Configure Basic Chunk Settings

Overview

Flow ID: configure-basic-chunks
Category: Blockify Processing
Estimated Duration: < 1 minute
User Role: All Users
Complexity: Simple

Purpose: Provides a simplified interface for adjusting document splitting parameters. Unlike advanced settings, this view offers preset recommendations or simple sliders without the complex preview visualization, suitable for most users.

Trigger

What initiates this flow:

• User manually initiates

Specific trigger: During job setup, locating the “Chunk Settings” section (often expanded by default).

Prerequisites

Before starting, users must have:

• Files uploaded to the job

User Intent Analysis

Primary Intent

Ensure documents are split into reasonable sizes for search without getting bogged down in technical details.

Secondary Intents

• Trusting system defaults (“Just make it work”)
• Slight adjustments for “Dense” vs “Sparse” content

Step-by-Step Flow

Main Path (Happy Path)

Step 1: Locate Settings

• User Action: Scroll to Chunk Settings section in Job Creation screen.
• System Response: Shows current parameters (e.g., “Size: 512 tokens”, “Overlap: 50 tokens”).

Step 2: Choose Preset (If available)

• User Action: Select a preset like “Balanced” (Default), “Granular” (Small chunks), or “Coherent” (Large chunks).
• System Response: Sliders automatically move to preset positions.

Step 3: Manual Adjustment (Optional)

• User Action: Move Size slider slightly.
• System Response: Value updates.
• Visual Cues: Helper text explaining trade-off (e.g., “Smaller chunks = more precise search, less context”).

Error States & Recovery

Error 1: Invalid Values

Cause: Typing manual numbers outside allowed range
User Experience: Input turns red, “Value must be between 100 and 2048”
Recovery: Use sliders or correct the number.

Pain Points & Friction

1. “What is a token?”: Users don’t know what 512 tokens means.
   • Mitigation: Display “Approx. X words” next to the token count.

Design Considerations

• Accordions: Keep this section collapsible if “Smart Defaults” are reliable, to reduce visual noise.
• Tooltips: Explain “Overlap” simply (e.g., “Repeats text to ensure context isn’t cut off”).

Related Flows

• Configure Advanced Chunks - For more control
• Create Chunking Job - Context of use

Technical References

• src/components/blockify-corpus/basic-settings.js
• src/constants/processing-defaults.js