[FEAT]: Incident Timeline Extraction for Structured Emergency Reports #286
Description
name: 🚀 Feature Request
about: Suggest an idea or a new capability for FireForm.
title: "[FEAT]: Incident Timeline Extraction for Structured Emergency Reports"
labels: enhancement, ai, data-structure
assignees: ''
📝 Description
Fire incident reports often contain multiple events that occur in chronological order, such as unit arrival, fire containment, medical response, and incident resolution. Currently, FireForm extracts structured fields from natural language descriptions, but it does not capture the temporal sequence of events within an incident.
This feature proposes introducing an Incident Timeline Extraction component that automatically identifies time-based events from incident descriptions and converts them into a structured chronological timeline.
The timeline can then be used to populate sections of agency reports, assist in incident review, and enable additional validation or analytics across reports.
💡 Rationale
Emergency reports frequently require a clear timeline of events. However, this information is usually embedded in free-form text and must be manually interpreted.
Automatically extracting timelines provides several benefits:
- Improves report clarity by structuring events chronologically
- Enables validation of event ordering (e.g., arrival before containment)
- Allows downstream analytics or incident comparisons
- Enhances FireForm’s ability to produce structured, multi-agency reports
- Supports more advanced analysis of incident response patterns
This capability aligns with FireForm’s mission of transforming natural language incident descriptions into structured data that can populate multiple agency forms.
🛠️ Proposed Solution
Introduce a timeline extraction module that processes incident descriptions after the initial LLM extraction stage.
Proposed pipeline:
Incident Description
→ LLM Structured Extraction
→ Timeline Extraction Layer
→ Structured JSON + Timeline
→ PDF Form Filling
Example input:
"Engine 12 arrived at 3:10 PM. Fire contained at 3:25 PM. Police arrived at 3:30 PM."
Example structured output:
{
"timeline": [
{"event": "engine_arrival", "time": "15:10"},
{"event": "fire_contained", "time": "15:25"},
{"event": "police_arrival", "time": "15:30"}
]
}