Ask HN: Paul Graham Parallel Corpus
TLDR: Are you willing to rewrite a pg essay in your own style? Goal is parallel style corpus for authorship obfuscation tools.
PG on Twitter today: "writers are leaving a damning trail that they don't realize they're leaving. This text is going to be around forever for software to analyze." [1]
People must be free to speak anonymously, without retaliation from monopolist employers.
Authorship obfuscation is currently weak. Research could be improved by building parallel corpus with single target style, many source styles. Diversity of source styles likely important.
This is not much work for a big crowd.
[1] https://twitter.com/paulg/status/1320703313818619905
No comments yet.