Hacker News new | comments | show | ask | jobs | submit login

Looking at the code, it does less than is claimed. The add-on scrapes a Facebook page and calls their server. All the real work is in [1]. There seem to be just two tests, both applied to images - 1) is it an "adult" image, and 2) if it has text, does the text match the indicated Twitter feed?

It uses some Microsoft Project Oxford APIs to do all the image work. There's an "adult image" test API, and an OCR API.

[1] https://github.com/anantdgoel/HackPrincetonF16/blob/master/b...




It also checks the url of the post content in https://www.mywot.com/, which is some kind of trust service. I don't know the specifics of it, but it looks like it's basically a whitelist of known trustworthy sources. And I have no idea about the quality of the results.

I'd say that piggybacking this service would match the goal of the project, and the other stuff is for image macros/memes etc.

Actually pretty decent for a small project like this, but the title definitely makes it sound like it's looking at arbitrary text and evaluating it for truthfulness. Really it's just seeing if the source is reliable.


"Really it's just seeing if the source is reliable"

Not sure about that.

Does adult content make a source unreliable? Remember this? [1] It was flagged for adult content on fb, and there was a lot of controversy about how it shouldnt have been removed.

It also checks if there are tweets referenced in the article, and if so, checks to see it actually exists on twitter. What about a hypothetical story about someone saying something controversial on twitter, then deleting it? Is that news? Because that story would be marked as fake since the tweet no longer exists.

Then there is this story. The article "makes it sound like it's looking at arbitrary text and evaluating it for truthfulness". So would this be marked as fake news? Does the wapo domain get added to a blacklist?

1. www.nytimes.com/2016/09/10/technology/facebook-vietnam-war-photo-nudity.html




Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | DMCA | Apply to YC | Contact

Search: