COMPASS Transfer GPT Tutorial
This GPT performs a structured, concept-level comparison using your curriculum map and a syllabus: COMPASS Transfer Analyzer.
For the tutorial below, I use the following testing data.
- A curriculum map for the department that is evaluating an incoming syllabus
- An incoming syllabus from an external institution (this was downloaded from the web)
Required Inputs
COMPASS requires two inputs:
| # | Required input | Must include | Allowed values / meaning |
|---|---|---|---|
| 1 | Curriculum Map CSV file | A column named concept; one column per course |
Cell values must be 0, 1, 2, or 3: 0 = not taught; 1 = introduced; 2 = developed; 3 = mastered |
| 2 | Incoming transfer course syllabus | A complete syllabus for the transfer course | N/A (must be complete enough to extract topics/skills for mapping) |
If either input is missing, COMPASS will stop and request the missing information.
Step-by-Step Workflow
| Step | Name | What you do | Output |
|---|---|---|---|
| 1 | Upload curriculum map CSV | Upload a CSV that includes a concept column and one column per course with values 0–3 |
Curriculum concept list + course coverage matrix ready for mapping/scoring |
| 2 | Provide full syllabus text | Paste the complete transfer-course syllabus text (topics, weekly schedule, textbook, learning outcomes) | Syllabus content available for topic extraction |
| 3 | Concept Mapping | Extract topics from the syllabus → map to official concept labels; prefer exact matches; allow conservative near matches only with clear wording overlap; label the rest Unmapped | Mapped concept set \((A)\) (all mapped labels; excludes Unmapped) |
| 4 | Overlap Scoring | For each course \((Y)\) compute: \(s(Y)=|A \cap X_Y|\) (shared concepts). If tied, compute: \(w(Y)=\sum_{c\in A}\mathrm{level}(c,Y)\). | Recommended course = highest \(s(Y)\); ties broken by \(w(Y)\) |
Understanding the Output
COMPASS produces four structured sections:
| Section | What to include | Fields / measures |
|---|---|---|
| 1) Executive Summary | Recommended equivalency + why | Recommended equivalency, Best-match course set \(\Phi_A\), Reason for recommendation |
| 2) Mapping Table | How syllabus content maps to curriculum concepts | Syllabus phrase, Matched concept, Match type (Exact / Near), Justification |
| 3) Ranked Course Table | How each course scores against the mapped set | Overlap score \(s(Y)\), Weighted score \(w(Y)\) |
| 4) Diagnostics | Evidence and gaps supporting the decision | Overlap concepts, Missing concepts, Extra concepts, Coverage level profile |
All results are fully auditable.
Optional Strict Mode
If you type “Strict mapping”, only exact matches are allowed.
This is recommended for high-stakes articulation reviews.
Best Practices
– Keep your curriculum map clean and consistent.
– Use precise concept labels.
– Expand your concept list if critical topics frequently appear as “Unmapped”.
– Remember: Proposed extensions are not used in scoring.
Mathematical Framework (Advanced)
| Symbol / term | Meaning |
|---|---|
| \((L)\) | Official concept list (the set of concept labels from the curriculum map’s concept column) |
| \((A)\) | Mapped syllabus concepts (the set of concept labels successfully mapped from the incoming syllabus; excludes Unmapped) |
| \((X_Y)\) | Concepts taught in course \((Y)\) (all concepts in \((L)\) whose curriculum-map value for course \((Y)\) is \(>0\)) |
| \((s(Y))\) | Overlap score for course \((Y)\): \(s(Y)=|A \cap X_Y|\) |
| \((w(Y))\) | Weighted overlap score for course \((Y)\): \(w(Y)=\sum_{c\in A}\mathrm{level}(c,Y)\) (sum of coverage levels for concepts in \((A)\) for course \((Y)\); equivalently sum over overlapping concepts since non-overlaps contribute 0) |
| \((\Phi_A)\) | Best-match course set: all courses with maximum \(s(Y)\) across courses (ties possible) |
This ensures structured, consistent transfer decisions.