github-starred/open-webui
2
0
Fork 0
mirror of https://github.com/open-webui/open-webui.git synced 2026-05-08 04:16:03 -05:00
Code Issues 1.3k Packages Projects Releases 126 Wiki Activity

Improve Web Search Intelligence in LLM Responses #1790

New Issue
Closed
opened 2025-11-11 14:53:20 -06:00 by GiteaMirror · 0 comments
GiteaMirror commented 2025-11-11 14:53:20 -06:00
Owner
Copy Link

Originally created by @odefta on GitHub (Aug 15, 2024).

Current Behavior

When web search is enabled and a user asks a specific question, the system performs a literal search of the entire question. For example:

Query: "I don't know how to modify the column type of a materialized view in Oracle 19c from char(5) to varchar2(100)... please help me"

Current search: [entire user query]

This often leads to poor or irrelevant search results, as the search query is too specific and unlikely to match existing web content directly.

Proposed Behavior

  1. The LLM should first analyze the user's query to determine the core information need.
  2. Based on this analysis, it should generate a more generalized and effective search query.
  3. It should then perform the web search using this optimized query.
  4. Finally, it should use the search results to inform its response to the user's original question.

Example:

User Query: "I don't know how to modify the column type of a materialized view in Oracle 19c from char(5) to varchar2(100)... please help me"

  • Current Search: [entire user query]
  • Proposed Search: "alter materialized view column type oracle 19c"

The LLM would then use the results from this more targeted search to construct a detailed response to the user's specific question.

Benefits

  • More relevant and useful web search results
  • Improved ability of the LLM to provide accurate and helpful responses
  • Better user experience, especially for technical or specific queries
  • Closer alignment with advanced AI platforms like Abacus.AI

Technical Considerations

  • This feature will require the LLM to perform two-step processing:
    1. Analyze the query and generate a search term
    2. Use the search results in formulating a response
  • The system may need to implement a form of query understanding or intent classification to generate effective search terms.
  • Consider implementing this as an optional feature that can be toggled on/off, as some users might prefer the current literal search in certain scenarios.

This enhancement would significantly improve the practical utility of the web search feature in OpenWebUI, especially for technical and specific queries (where targeted information is crucial).

Originally created by @odefta on GitHub (Aug 15, 2024). ## Current Behavior When web search is enabled and a user asks a specific question, the system performs a **literal search of the entire question**. For example: Query: "I don't know how to modify the column type of a materialized view in Oracle 19c from char(5) to varchar2(100)... please help me" Current search: [entire user query] This often leads to **poor or irrelevant search results**, as the search query is too specific and unlikely to match existing web content directly. ## Proposed Behavior 1. The LLM should **first analyze the user's query** to determine the core information need. 2. Based on this analysis, it should **generate a more generalized and effective search query**. 3. It should then **perform the web search using this optimized query**. 4. Finally, it should **use the search results to inform its response** to the user's original question. ### Example: User Query: "I don't know how to modify the column type of a materialized view in Oracle 19c from char(5) to varchar2(100)... please help me" - Current Search: [entire user query] - **Proposed Search: "alter materialized view column type oracle 19c"** The LLM would then use the results from this more targeted search to construct a detailed response to the user's specific question. ## Benefits - **More relevant and useful web search results** - **Improved ability of the LLM to provide accurate and helpful responses** - Better user experience, especially for technical or specific queries - Closer alignment with advanced AI platforms like Abacus.AI ## Technical Considerations - This feature will require the LLM to perform **two-step processing**: 1. Analyze the query and generate a search term 2. Use the search results in formulating a response - The system may need to implement a form of **query understanding or intent classification** to generate effective search terms. - Consider implementing this as an **optional feature** that can be toggled on/off, as some users might prefer the current literal search in certain scenarios. --- This enhancement would significantly improve the practical utility of the web search feature in OpenWebUI, especially for technical and specific queries (where targeted information is crucial).
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
bug
confirmed
confirmed issue
core
documentation
enhancement
good first issue
help wanted
non-core
pull-request
Mirrored from GitHub Pull Request
 python
question
testing wanted
No Label
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
GiteaMirror ninjasurge
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/open-webui#1790
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?