Overview
What this challenge is about.
Build a knowledge-graph recommender for a Lisbon bookstore. Return top 10 related titles using traversal rules and similarity scores. Earn a verifiable certificate.
The scenario
The file documents an internal request at Página Lusa, a de-identified Lisbon online bookstore of about 22 staff and roughly EUR 4M revenue focused on Portuguese-language literature, whose catalog of about 8,000 titles is small but carries high-quality author, genre, era, theme, and award metadata. To date, related-title suggestions have been produced by hand as weekly 'staff picks', which the team can no longer scale.
The Brief
What you'll do, and what you'll demonstrate.
Can a rules-plus-similarity recommender built over a catalog knowledge graph match the team's hand-curated staff-pick quality, measured as precision-at-10 within 15 percent, without using any cross-customer purchase history?
Earning criteria — what you'll demonstrate
- Represent a literary catalog as a knowledge graph with clear node types and edge semantics.
- Combine symbolic traversal rules with numeric similarity scoring into a single ranked recommendation.
- Design and run a precision-at-10 evaluation against a curated baseline using a held-out split.
- Translate a working service into an integration spec a frontend team can build against unaided.
Program Fit
Where this fits in your program.
Sharpens the same skills your degree expects you to demonstrate.
Aligned coursework coming soon.
Skills
Skills you'll demonstrate.
Each one shows up on your verified credential.
- Knowledge Representation
Apply knowledge representation to solve real industry problems and demonstrate production-level capability.
- Knowledge Graphs
Apply knowledge graphs to solve real industry problems and demonstrate production-level capability.
- Python Programming
Apply python programming to solve real industry problems and demonstrate production-level capability.
- Algorithm Evaluation
Apply algorithm evaluation to solve real industry problems and demonstrate production-level capability.
- Recommender Systems
Apply recommender systems to solve real industry problems and demonstrate production-level capability.
- Intelligent Agents
Apply intelligent agents to solve real industry problems and demonstrate production-level capability.
Careers
Career paths this challenge builds toward
Completing this challenge demonstrates skills that transfer directly to these roles:
Recommender Systems Engineer
This challenge mirrors real recommender work: turning rich item metadata into a ranked, explainable recommendation and proving its quality against a baseline. You leave with a portfolio piece showing you can ship a content-based recommender end to end.
This challenge sharpens
- recommender-systems
- knowledge-graphs
- algorithm-evaluation
Knowledge Engineer
Modeling a domain as nodes, edges, and traversal rules is the core of knowledge engineering. By designing a clean schema and reasoning over it, you practice the representation and rule-design skills these roles depend on daily.
This challenge sharpens
- knowledge-representation
- knowledge-graphs
- intelligent-agents
Backend Engineer (Search & Discovery)
Discovery teams need backend engineers who can build a service, score results, and hand a clean integration spec to frontend. This challenge exercises exactly that path from data model to deployable service.
This challenge sharpens
- python-programming
- recommender-systems
- algorithm-evaluation