Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Show HN: I made a neural net that analyzes privacy policies

useguard.com

Post reply on HN
Print view View original

111–120 of 122 posts

Re: Show HN: I made a neural net that analyzes privacy policies

#111

Post by BrS96bVxXBLzf5B »

This is a good tool, but the execution of the website is disappointing.

1. Only showing excerpts of the highest threat levels. Trying to view the less severe threats asks us to email in. If you're willing to volunteer the information, why the hoops?

2. "Play a short game to continue using this tool" ensures I'm not going to share this with anyone. Putting a stranglehold on users is _never_ the way forward. I might have volunteered my time if I were at home and browsing through. I can't when I'm quickly flicking through during taking a five minute break from looking at work. But it's left me with a final negative impression before being unceremoniously blocked off.

Re: Show HN: I made a neural net that analyzes privacy policies

#112

Post by martin__ »

Hello, if the webmaster for this site is reading this, your `change.org` file is getting a Content-Type of `application/octet-stream` instead of `text/html`, which is giving me (in firefox) a prompt to download a file instead of displaying the page.

The problem is probably with your type mappings: https://nginx.org/en/docs/http/ngx_http_core_module.html#typ...

Re: Show HN: I made a neural net that analyzes privacy policies

#113

Post by alexcnwy »

rameerez wrote: post #72
goldemerald wrote: post #71

Earlier quoted context omitted.

Does your work integrate pretrained LMs like BERT or GPT2?

Yes, but not Transformer-based like these two, rather LSTM-based like ULMFiT

Great work - would love to read your thesis if it’s available online?

Re: Show HN: I made a neural net that analyzes privacy policies

#114

Post by mceachen »

raynr wrote: post #94
foxes wrote: post #87

The "game" (training) that asks you to analyse "privacy threats" is a bit strange. It feels like it takes two random excerpts from a privacy policy and asks you to compare them, but with this, it feels like it is missing some global information, you are just looking at local details. Eg one policy might be disclosing what they do (but its actually relevant to collect data, eg password manager) while the other just sa…

Yeah, it's not comparing like for like. Feels like the system is trying to collect training data from users. To elaborate on your last sentence, context is critical in assessing whether a clause is pro- or anti- privacy. Is the collection of information critical to the provision of the service? What is collected, and how much? And so on.

Agreed with parent and gp. I gave up with the 10 questions, as the sentence comparisons were almost comically incomparable. I fear your model is going to be a random number generator.

Re: Show HN: I made a neural net that analyzes privacy policies

#115

Post by contravariant »

rameerez wrote: post #17
the_watcher wrote: post #14

Telegram has a 105% score? Is that expected or a bug?

It should be pretty close to 100%, though, they seem to be super privacy friendly!

>BIGGEST THREAT: «We never delete your funny cat pictures, we love them too much»

I don't know, seems like a GDPR violation to me ;-).

Re: Show HN: I made a neural net that analyzes privacy policies

#116

Post by abathur »

epoch_100 wrote: post #28

Hi. A related project that takes a more human-powered approach is PrivacySpy ( https://privacyspy.org ). Would be neat to see how these tools intersect. PrivacySpy is open source, community run, and more about grading policies on a standardized rubric (as opposed to entrusting that to ML), so these tools might complement one another. (Full disclosure: I'm a contributor to PrivacySpy.)

On the off chance the various responders in this sub-thread see this comment, for a while I've hoped someone would advocate for privacy/TOS policies to follow a similar model to OSS licensing.

Normal people could conceivably read and understand a given policy if the knowledge scaled.

Any substantial adoption would help focus effort/resources on services that deviate from the terms.

Sites that are already repositories of this knowledge could play some part in codifying best-practices, advocating for adoption, and tracking progress.

Re: Show HN: I made a neural net that analyzes privacy policies

#117

Post by fny »

rameerez wrote: post #53
brianberns wrote: post #36

Earlier quoted context omitted.

How did you create a data set large and accurate enough to be useful in training a model?

Some friends run an AI bootcamp and helped me finding the initial set of users to help me with labelling. Initial labelled data was generated mostly through them, both manually labelling and with the approach described in https://useguard.com/experiment Also, the model I'm using relies heavily in transfer learning and achieves very reasonable results with few labelled items (the paper in which the technique is descri…

What paper is this? Is it the UDA paper?

Re: Show HN: I made a neural net that analyzes privacy policies

#119

Post by tboyd47 »

Hi there. This is really neat. It reminds me of a talk I listened to recently on digital privacy, where the guy was using the price of "privacy products" as a way to measure how much people value their privacy. This seems like it would be one of those.

Did you find that there's one single variable, like length or the presence of certain words, that the system relies on heavily?

Re: Show HN: I made a neural net that analyzes privacy policies

#120

Post by mcbits »

gempir wrote: post #99
simongr3dal wrote: post #42

Earlier quoted context omitted.

More and more websites are so advanced that they can’t even use an tag anymore. Instead they do some convoluted onclick-scripting that breaks all standard behavior and accessibility functionality.

I think this is mostly a problem of the tools used. An anchor tag by definition is an inline element, so it shouldn't really be a giant box that's clickable, so you default back to an onClick. onClick in popular frameworks just means left click and nothing more, which makes sense, except for that use case of opening in a new tab with middle mouse button or right clicking etc. So you have to add a lot of logic to supp…

Styling an anchor as a block element has never violated the spec, and HTML5 deliberately added support for wrapping anchors around block elements because people had been doing that anyway, even though the browser wasn't required to make it work as intended (but it now is).
Post reply on HN