GraphWise

AI Instruction (Step 1)

You are an expert in **concept mapping**—the practice of representing understanding as discrete **concepts** (nouns) connected by **verbs** (relationships) and organized in a logical hierarchy from general to specific.

Read the text carefully and extract only the information explicitly contained in it to build a **concept map** (a knowledge graph that expresses relationships of meaning rather than simple co-occurrence).  
Do not add external knowledge or assumptions.

Output MUST be in Markdown using exactly the following section headings and formats, in this order:

## Topic
<one concise sentence naming the main subject or focus of the text>

## Framing
<one conceptual "armature" sentence that expresses the central relationship or model—
how the topic connects to its 2–3 most important ideas using verbs.
This framing should suggest the map's top-level structure and main branches.>

## Summary
<one short paragraph (3–5 sentences) synthesizing the key ideas, their relationships, and the overall significance of the text.>

## Nodes
Each node represents a single, well-defined concept.
Use a **noun or short noun phrase** (1–4 words) in **Title Case** that names an idea, object, or process—no verbs, clauses, or descriptive sentences.

Format:
- id: <unique lowercase-hyphenated string>; label: <Canonical Concept Name>; group: <Theme>

Example:
- id: concept-map; label: Concept Map; group: Purpose
- id: shared-understanding; label: Shared Understanding; group: Output

Normalization Rules:
- Node ids must be lowercase, kebab-case, unique, and derived deterministically from the label.
- Labels must be in Title Case with only letters, digits, spaces, and hyphens.
- Each node must belong to one short thematic group (1–3 words).
- Typical total size: 8–20 nodes; prefer fewer if the text is short or repetitive.

## Links
Each link expresses a **meaningful relationship** between two nodes.

Format:
- source: <id>; target: <id>; label: <active verb phrase>; weight: <number 0.0–1.0>

Guidelines:
- Relation labels must begin with an active verb (e.g., "enables", "depends on", "is part of", "causes", "leads to").
- Direction reflects conceptual flow (cause → effect, whole → part, general → specific).
- Weights represent relative strength or importance:
  - 1.0 = explicit core relation
  - 0.7 = strong implication
  - 0.4 = weak implication
  - 0.1 = tentative or minor
- Each node must appear in at least one link (as source or target).

## Hierarchy
Represent the conceptual hierarchy as an indented list from general to specific.

Example:
- name: <Topic>
  - name: <Major Branch>
    - name: <Subconcept>

Rules:
- Each node label must appear exactly once somewhere in the hierarchy.
- The hierarchy root must be the **Topic**.
- First-level branches must align with the main ideas from the **Framing**.
- Numeric IDs must be quoted as strings when used in IDs.
- Maintain consistency between hierarchy names and node labels.

If the text lacks sufficient information, produce the smallest faithful graph (≤6 nodes) and note uncertainty in the Summary.

Input Text:
{{{text}}}

Source Content

Step 1 Model

Step 1: Intermediate Text

Step 2 Model

Target JSON Schema (Step 2)

{

"$ref":"#/definitions/GraphData"

"definitions":Object { ... }

"$schema":"http://json-schema.org/draft-07/schema#"

}

Step 2: Final JSON Output

Validated JSON output will appear here.