BigQuery Graph Peer Review Data Demo

This repository contains SQL DDL, seed data, and property graph definitions for demonstrating BigQuery Graph modeling, cross-domain data mesh queries, and GQL pattern matching for academic peer review and editorial workflows.

---

Architecture Overview

The demo models an academic editorial and peer review ecosystem split across two decoupled BigQuery datasets:

1. academic_editorial (Public / Editorial Domain):
   • Nodes: Paper, Subject
   • Edges: Cites (Paper → Paper), HasSubject (Paper → Subject)
2. peer_review (Sensitive Profile / Reviewer Domain):
   • Nodes: Author, Institution
   • Edges: Writes (Author → Paper), Affiliated (Author → Institution), SpecializesIn (Author → Subject)

A logical Property Graph catalog (academic_editorial.ScholarlyCentralGraph) is layered directly on top of these standard relational tables without data replication or proprietary graph infrastructure.

---

Execution Workflow

Execute the SQL scripts in numerical order within BigQuery:

1. demo_central_01_create_tables.sql

Creates the underlying physical relational tables across both academic_editorial and peer_review datasets, including primary and foreign key constraints (not enforced).

2. demo_central_02_seed_data.sql

Populates entity nodes and relationship edges. Sets up a scenario featuring:

• Co-authorship and institutional relationships (Dr. Alice Smith and Dr. Bob Johnson at Harvard).
• Independent researcher with matching specialization (Dr. Charlie Brown at Cambridge).

3. demo_central_03_create_property_graph.sql

Defines the CREATE OR REPLACE PROPERTY GRAPH academic_editorial.ScholarlyCentralGraph DDL catalog entry, mapping relational tables to nodes and edges.

4. demo_central_04_verify_queries.sql

Executes native GQL queries using BigQuery's GRAPH_TABLE operator:

• Query 1: Multi-edge researcher expertise and institutional directory traversal.
• Query 2: Multi-hop citation network traversal (Paper -[:Cites]-> Paper).
• Query 3: Peer reviewer matching with Conflict of Interest (COI) elimination (excluding co-authors and institutional conflicts).