Dear all,
I started using WikiData for Private Information Retrieval. This allows answering certain questions while maintaining a high degree of secrecy. Suppose you wanted to know when Einstein was born, but for some reason you must keep the fact that you want to know this a secret. In this case, we assume a threat model with perfect knowledge about the computer in use, not just that someone managed to log the Wikidata requests.
One solution would be to request a table of every human who ever won the Nobel Prize (this requires the common knowledge of Einstein being a winner of said) with the kind of Nobel Prize, date awarded, date of birth, date of death unless living, nationality etc. If we let this table scroll across the screen and read the required entry, there would be - as far as I can tell - no way to learn which entry (and how many of them) is of our interest.
I have written a simple PHP script that allows one to enter a SPARQL request and have it displayed as a scrolling table. Please be advised that this is in a very informal alpha state, and I am no professional Web Developer. It's a mere proof-of-concept, but feel free to try it out if the API quota lets you.
https://darius-runge.eu/otp/request.php
My question is, whether anyone of you might be interested in working with me on discussing practical implications of this method (how should requests be written to allow for the desired privacy?) or even making a better implementation of a tool that allows viewing the scrolling table of printing it out.
Feel free to reply to this mailing list entry or contact me privately with the postal or telecommunication data provided in the footer in case you don't want to discuss it in public.
Best, Darius
---------------------------------------- Darius Runge Postfach 3 72669 Unterensingen Germany
Tel +49 7022 5064970 Fax +49 7022 5064971 Vox +49 7022 5064998 (2 min)
All up-to-date contact data: https://darius-runge.eu
Aren't similar methods too much resource intensive from Wikidata's perspective?
Vito
Il giorno sab 29 gen 2022 alle ore 13:44 Darius Runge < darius-runge@magenta.de> ha scritto:
Dear all,
I started using WikiData for Private Information Retrieval. This allows answering certain questions while maintaining a high degree of secrecy. Suppose you wanted to know when Einstein was born, but for some reason you must keep the fact that you want to know this a secret. In this case, we assume a threat model with perfect knowledge about the computer in use, not just that someone managed to log the Wikidata requests.
One solution would be to request a table of every human who ever won the Nobel Prize (this requires the common knowledge of Einstein being a winner of said) with the kind of Nobel Prize, date awarded, date of birth, date of death unless living, nationality etc. If we let this table scroll across the screen and read the required entry, there would be - as far as I can tell - no way to learn which entry (and how many of them) is of our interest.
I have written a simple PHP script that allows one to enter a SPARQL request and have it displayed as a scrolling table. Please be advised that this is in a very informal alpha state, and I am no professional Web Developer. It's a mere proof-of-concept, but feel free to try it out if the API quota lets you.
https://darius-runge.eu/otp/request.php
My question is, whether anyone of you might be interested in working with me on discussing practical implications of this method (how should requests be written to allow for the desired privacy?) or even making a better implementation of a tool that allows viewing the scrolling table of printing it out.
Feel free to reply to this mailing list entry or contact me privately with the postal or telecommunication data provided in the footer in case you don't want to discuss it in public.
Best, Darius
Darius Runge Postfach 3 72669 Unterensingen Germany
Tel +49 7022 5064970 Fax +49 7022 5064971 Vox +49 7022 5064998 (2 min)
All up-to-date contact data: https://darius-runge.eu_______________________________________________ Wikidata mailing list -- wikidata@lists.wikimedia.org To unsubscribe send an email to wikidata-leave@lists.wikimedia.org
Dear Vito,
I thought about that concern already. While I cannot provide you with a conclusive answer, many requests I did were below the 60 second timeout limit and are similar to the SPARQL examples users are encouraged to try out.
If this becomes a real thing, I would, however, prefer to host a SPARQL endpoint soley for this purpose (and with a better timeout) myself.
Best, Darius
--- Original-Nachricht --- Von: Vi to Betreff: [Wikidata] Re: Private Information Retrieval Datum: 29. Januar 2022, 14:24 An: Discussion list for the Wikidata project
Aren't similar methods too much resource intensive from Wikidata's perspective?
Vito
Il giorno sab 29 gen 2022 alle ore 13:44 Darius Runge < darius-runge@magenta.de mailto:darius-runge@magenta.de > ha scritto: Dear all,
I started using WikiData for Private Information Retrieval. This allows answering certain questions while maintaining a high degree of secrecy. Suppose you wanted to know when Einstein was born, but for some reason you must keep the fact that you want to know this a secret. In this case, we assume a threat model with perfect knowledge about the computer in use, not just that someone managed to log the Wikidata requests.
One solution would be to request a table of every human who ever won the Nobel Prize (this requires the common knowledge of Einstein being a winner of said) with the kind of Nobel Prize, date awarded, date of birth, date of death unless living, nationality etc. If we let this table scroll across the screen and read the required entry, there would be - as far as I can tell - no way to learn which entry (and how many of them) is of our interest.
I have written a simple PHP script that allows one to enter a SPARQL request and have it displayed as a scrolling table. Please be advised that this is in a very informal alpha state, and I am no professional Web Developer. It's a mere proof-of-concept, but feel free to try it out if the API quota lets you.
https://darius-runge.eu/otp/request.php
My question is, whether anyone of you might be interested in working with me on discussing practical implications of this method (how should requests be written to allow for the desired privacy?) or even making a better implementation of a tool that allows viewing the scrolling table of printing it out.
Feel free to reply to this mailing list entry or contact me privately with the postal or telecommunication data provided in the footer in case you don't want to discuss it in public.
Best, Darius
---------------------------------------- Darius Runge Postfach 3 72669 Unterensingen Germany
Tel +49 7022 5064970 Fax +49 7022 5064971 Vox +49 7022 5064998 (2 min)
All up-to-date contact data: https://darius-runge.eu______________________________ _________________ Wikidata mailing list -- wikidata@lists.wikimedia.org mailto:wikidata@lists.wikimedia.org To unsubscribe send an email to wikidata-leave@lists.wikimedia.org mailto:wikidata-leave@lists.wikimedia.org
_______________________________________________ Wikidata mailing list -- wikidata@lists.wikimedia.org To unsubscribe send an email to wikidata-leave@lists.wikimedia.org
It is an interesting idea! Are you concerned at all about how a series of queries from a known origin might be interpreted? I guess you can always route your queries through Tor, assuming Wikidata continues to allow that?
Privacy doesn’t come with that much convenience. When it does the privacy is typically compromised in other ways.
I wouldn’t use such a service but instead just process the dump files with local tools like Wikibase or KGTK or Oxigraph etc.
Darius,
I think A better effort would be to perhaps build a Wikidata extension that has those features. THAT would be more useful for all.
On Sat, Jan 29, 2022 at 9:16 AM Thad Guidry thadguidry@gmail.com wrote:
Privacy doesn’t come with that much convenience. When it does the privacy is typically compromised in other ways.
I wouldn’t use such a service but instead just process the dump files with local tools like Wikibase or KGTK or Oxigraph etc. -- Thad https://www.linkedin.com/in/thadguidry/ https://calendly.com/thadguidry/
Ugh, sorry Darius,
I should have said Wikibase client (not Wikidata) extension. Showcases: https://wikiba.se/
Besides PHP, there is Lua available which you might try your hand at (given you have some knowledge of Javascript as well for client side UI customization) 1. https://www.mediawiki.org/wiki/Extension:Wikibase_Client 2. https://doc.wikimedia.org/Wikibase/master/php/namespaceWikibase_1_1Client_1_... 3. https://doc.wikimedia.org/Wikibase/master/php/md_docs_components_client-data...
Thad https://www.linkedin.com/in/thadguidry/ https://calendly.com/thadguidry/
I started using WikiData for Private Information Retrieval.
What is the "root cause" of your problems? ( https://en.wikipedia.org/wiki/Root_cause_analysis )
Now the "privacy policy" is based on U.S. law ( https://foundation.wikimedia.org/wiki/Privacy_policy ) and as I know this is not a GDPR friendly. *"The Wikimedia Foundation is a non-profit organization based in San Francisco, California, with servers and data centers located in the U.S. If you decide to use Wikimedia Sites, whether from inside or outside of the U.S., you understand that your Personal Information will be collected, transferred, stored, processed, disclosed and otherwise used in the U.S. as described in this Privacy Policy."*
IMHO: - for most EU people will be enough a GDPR friendly ".eu" wikidata query service mirror - adding extra noise for the wikidata query is not an environmentally friendly solution.
best, Imre
Darius Runge darius-runge@magenta.de ezt írta (időpont: 2022. jan. 29., Szo, 13:44):
Dear all,
I started using WikiData for Private Information Retrieval. This allows answering certain questions while maintaining a high degree of secrecy. Suppose you wanted to know when Einstein was born, but for some reason you must keep the fact that you want to know this a secret. In this case, we assume a threat model with perfect knowledge about the computer in use, not just that someone managed to log the Wikidata requests.
One solution would be to request a table of every human who ever won the Nobel Prize (this requires the common knowledge of Einstein being a winner of said) with the kind of Nobel Prize, date awarded, date of birth, date of death unless living, nationality etc. If we let this table scroll across the screen and read the required entry, there would be - as far as I can tell - no way to learn which entry (and how many of them) is of our interest.
I have written a simple PHP script that allows one to enter a SPARQL request and have it displayed as a scrolling table. Please be advised that this is in a very informal alpha state, and I am no professional Web Developer. It's a mere proof-of-concept, but feel free to try it out if the API quota lets you.
https://darius-runge.eu/otp/request.php
My question is, whether anyone of you might be interested in working with me on discussing practical implications of this method (how should requests be written to allow for the desired privacy?) or even making a better implementation of a tool that allows viewing the scrolling table of printing it out.
Feel free to reply to this mailing list entry or contact me privately with the postal or telecommunication data provided in the footer in case you don't want to discuss it in public.
Best, Darius
Darius Runge Postfach 3 72669 Unterensingen Germany
Tel +49 7022 5064970 Fax +49 7022 5064971 Vox +49 7022 5064998 (2 min)
All up-to-date contact data: https://darius-runge.eu_______________________________________________ Wikidata mailing list -- wikidata@lists.wikimedia.org To unsubscribe send an email to wikidata-leave@lists.wikimedia.org
On Sat, 29 Jan 2022 at 07:44, Darius Runge darius-runge@magenta.de wrote:
Dear all,
I started using WikiData for Private Information Retrieval. This allows answering certain questions while maintaining a high degree of secrecy. Suppose you wanted to know when Einstein was born, but for some reason you must keep the fact that you want to know this a secret. In this case, we assume a threat model with perfect knowledge about the computer in use, not just that someone managed to log the Wikidata requests.
One solution would be to request a table of every human who ever won the Nobel Prize (this requires the common knowledge of Einstein being a winner of said) with the kind of Nobel Prize, date awarded, date of birth, date of death unless living, nationality etc. If we let this table scroll across the screen and read the required entry, there would be - as far as I can tell - no way to learn which entry (and how many of them) is of our interest.
I have written a simple PHP script that allows one to enter a SPARQL request and have it displayed as a scrolling table. Please be advised that this is in a very informal alpha state, and I am no professional Web Developer. It's a mere proof-of-concept, but feel free to try it out if the API quota lets you.
https://darius-runge.eu/otp/request.php
My question is, whether anyone of you might be interested in working with me on discussing practical implications of this method (how should requests be written to allow for the desired privacy?) or even making a better implementation of a tool that allows viewing the scrolling table of printing it out.
Feel free to reply to this mailing list entry or contact me privately with the postal or telecommunication data provided in the footer in case you don't want to discuss it in public.
Best, Darius
This is interesting. Both privacy and wikidata service load concerns may be addressed best by the various efforts underway towards making it easier to run clones of the wikidata dataset in a variety of environments - cloud platforms, local machines, etc.
Eg https://addshore.com/2019/10/your-own-wikidata-query-service-with-no-limits/
Dan
Darius Runge Postfach 3 72669 Unterensingen Germany
Tel +49 7022 5064970 Fax +49 7022 5064971 Vox +49 7022 5064998 (2 min)
All up-to-date contact data: https://darius-runge.eu_______________________________________________ Wikidata mailing list -- wikidata@lists.wikimedia.org To unsubscribe send an email to wikidata-leave@lists.wikimedia.org
Thank you for the link, Dan! I'm sure there are many reasons to make it easier to run arbitrarily long queries on ones own machine. Anyway, I'm sure this will turn out to be helpful.
Darius
--- Original-Nachricht --- Von: Dan Brickley Betreff: [Wikidata] Re: Private Information Retrieval Datum: 30. Januar 2022, 2:32 An: Discussion list for the Wikidata project
On Sat, 29 Jan 2022 at 07:44, Darius Runge <darius-runge@magenta.de mailto:darius-runge@magenta.de > wrote: Dear all,
I started using WikiData for Private Information Retrieval. This allows answering certain questions while maintaining a high degree of secrecy. Suppose you wanted to know when Einstein was born, but for some reason you must keep the fact that you want to know this a secret. In this case, we assume a threat model with perfect knowledge about the computer in use, not just that someone managed to log the Wikidata requests.
One solution would be to request a table of every human who ever won the Nobel Prize (this requires the common knowledge of Einstein being a winner of said) with the kind of Nobel Prize, date awarded, date of birth, date of death unless living, nationality etc. If we let this table scroll across the screen and read the required entry, there would be - as far as I can tell - no way to learn which entry (and how many of them) is of our interest.
I have written a simple PHP script that allows one to enter a SPARQL request and have it displayed as a scrolling table. Please be advised that this is in a very informal alpha state, and I am no professional Web Developer. It's a mere proof-of-concept, but feel free to try it out if the API quota lets you.
https://darius-runge.eu/otp/request.php
My question is, whether anyone of you might be interested in working with me on discussing practical implications of this method (how should requests be written to allow for the desired privacy?) or even making a better implementation of a tool that allows viewing the scrolling table of printing it out.
Feel free to reply to this mailing list entry or contact me privately with the postal or telecommunication data provided in the footer in case you don't want to discuss it in public.
Best, Darius
This is interesting. Both privacy and wikidata service load concerns may be addressed best by the various efforts underway towards making it easier to run clones of the wikidata dataset in a variety of environments - cloud platforms, local machines, etc.
Eg https://addshore.com/2019/10/your-own-wikidata-query-service-with-no-limits/
Dan
---------------------------------------- Darius Runge Postfach 3 72669 Unterensingen Germany
Tel +49 7022 5064970 Fax +49 7022 5064971 Vox +49 7022 5064998 (2 min)
All up-to-date contact data: https://darius-runge.eu______________________________ _________________ Wikidata mailing list -- wikidata@lists.wikimedia.org mailto:wikidata@lists.wikimedia.org To unsubscribe send an email to wikidata-leave@lists.wikimedia.org mailto:wikidata-leave@lists.wikimedia.org
_______________________________________________ Wikidata mailing list -- wikidata@lists.wikimedia.org To unsubscribe send an email to wikidata-leave@lists.wikimedia.org