Managing Vector Content

How to Upload Documents

Prerequisites: You have documents in supported formats (PDF, DOCX, TXT, MD, CSV, XLSX, XLS, or JSON).

Steps:

1. Click Add, choose Documents

1. Select your file(s) using the file picker

1. Configure chunking settings:

Setting	Description	Default	Range
Chunking Method	How text is divided	Word	Letter, Word, Sentence, Passage
Chunk Size	Size of each text segment	200	0-4500
Chunk Overlap	Overlap between chunks	3	0-4500

1. Choose destination (Save to groups): Select an existing group

1. Click Create

Expected Result: Document uploads and begins processing. Status shows “Processing” then “Completed”.

Tip: Use larger chunk sizes (400-500) for technical documents, smaller sizes (100-200) for conversational content.

Error Handling:

• File type not supported: Ensure your file is PDF, DOCX, TXT, MD, CSV, XLSX, XLS, or JSON format

• File is larger than 50MB: Compress your file or split into smaller sections

• Processing failed: Check file integrity and try re-uploading or “Rejected” if over limitations.

• The desired group not be visible: Please refresh the page to obtain the most current list.

How to Create Articles

Prerequisites: You want to add custom text content to your knowledge base.

Steps:

1. Click Add, choose Articles

1. Enter article details:

• Name: Required field, must be unique

• Content: Use the rich text editor to format your content

1. Configure chunking settings:

Setting	Description	Default	Range
Chunking Method	How text is divided	Word	Letter, Word, Sentence, Passage
Chunk Size	Size of each text segment	200	0-4500
Chunk Overlap	Overlap between chunks	3	0-4500

1. Choose destination (Save to groups): Select an existing group

1. Click Create

Expected Result: Article is created and processed for vectorization.

Content Editor Features:

• Text formatting (bold, italic, strikethrough, underline, square code, headers, and similar formatting options)

• Bullet points and numbered lists

• Image/video insertion

• Link embedding

Error Handling:

• Limitation: Articles should not exceed 300,000 words.

• The desired group not be visible: Please refresh the page to obtain the most current list.

How to Add URLs and Crawl Websites

Option 1: Individual URLs

Prerequisites: You have specific web pages to add to your knowledge base.

Steps:

1. Click Add, choose URLs

1. Select Add individual links

1. Enter the URL (must include http:// or https://)

1. Click Add another link for multiple URLs

2. Configure chunking settings:

Setting	Description	Default	Range
Chunking Method	How text is divided	Word	Letter, Word, Sentence, Passage
Chunk Size	Size of each text segment	200	0-4500
Chunk Overlap	Overlap between chunks	3	0-4500

1. Choose destination (Save to groups): Select an existing group

1. Click Create

Option 2: Crawl Website Sitemap

Prerequisites: The target website has an accessible sitemap.

Steps:

1. Click Add, choose URLs

1. Select Crawl website sitemap

1. Enter the website URL (e.g., https://example.com)

1. Click Fetch

2. Wait for sitemap retrieval (loading screen appears)

3. Review the URL list:

• Deselect unwanted URLs by unchecking boxes

• URLs may display as relative paths (e.g., /products/item1 instead of full URL)

1. Configure chunking settings:

Setting	Description	Default	Range
Chunking Method	How text is divided	Word	Letter, Word, Sentence, Passage
Chunk Size	Size of each text segment	200	0-4500
Chunk Overlap	Overlap between chunks	3	0-4500

1. Choose destination (Save to groups): Select an existing group

1. Click Create

Expected Result: Selected URLs are queued for processing and appear with “Processing” status, then “Completed” when ready, or “Rejected” if over limitations.

Domain Parsing Example:

• Input: https://dev.pp1.fr/tnc

• Crawled URL: https://dev.pp1.fr/tnc/vehicule/porsche-911

• Display: vehicule/porsche-911

Troubleshooting URL Issues:

• Failed to fetch sitemap: Verify URL is correct and sitemap is accessible

• No valid URLs found: The website may not have a standard sitemap format

• Processing failed: The URL content may be unavailable, not supported, or beyond the allowed crawling limit.