How to Configure Segments in FastLinePro?

Segments are text fragments extracted from documents in the knowledge base. Configuring segments allows you to manage how documents are split, edit individual segments, and optimize search accuracy. This instruction will show how to configure segment settings and manage segments in FastLinePro.

When Needed

• You need to configure how documents are split into segments for better search.
• You want to edit or delete specific segments from a document.
• You need to optimize segment settings to improve search accuracy.
• You want to review which segments are most frequently used (hit count).

What's Important to Know

• Segment — a text fragment extracted from a document for indexing and search.
• Segment identifier — separator used to split documents into segments (for example, --- or \n\n).
• Hit count — number of times a segment was retrieved during searches (shows segment relevance).
• Segment settings — configuration that determines how documents are split and processed.
• Segments can be enabled or disabled individually.

Before You Start

• You are logged in with integrator or administrator rights.
• You have navigated to the FastLinePro module via Menu -> FastLinePro.
• You have a knowledge base with uploaded documents.
• You have opened the dataset (knowledge base) for which you need to configure segments.

Step-by-Step Instructions

1. Open Segment Settings

1. In the FastLinePro module, go to the "Datasets" or "Knowledge Bases" section.
2. Select the dataset (knowledge base) you need.
3. Open the dataset settings or go to the "Segments" tab.
4. The system displays a list of segments extracted from documents.

2. Configure Segment Settings (Fragmentation)

1. In the dataset settings, find the "Segment Settings" or "Fragmentation Settings" section.
2. Configure the "Segment Identifier" (separator):
   • Enter the character or sequence of characters used to split text (for example, ---, \n\n, !--).
   • Common separators:
     • --- (three dashes) — standard separator for topics
     • \n\n (double newline) — for paragraphs
     • \n (single newline) — for lines
     • Custom separators like !-- or ***
   • The separator must exactly match what is used in your documents.
3. Configure "Text Preprocessing Rules":
   • "Replace consecutive spaces, newlines, and tabs" — enables automatic cleanup of extra whitespace.
   • "Remove all URLs and email addresses" — removes URLs and emails from segments (optional).
4. Configure "Max Segment Length" (if available):
   • Set the maximum number of characters per segment.
   • Prevents segments from being too long.
5. Enable "Hybrid Search" (recommended):
   • Hybrid search combines vector and full-text analysis.
   • Improves search accuracy for relevant information.
   • Recommended to keep enabled.
6. Configure "Top K" parameter:
   • Determines how many segments the AI agent considers when formulating a response.
   • Recommended: 2–3
   • Larger values increase chance to find answers but use more tokens.
7. Click "Save" or "Apply" to save segment settings.

3. View Segments List

1. In the dataset, go to the "Segments" tab or section.
2. The system displays a list of all segments from documents in the dataset.
3. For each segment, you can see:
   • Content — text of the segment
   • Word count — number of words in the segment
   • Hit count — how many times the segment was retrieved
   • Source document — which document the segment came from
   • Status — enabled or disabled

4. Edit Individual Segment

1. In the segments list, find the segment you want to edit.
2. Click the "Edit" button or click on the segment.
3. In the edit dialog:
   • Modify the segment text if needed.
   • Review segment information (word count, source document).
4. Click "Save" to save changes.

5. Enable or Disable Segment

1. In the segments list, find the segment you want to enable or disable.
2. Use the toggle switch or checkbox to enable/disable the segment.
3. Disabled segments are not used in search but remain in the dataset.
4. Changes take effect immediately.

6. Delete Segment

1. In the segments list, find the segment you want to delete.
2. Click the "Delete" button.
3. Confirm deletion in the dialog.
4. The segment is permanently deleted from the dataset.

What Happens After

After configuring segment settings:

• Documents are processed and split into segments according to your settings.
• Segments are indexed for search.
• The AI agent can use segments to find relevant information.

After editing segments:

• Changes are saved and immediately available for search.
• Hit counts continue to track segment usage.

After disabling segments:

• Disabled segments are excluded from search.
• They remain in the dataset and can be re-enabled later.

How to Verify It Worked

• Check that segments are created according to your separator settings.
• Verify that segment content matches document structure.
• Test search to ensure segments are being used correctly.
• Review hit counts to see which segments are most relevant.

Configuration Recommendations

Segment Identifier

• Use --- as the standard separator for best compatibility.
• Ensure the separator exactly matches what is in your documents.
• If documents have no separators, the system will auto-split, but manual separators give better control.

Text Preprocessing

• Enable "Replace consecutive spaces" — helps clean up formatting issues.
• Enable "Remove URLs and emails" — useful if you don't want URLs/emails in search results.

Hybrid Search

• Always enable — provides best search accuracy.
• Combines advantages of both vector and full-text search.

Top K

• For simple questions: 2 segments
• For complex questions: 3–4 segments
• For very complex questions: 4–5 segments (increases token costs)

Related Materials

• Configure chunking — detailed guide on chunking configuration
• Create knowledge base — how to create a knowledge base
• Upload files to KB — how to upload files to knowledge base
• Connect knowledge base to agent — how to connect KB to AI agent

Important Notes

• ⚠️ Segment identifier: Must exactly match separators in your documents. If it doesn't match, segments may not be created correctly.
• 📊 Hit count: Shows which segments are most useful. Segments with high hit counts are frequently retrieved and likely contain important information.
• 🔄 Re-processing: If you change segment settings, you may need to re-process documents to apply new settings.
• 🚫 Deleting segments: Deletion is permanent. Make sure you really want to delete a segment before confirming.