Skip to main content

Keyword Management

Keyword Management is the classification rules editor for comments and findings. It defines categories (Safety, Major, Minor, Maintenance, etc.) and the keywords that map a comment into each category. When keyword-based color coding is enabled in Smart Comments, the renderer uses these rules to color-code every finding card in the report. Keyword Management lives inside the Smart Comments page as an embedded section (not a separate route). This article documents the keyword-rule subsystem in depth.

Why classify by keyword

Inspection reports communicate severity. Clients respond to visual signals: a red border around a finding feels urgent in a way that black text doesn’t. Keyword Management lets you build the rule that “any finding whose text contains ‘leak’, ‘water damage’, ‘staining’, or ‘dripping’ gets the Major-Concern color”. The system is intentionally simple — keyword matching, not AI classification. You define the rules; the renderer applies them deterministically.

Screen layout

[SCREENSHOT: keyword-management.png — desktop, the Keyword Management section showing Add Keywords on the left with Category picker and comma-separated input, Keyword Rules list on the right, and a Test Keyword Classification panel below.] Three main areas:
  1. Add Keywords — form for adding new rules
  2. Keyword Rules — list of existing categories and their keywords
  3. Test Keyword Classification — try a sample comment to see which category matches

Add Keywords form

The left-side form lets you add keywords to a specific category.

Fields

After filling, click Add Keywords. The keywords are appended to the chosen category’s rule and immediately active.

Default categories

Tenants typically start with these categories pre-configured: These are configurable at the tenant level (your administrator can rename, add, or remove categories).

Common keyword starters

By category:

Keyword Rules list

The right side shows every defined category and its keywords as a searchable list.

Editing keywords

Click any keyword in the list to inline-edit it. Changes save immediately on blur or Enter.

Deleting a keyword

The 🗑 icon next to each keyword removes it from the category. No confirmation — single-keyword deletes are easily reversible by re-typing.

Test Keyword Classification

A test pane at the bottom of the section. Useful for verifying your rules behave correctly without saving a real inspection.

Fields

The result panel might show:
Classification: Major Concern Matched keywords: “leak”, “staining”
If no keywords match:
Classification: Uncategorized No matching keywords found.
Uncategorized comments render in the neutral default style.

How classification actually works

When the renderer encounters a finding’s narrative:
  1. The narrative text is lowercased
  2. Each category’s keywords are checked for substring presence (also lowercased)
  3. The first matching category in priority order wins (typically Safety > Major > Minor > Maintenance > Informational)
  4. The renderer applies that category’s color to the finding card

Priority order

Categories have an implicit priority based on their creation/edit order. If a comment matches both “Safety” and “Major Concern”, Safety wins (because more severe). To change priorities, you may need to delete and re-create categories in the desired order — this is platform-config rather than UI, so contact support for re-ordering.

Multi-keyword matches

If a comment contains “leak” (Major Concern) AND “fire risk” (Safety), Safety wins because of priority. The Test Classification shows you which keywords matched, but the rendered output uses only the highest-priority match.

Best practices

  • Start narrow. Add only the most obvious keywords per category — overly broad keywords (e.g. “old”) trigger false matches across many comments.
  • Test with real narratives. Use the Test Classification pane against actual narratives from your library to validate.
  • Be careful with negations. The matcher is keyword-substring only — it doesn’t understand “no leak” vs “leak”. A comment that says “No active leak observed” matches leak. Avoid keywords prone to negation.
  • Don’t over-categorize. 5–7 categories with focused keywords beats 15 categories with overlapping keywords.
  • Document your category meanings. Add a description to each category (where supported) so future inspectors know what a Major Concern means to your business.

When keyword-based color coding is OFF

If Smart Comments has the Keyword-based color coding toggle off:
  • The renderer ignores these keyword rules
  • All findings render in the neutral default style
  • The rules sit dormant — they’re saved but not applied
This is useful during initial setup: define rules, leave color coding off, verify in test mode, then turn on when satisfied.

Errors and recovery


What this page does NOT do

  • It does not edit narratives themselves. That’s Narrative Collections.
  • It does not run AI classification. Keyword matching is deterministic substring matching, not LLM categorization.
  • It does not affect mobile. The classification is rendered-PDF only. Mobile doesn’t show color-coded findings during capture — only the published report does.
  • It does not override severity rated by inspectors. A finding rated D (Deficient) on mobile keeps that rating regardless of keyword classification — the rendering color is additive, not corrective.