USC CSCI 585 · Database Systems

CSCI 585 — Database Systems (USC, Spring 2023)

Two interactive demos from Yan Liu's CSCI 585 alongside the original homework source. The Bay Area photo spatial demo plots 13 geotagged photos on a real map and runs convex-hull / k-NN queries client-side via turf.js. The YouTube SQL demo runs the original 7-table relational schema on PGlite (real PostgreSQL compiled to WebAssembly), letting you execute the 6 query exercises live in the browser. Original homework source lives alongside in HWs/HW{1..5}/.

View on GitHub
Works
Demos
HW3 — Bay Area Photo Spatial QueriesHW2 — YouTube SQL Schema (PGlite)
Loading embed…
Open

13 geotagged photos around Fremont/Pleasanton/Livermore on a maplibre-gl basemap. Toggle convex hull, click anywhere to find the 4 nearest photos, enable density to see a Gaussian kernel-density visualization. Same dataset as the original PostGIS HW3, computed client-side via turf.js (no server-side DB required).
sql postgresql pglite spatial postgis geospatial wasm usc

Overview

README.md ↗

My homework + project work from Yan Liu's CSCI 585 at USC.

Repo history note: the original HWs/HW4/.env and HWs/HW3/spatialdbqueries.py contained live credentials for a personal MongoDB Atlas cluster and a Supabase Postgres instance. Both were rotated and the strings scrubbed from git history via git filter-repo on 2026-05-28 before the repo was made public. The current .env.example and env-driven Python code show how to point the demos at your own DB. See the audit notes inside HWs/HW3/spatialdbqueries.py.

Live demos

The embeds/ directory holds two interactive visualizations:

  • Bay Area Photo Spatial Queries (HW3) — 13 geotagged photos, convex hull, k-NN, KDE. Same logic as the original PostGIS code, but computed in-browser with turf.js so no DB is needed.
  • YouTube SQL Schema (HW2) — the 7-table relational schema running on real PostgreSQL via PGlite. All 6 HW2 query exercises are pre-loaded.

These are rendered on codeseys.io/projects/csci-585 — the demo URLs are versioned per git-sha so old commits stay reachable.

Layout

Path What
HWs/HW1/ Conceptual modeling (ER → relational)
HWs/HW2/ YouTube schema + SQL queries (q1.sql … q6.sql)
HWs/HW3/ KML + PostGIS spatial queries
HWs/HW4/ MongoDB Atlas + KML loader (env-driven; see .env.example)
HWs/HW5/ Image classification (chihuahua-vs-muffin etc.)
Lectures/ Annotated lecture notes
Exams/ Midterm + final review notes
embeds/ Interactive web demos (rendered on codeseys.io)
web.codeseys.json Embed manifest (consumed by personal-site discovery)