Upload ml/schema.sql with huggingface_hub
Browse files- ml/schema.sql +57 -0
ml/schema.sql
ADDED
|
@@ -0,0 +1,57 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
-- Crisis Correlation Schema Extensions
|
| 2 |
+
-- Adds historical events and document-event linking
|
| 3 |
+
-- Existing tables: entities, document_topics, document_keywords, document_features, page_features
|
| 4 |
+
|
| 5 |
+
BEGIN;
|
| 6 |
+
|
| 7 |
+
-- Historical events reference table
|
| 8 |
+
CREATE TABLE IF NOT EXISTS historical_events (
|
| 9 |
+
id SERIAL PRIMARY KEY,
|
| 10 |
+
event_name TEXT NOT NULL UNIQUE,
|
| 11 |
+
start_date DATE NOT NULL,
|
| 12 |
+
end_date DATE,
|
| 13 |
+
category VARCHAR(50),
|
| 14 |
+
description TEXT,
|
| 15 |
+
keywords JSONB DEFAULT '[]'
|
| 16 |
+
);
|
| 17 |
+
|
| 18 |
+
-- Many-to-many: documents <-> events with relevance scoring
|
| 19 |
+
CREATE TABLE IF NOT EXISTS document_events (
|
| 20 |
+
document_id INT NOT NULL REFERENCES documents(id) ON DELETE CASCADE,
|
| 21 |
+
event_id INT NOT NULL REFERENCES historical_events(id) ON DELETE CASCADE,
|
| 22 |
+
relevance_score REAL NOT NULL DEFAULT 0,
|
| 23 |
+
match_methods JSONB DEFAULT '[]',
|
| 24 |
+
details JSONB DEFAULT '{}',
|
| 25 |
+
created_at TIMESTAMP DEFAULT NOW(),
|
| 26 |
+
PRIMARY KEY (document_id, event_id)
|
| 27 |
+
);
|
| 28 |
+
|
| 29 |
+
CREATE INDEX IF NOT EXISTS idx_document_events_event ON document_events(event_id);
|
| 30 |
+
CREATE INDEX IF NOT EXISTS idx_document_events_score ON document_events(relevance_score DESC);
|
| 31 |
+
|
| 32 |
+
-- Document-level date estimates (extracted from filename, OCR, entities)
|
| 33 |
+
CREATE TABLE IF NOT EXISTS document_dates (
|
| 34 |
+
document_id INT PRIMARY KEY REFERENCES documents(id) ON DELETE CASCADE,
|
| 35 |
+
estimated_date DATE,
|
| 36 |
+
date_source VARCHAR(30),
|
| 37 |
+
date_confidence REAL,
|
| 38 |
+
date_range_start DATE,
|
| 39 |
+
date_range_end DATE,
|
| 40 |
+
congress_session INT,
|
| 41 |
+
created_at TIMESTAMP DEFAULT NOW()
|
| 42 |
+
);
|
| 43 |
+
|
| 44 |
+
CREATE INDEX IF NOT EXISTS idx_document_dates_date ON document_dates(estimated_date);
|
| 45 |
+
CREATE INDEX IF NOT EXISTS idx_document_dates_congress ON document_dates(congress_session);
|
| 46 |
+
|
| 47 |
+
-- Track pipeline runs
|
| 48 |
+
CREATE TABLE IF NOT EXISTS ml_pipeline_log (
|
| 49 |
+
id SERIAL PRIMARY KEY,
|
| 50 |
+
pipeline TEXT NOT NULL,
|
| 51 |
+
started_at TIMESTAMP DEFAULT NOW(),
|
| 52 |
+
finished_at TIMESTAMP,
|
| 53 |
+
docs_processed INT DEFAULT 0,
|
| 54 |
+
status TEXT DEFAULT 'running'
|
| 55 |
+
);
|
| 56 |
+
|
| 57 |
+
COMMIT;
|