Note‐Search‐Guide

Note Search Function Guide

Overview

Knot offers two separate search modes located in the top toolbar and bottom toolbar of the note page, with distinct matching logic and activation requirements:

1. Top search: Exact text keyword matching, available without extra setup;
2. Bottom semantic search: Vector-based semantic retrieval powered by local embedding models. The button stays hidden if no local model is configured.

Search Button Differentiation

1. Top Toolbar Search (Exact Text Match)

Location: Magnifying glass icon in the top navigation bar of note page

• Matching Logic: Exact character matching across note titles, body text and tags; returns notes containing the exact keyword you typed.
• Prerequisite: No extra configuration required, usable at all times.
• Best For: Locating notes with exact words, filenames or specific terms you remember clearly.

2. Bottom Toolbar Search (Semantic Vector Search)

Location: Magnifying glass icon in the bottom operation bar (blue highlighted toggle in screenshot)

• Matching Logic: Semantic similarity matching powered by local GGUF embedding models. It recognizes synonyms and related concepts instead of strict literal matches.
• Mandatory Prerequisite: Complete local embedding model setup first, place multilingual-e5-small-fp32.gguf inside the .Knot/model directory on your device.
• Visibility Rule: The bottom search button remains hidden and unclickable if no valid embedding model is detected. It will only appear after a valid model is loaded successfully.
• Best For: Retrieving notes by vague descriptions or general ideas when exact keywords are forgotten.

Step-by-Step Usage

Exact Text Search (Top Toolbar)

1. Tap the magnifying glass icon in the top toolbar;
2. Type your keyword, the list instantly filters all notes with matching text;
3. Navigate through matched entries or close the search panel.

Semantic Vector Search (Bottom Toolbar)

1. Deploy the local embedding model following the setup guide and restart Knot to load the model;
2. The semantic search button will appear in the bottom toolbar; click to open the search input;
3. Type descriptive sentences, and Knot returns semantically related notes based on vector similarity;
4. Browse matched results or close the search panel to return to full note list.

Core Comparison Table

Search Entry	Retrieval Type	Setup Requirement	Button Visibility
Top Toolbar Search	Exact Text Match	None, always available	Permanently visible
Bottom Toolbar Search	Semantic Vector Search	Valid local GGUF embedding model required	Hidden without model, visible after model loaded

Troubleshooting

1. Semantic search button missing: Check if valid .gguf embedding file exists inside the hidden .Knot/model folder, restart Knot to reload files;
2. Poor semantic matching results: Try a larger embedding model and wait for full model initialization before searching;
3. Zero hits from exact search: Double-check keywords, spaces and punctuation — exact match respects all text symbols.