4 ms·
Google has developed advanced NLP algorithms, so it should be easy for them to corroborate such data before it gets pushed to featured snippets. On second thoug
by superasn 10y ago
Google has developed advanced NLP algorithms, so it should be easy for them to corroborate such data before it gets pushed to featured snippets. On second thought, maybe they're already doing it and sites including authority are just copying off each other.
- rspeer 10y agoWhat makes you think that "advanced NLP algorithms" can distinguish true statements from false ones? They're not even close. You'd have to dramatically simplify the scope of the problem to make it not require artificial general intelligence. The "Fake News Challenge" [1] is the first attempt I know of to seriously evaluate NLP that could eventually be able to fact-check real-world statements, and the current task is about stance detection, because the goal of fact-checking is not considered attainable yet. [1] http://www.fakenewschallenge.org/ http://www.fakenewschallenge.org/
- superasn 10y agoThey were building a knowledge graph for this purpose[1] [1] http://searchengineland.com/demystifying-knowledge-graph-201976 http://searchengineland.com/demystifying-knowledge-graph-201...
- rspeer 10y agoI'm pretty sure these horrible snippets come from the Google Knowledge Graph.
- elcapitan 10y agoNot false or true, but it would already help to check if the text is an opinion piece. Lingustic hints for that should be relatively clear, as opposed to for example encyclopaedic style. That way they could both exclude most Breitbart stuff as well as editorials from actual newspapers, which shouldn't make it into quick-info fact boxes either.