metadata using document's uid, ensuring the improvements are immediately reflected in the search results.

Generate Titles, Descriptions and Keywords using LLM

Using a Large Language Models (LLMs), we can now create titles, descriptions and keywords for any document.

📘

NOTE:

Large Language Models (LLMs) improves the relevance of the search results even when the quality of the document's metadata is poor.

The key difference between the already existing titles, descriptions, or keywords and the newly generated ones is that the new ones are generated systematically using LLMs and can be more comprehensive regarding the content of the document.

SearchBlox provides two ways to create titles, descriptions and keywords based on Llama-2-based model: