Launching today
resolve()
The entity API that knows when not to answer
1 follower
The entity API that knows when not to answer
1 follower
resolve() tells your agent which real-world thing a name refers to: Jaguar the cat or the carmaker, Orange the telecom or the French town. Send text and get a stable ID for each mention, from Wikidata, your knowledge graph or your own vocabulary. When the evidence isn't there, it returns “unresolved” with a reason instead of guessing. Try three live agent demos with no sign-up. Free for individual developers: 1,000 credits a month. The pipeline and clients are open source (Apache-2.0).
Hi Product Hunt, I'm Andrea, co-founder of @wordliftit.
A jaguar swam through the Amazon rainforest. In our demo, a search-based agent files that story under automotive news, because the first hit for "jaguar" is Jaguar Cars. Nothing crashes. The wrong fact just moves into the graph, the markup and the agent's memory.
We have spent years linking text to knowledge graphs, and that one decision, "which thing is this?", is where it goes wrong. resolve() is that decision as an API:
Send text and get a stable ID for every mention, from Wikidata by default
When the evidence is thin, it answers "unresolved" and tells you why
Point it at your own knowledge graph or vocabulary and get your own IDs back
You can run three live agent demos right now with no sign-up, and edit the text to test your own cases. A free key gives individual developers 1,000 credits a month. The pipeline and clients are open source.
We also published our benchmark against Google, including the languages where we lose: https://wordlift.io/resolve/#benchmarks
One thing I would love to hear: where would a wrong identity hurt most in what you are building?
Demos: https://wordlift.io/resolve/demos/
Free key: https://wordlift.io/resolve/#access
GitHub: https://github.com/wordlift/content-analysis