Regulatory Topics

What are Topics?

Topics are cross-cutting regulatory domains that span multiple agencies and CFR titles. Unlike agencies (which own specific chapters), topics identify thematic areas like "Environmental Protection" or "Privacy & Data" across the entire regulatory landscape.

Search Technology:

Topics use SQLite FTS5 (Full-Text Search) with BM25 ranking—the same relevance algorithm used by Elasticsearch and Apache Solr. This provides intelligent keyword matching with several advantages over simple substring search:

• Word boundary matching: Searches for complete words only. "air" matches "air quality" but not "chairman" or "repair"
• Automatic stemming: Porter stemmer matches word variations: "environment" finds "environmental", "environments", "environmentally"
• Relevance ranking: BM25 scores sections by keyword frequency and document length, surfacing the most relevant matches first
• Boolean queries: Topics combine multiple keywords with OR logic to capture related regulatory domains

Data Quality:

The FTS5 implementation significantly improves precision compared to naive substring matching. Example: "Environmental Protection" now matches 2,541 sections instead of 3,796—eliminating ~33% false positives (1,255 sections that contained substrings like "chair" or "fair" but weren't actually about environmental regulation).

This means you can trust that sections tagged with a topic are genuinely relevant to that regulatory domain, making cross-agency analysis more reliable.