--- In yws-flickr%40yahoogroups.com">yws-flickr
yahoogroups.com, "J. Daniel Ashton" <jdashton...>
wrote:
>
> gvendurjaki wrote:
> >
>; > See, my problem is a little bit more complex. I am doing a survey for
> > Flickr users for a university research project. When people finish the
> > survey they are asked to allow us to access their account to read
>; > their Flickr information. For that they log onto Flickr (Yahoo!).
> >
>; > So, what I wanted to do was to use the authentication that has taken
> > place with the API and browse the user's authenticated pages. I'm
> > guessing that can't be done?
> >
>; >
>; >
>; Interesting project. I regret that this is all beyond my limited Flickr
> knowledge and experience.
>
> I would be rather surprised if an authentication token could be
> converted into web sign-in access. As far as I know, there is no way to
> give any person "read-only" access to the web pages that hold your
>; information on Flickr, so I wouldn't want there to be a way to convert a
> token (certainly not a read token, at least) into web-page access.
Yeah, makes sense, although I was hoping it would somehow be possible.
>
> /All/ is not lost, of course, because with a user's authentication you
> can get 95% of the same information through the API that they see in
> their web pages. An API to get the reverse contacts probably exists,
> undocumented. We can hope that a Flickr wizard will chime in here with
>; that information.
Well, I had already asked about an API method on this list and got the
answer that none was available. That's why I was gunning for some
other method to find it. Maybe there is some undocumented method?
Anyone? 
>
> I believe you can get some pretty good clues about a person's reverse
> contacts by checking on anyone who has commented on or faved that user's
> pictures. I believe there is a very good chance that a user's contacts
> will have left at least one comment or fave for that user, in most
>; cases. (Actually, my mom and my sister both violate that rule, but I'm
> still working on them. 
)
>
> To make a best-guess approximation of people who consider Photog-A to be
> a contact, start by checking who has faved or commented-on Photog-A's
> pictures.
>
> You can get a list of Photog-B's contacts using the url
> http://www.flickr.com/people/vlashton/contacts/?filter=family for
> family and
> http://www.flickr.com/people/vlashton/contacts/?filter=contacts for
> contacts
>
> (I haven't found the correct value for friends yet, strangely.)
>
> However, you might have to be a signed-in user (again, a Perl script can
> help you to automate this), and you might need to be one of Photog-B's
> contacts: I'm not sure about that part.
Interesting thought. Although, I already have a method to estimate
reverse contacts. What I need, is the correct number for some dataset
so I can use that to correlate my approximate reverse contacts.
>
> Doing it this way would be database intensive for Flickr, since you
> would have to make API calls for every one of Photog-A's pictures, plus
>; page requests for each person who might be a contact. It would save
>; Flickr a bunch of CPU cycles, database accesses, API calls and web page
>; requests if they would tell you about the API for getting reverse
> contacts for an authed user.
Yup, although I don't want to be abusing Flickr's database. Lets hope
there is something in the API I can use. Anyone know of any method?
>
> (By the way, Flickr has stated that programs that slam the API with too
> many calls, too fast, may eventually get their key revoked. The
> threshold of safety I've seen published is one call per second. I think
> my program is making about five calls in two seconds, but only in rare
>; bursts. So far, this level seems to have been acceptable.)
My estimation program (still to be written) will be database
intensive, but only for one or two calls (while I'm fetching all the
info). Is that acceptable?
>
> Another approach would be to read through the user's inbox, /if/ they
>; haven't deleted all the "You are so-and-so's newest contact" e-mails.
> However, as far as I can tell, there's no published API for reading the
> inbox either.
>
> (I've thought about these things because I wanted a way to check, every
> so often, to see whether I had made a contact of every person who
> contacted me. Checking the API, I came up dry in both the inbox and
> reverse contacts areas.)
>
> A potential problem in this area is privacy: it makes a lot of sense to
> let one person see their reverse contacts, but permitting a third-entity
> to access this data, especially on a large-scale, conglomerative basis,
> seems like something of a privacy risk.
Makes sense. I hadn't thought of that.
>
> If it is sufficient for you to just know /how many/ reverse contacts a
> user has, you could ask your survey participants to just report the
> number of reverse contacts, as given at the bottom of their reverse
> contact page.
Well, the number would be a good start, and perfectly acceptable if
that is the only thing I can get. Asking the users seems the best way
at the moment. Although we want to ask the user as little as possible
(less work for user, less chance of error).
>
> Sorry not to be more helpful.
Not helpful? You've been extremely helpful. Thanks alot.
>
> Kind regards,
>
> Daniel
>
> --
> Daniel Ashton PGP key available http://Daniel.AshtonFam.org
> mailto
aniel... http://ChamberMusicWeekend.org
> AIM: FirstFiddl ICQ# 9445142 http://MDMusic.org
>
Kind regards,
Gummi
.
