MaplePath
Benchmark

Does the occupation search actually work?

GET /evaluate

Classifying a job description into one of the NOC 2021 unit groups is the one place in MaplePath where a wrong answer is plausible and consequential — it decides TEER, and TEER decides program eligibility. So retrieval is measured, not asserted. 9 job descriptions, hand-labelled with their correct NOC code, run through four retrieval methods with identical queries.

Correct NOC in top 10
100%
9 of 9 descriptions
Ranking quality
0.673
NDCG@10, shipped method
Lift over keyword search
+0.082
NDCG vs BM25 alone
Misses
0
hybrid method, outside top 10
Each method against the same 9 labels

Higher NDCG@10 means the correct code sits nearer the top of the list. The shipped method is the last row.

BM25 full-textkeyword matching only0.591hit 1.00
pgvector semanticembedding similarity only0.663hit 1.00
Reciprocal rank fusionboth lists merged0.650hit 1.00
Hybrid + Cohere rerankshipped0.673hit 1.00
Per-example report

One row per labelled description, from examples_report. NDCG and hit are for the shipped hybrid + rerank method.

Job titleCase typeNDCGOutcome
Software Developerexact_title0.865Top 10
Software Engineerclose_sibling0.824Top 10
IT Support Technicianvague_title0.694Top 10
Bookkeeperteer_distinction0.476Top 10
Registered Nurseexact_title0.723Top 10
Licensed Practical Nurseclose_sibling0.498Top 10
Cookclose_sibling0.596Top 10
Electricianexact_title0.820Top 10
Transport Truck Driverexact_title0.558Top 10

Labels live in data/noc_eval_labels.csv. This report is produced live by the /evaluate endpoint on each visit.