Implement Eric Kansa's authoritative iSamples queries in Cesium tutorial

[rdhyee]

Summary

Replaces the previous UNION-based query approach with Eric Kansa's exact query implementations from open-context-py/isamples_explore.py.

Queries Implemented

✅ 1. get_samples_at_geo_cord_location_via_sample_event

• Path 1 only (backward walk from GeospatialCoordLocation via sample_location)
• INNER JOIN for site (required)
• Orders by has_thumbnail DESC
• Returns: sample details, geo coordinates, site context

✅ 2. get_sample_data_via_sample_pid

• Get full sample metadata including geo and site info
• Forward walk: sample → produced_by → event → sample_location → geo

✅ 3. get_sample_data_agents_sample_pid

• Get agent information (who collected/registered)
• Returns responsibility and registrant predicates

✅ 4. get_sample_types_and_keywords_via_sample_pid

• Get classification keywords and types
• Returns keywords, has_sample_object_type, has_material_category

Key Changes

Query Strategy

• Old: UNION of Path 1 + Path 2 (combined direct and site-based location paths)
• New: Path 1 only (exact match to Eric's implementation)

Behavior Changes

• Site marker locations (Path 2 only) now correctly return 0 results
• Example: geoloc_7a05216d388682536f3e2abd8bd2ee3fb286e461 (Larnaka site marker) returns 0
• Field collection points (Path 1) return samples as expected
• Example: geoloc_04d6e816218b1a8798fa90b3d1d43bf4c043a57f (PKAP) returns 5 samples

Testing

Tested all 4 queries with PKAP survey area data:

# Test Case 1: Geo location with Path 1 connections
geo_pid = "geoloc_04d6e816218b1a8798fa90b3d1d43bf4c043a57f"
✓ Returns 5 samples (PKAP Survey Area)

# Test Case 2: Sample metadata  
sample_pid = "ark:/28722/k2wq0b20z" 
✓ Returns 1 record with full metadata

# Test Case 3: Agent data
✓ Returns 3 agents (R. Scott Moore, etc.)

# Test Case 4: Keywords
✓ Returns 4 keywords (pottery, amphora, Artifact, ceramic)

Files Changed

• tutorials/parquet_cesium.qmd (+196, -54 lines)

🤖 Generated with Claude Code