Context Memory Search

Functionality

The "Context Memory Search" module is designed to provide more detailed information in response to a follow-up query after an initial document search. This module goes deeper into the already retrieved sources and extracts additional information without triggering a new search. It is required to have the document search module activated when using the context memory search.

Example scenario:

  1. User query: "What is the guideline saying about travels to Europe?"

  2. Initial document search result: An answer is provided based on internal documents.

  3. Follow-up query: "Please give more details on this."

The module then extracts more details from the previously retrieved sources to provide a more in-depth or extensive answer.


Input

A user query follow-up from a document search, requesting more detailed information about the retrieved documents.

Example input:

image-20240522-071917.png
image-20240522-072223.png

Output

A more in-depth or extensive answer to the user's follow-up question, based on the previously retrieved sources.


Configuration settings (technical)

This section contains all configurable parameters as seen in the JSON.

General parameters

Parameter

Description

Parameter

Description

languageModel: string

Specifies the language model used

Default: AZURE_GPT_35_TURBO_0613

chunkedSources: boolean

Indicates if each chunk is appended as an individual source (true) or if the chunks from the same document are merged (false)

Default: false

Prompts

Only adjust prompts if you are fully familiar with the code logic. Small changes can break the module or reduce the output quality.

Parameter

Description

Parameter

Description

systemPromptContextMemorySearch: string

System prompt

triggerPromptContextMemorySearch: string

Trigger prompt


Author

@Pascal Hauri

 

© 2024 Unique AG. All rights reserved. Privacy PolicyTerms of Service