Hi,
I’m evaluating EGroupware 26.x as an internal knowledge and editorial-management system for a small content website.
The public website contains a collection of interconnected reference articles. This is an example of the type of content involved:
Disclosure: I maintain the linked website. I’m sharing it only as the actual external content source I would like to make searchable through EGroupware.
The page links to individual reference articles, and those articles are updated periodically. I would like editors using EGroupware to be able to search the material semantically and ask questions such as:
- Which articles mention a particular symbol or theme?
- Which subjects are covered by multiple articles?
- Which pages may contain contradictory interpretations?
- Which source article was used for a particular internal note?
- Which articles have not been reviewed recently?
I’m trying to understand the correct architecture before importing anything.
1. Can RAG index external URLs?
Can the EGroupware RAG system crawl and index a public website directly, starting from a sitemap or a list of URLs?
Or does all source material need to exist inside an EGroupware application before it can be indexed?
2. Best EGroupware application for imported articles
If the content must be imported, which application would be the most appropriate source?
- KnowledgeBase
- File Manager with HTML, Markdown or PDF files
- InfoLog
- A custom EGroupware application
- An external BookStack instance connected through OpenID
I need to retain the original URL, article title, update date, subject category and review status.
3. Keeping the index synchronized
The public articles may change after they have been imported.
Is there an existing method for periodically checking the source URL and re-indexing only content that has changed?
My initial idea would be to store:
- Canonical URL
- Source content hash
- Last fetched date
- Last modified date
- EGroupware entry ID
- Indexing status
A scheduled job could fetch each page, compare the content hash and update the corresponding EGroupware record when necessary.
Would this fit the expected EGroupware architecture, or should synchronization be handled outside EGroupware?
4. Avoiding duplicate content
The website includes overview pages and individual detail pages that may repeat some definitions.
Does the RAG implementation provide any deduplication or source-priority controls, or should I remove navigation, repeated introductions and template content before saving each article?
5. Source attribution
When a RAG search returns an answer, can it also display the originating EGroupware entry and the original public URL?
For this use case, editors need to verify the source rather than receiving an answer without a citation.
6. Permissions
Some editorial notes would be private, while the imported website content would be available to the whole editorial group.
Does the RAG search automatically respect the permissions of the original EGroupware entries, including mixed search results from public reference material and restricted internal notes?
7. Built-in RAG search or OpenWebUI?
For searching and comparing this material, would you recommend using the native EGroupware RAG search first, or connecting EGroupware data to OpenWebUI?
The initial requirement is search and source retrieval rather than an autonomous assistant that creates or modifies records.
I’m mainly looking for guidance on the supported data-ingestion route. I do not need feedback about the subject of the linked website.