Show HN: I discovered roads in the US across > 1000 themes
A Show HN project maps US roads to over a thousand themes using Wikidata, though commenters quickly point out widespread spurious matches and automated noise.
summary is enough The concept is neat, but the discussion and implementation details reveal pervasive false positives that make full exploration unnecessary.
summary
A developer launched a tool that categorizes US roads across more than a thousand themes by cross-referencing Wikidata entities with street names. The comment thread quickly uncovers massive false positive rates, where common words and unrelated names are misattributed to specific pop culture or historical references. The creator admits that loose SPARQL queries and automated string matching allowed substantial noise into the results. Commenters criticize the lack of human curation and the poor quality of the output.
key points
- The web tool claims to map US road names across 1,068 themes using automated queries and string matching.
- Commenters demonstrated numerous egregious false matches, such as common words linked to TV show episodes or unrelated historical figures.
- The creator confirmed the use of automated SPARQL queries and BM25 matching, acknowledging that the association thresholds were far too broad.
who should read it
Engineers and data hobbyists interested in geographic data scraping, entity resolution, and the pitfalls of naive automated matching.
discussion skeptical
consensus
Commenters overwhelmingly criticize the project for generating low-quality, nonsensical matches through poorly tuned automated queries and unverified data generation.
pushback
A few participants found the core idea amusing or inquired about the methodology before seeing the flaws.
ouch, and also thank you for calling out the busted.
An LLM titles each theme, but otherwise this is all derived by checking wikidata entities that are 'most read' (according to qrank), then choosing from a set of sparql queries I put together (members of group X, places within Y, etc), for what produces the most results, and then string/substring/bm25 matching across all the streets in the country.
I think the trash smell is too generously allowing an association both in my sparql queries (what to search for), and whats considered a match :/
I wonder too if explaining how a street matched …
The project creator steps in to explain the exact technical pipeline using Wikidata, SPARQL, and BM25 matching, admitting the search criteria were too loose.
Generated by an LLM from the article and thread. Check the sources before relying on it.