suggestion for generic filtering

[diplix]

just an idea i had on my way to work today. i thought about filtering on a module basis. most of all anoyances happens when one posts a picture to instagram and also CCs it to facebook or twitter. then you have 3 posts with that instagram. facebook.class already has a hard wired filter for apps like ifft, friendfeed, etc. but twitter has not (yet). but filtering all tweets that link to instagram.com wouldn't be a good idea, because that would also filter intentional tweets to other's instagrams.

so how about a filter that checks whether a tweet contains an url that is already the original_guid of a post. an instagram will get the original_guid like http://instagram.com/p/k39QpwJCnL/ . the tweet would contain http://t.co/qTlmlRozVq but since we already enlongen those short links, we could actually see that we have a duplicate.

however, due to syncing and wp_cron, it might occur, that the instagram is not saved, when the tweet is beeing saved, since they both were send at the same time. i can't say for sure which one of these was saved to the wp database first: http://wirres.net/reclaim/man-kann-das-mit-den-wimpern-auch-uebertreiben-zara-instagram-compk39qpwjcnl/ http://wirres.net/reclaim/man-kann-das-mit-den-wimpern-auch-uebertreiben-zara/

so my suggestion would be an hourly cron job that checks the last 50 entries for duplicates. if it finds a tweet or facebook status containing an instagram, flickr or whatever link guid, the tweet or facebook entry status is set to draft or private. a private status would show the duplicate to a logged in user, but not to visitors.

the duplicate finder needs some sort of priority list, so that it can decide which entry to get rid of, if it finds a duplicate.

still, a mute filter for twitter or facebook would be helpful.

• let the user decide whether he or she wants to get retweets, mentions or filter certain hastags or tweets generated by certain apps
• let the user decide which facebook apps get filtered

[klischka]

Great idea, I was thinking of exactly the same mechanism.

Am 26.02.2014 13:39 schrieb diplix:

just an idea i had on my way to work today. i thought about filtering on a module basis. most of all anoyances like when i post a picture to instagram and also cc it to facebook or twitter. then i have 3 posts with that instagram. facebook already has a hard wired filter for apps like ifft, friendfeed, etc. but twitter has not. but filtering all tweets that link to instagram.com wouldn't be a good idea, because that would also filter intentional tweets to other's instagrams.

[danito]

For Twitter, I can't see where we filter out false negative when we filter on source app and author (when respective mods are active).

E.g Instagram, if instagram == active, twitter username == @me and app == instagram -> filter it out, but not for favorites. So that when I fav a tweet with an instagram it doesn't get filtered. And if I retweet a tweet with an instagram -> source app != instagram anymore.

For 4sq : same , if app == "foursquare" filter it when twitter username == @me.

The only thing is if Twitter decides to drop the source app name in their response, it wouldn't work anymore.

Dan

2014-02-26 13:39 GMT+01:00 diplix notifications@github.com:

just an idea i had on my way to work today. i thought about filtering on a module basis. most of all anoyances like when i post a picture to instagram and also cc it to facebook or twitter. then i have 3 posts with that instagram. facebook already has a hard wired filter for apps like ifft, friendfeed, etc. but twitter has not. but filtering all tweets that link to instagram.com wouldn't be a good idea, because that would also filter intentional tweets to other's instagrams.

so how about a filter that checks whether a tweet contains ab url that is already the original_guid of a post. an instagram will get the original_guid like http://instagram.com/p/k39QpwJCnL/ . the tweet would contain http://t.co/qTlmlRozVq but since we already enlongen those links we could actually see that we have a duplicate.

however, due to syncing and wp_cron it might occur, that the instagram is not saved, when the tweet is beeing saved, since they both were send at the same time:

http://wirres.net/reclaim/man-kann-das-mit-den-wimpern-auch-uebertreiben-zara-instagram-compk39qpwjcnl/

http://wirres.net/reclaim/man-kann-das-mit-den-wimpern-auch-uebertreiben-zara/

so my suggestion would be an hourly cron job that checks the last 50 entries for duplicates. if it finds a tweet or facebook status containing an instagram, flickr or whatever link guid, the tweet or facebook entry status is set to draft or private. private would show the duplicate to a logged in user, but not to visitors.

the duplicate finder needs some sort of priority list, so that it can decide which entry to get rid of, if it finds a duplicate.

Reply to this email directly or view it on GitHubhttps://github.com/espresto/reclaim-social-media/issues/127 .

Daniel Nix

[diplix]

you're right but my suggestion seems more effective. it would work for all services right out of the box and does not have to differentiate whether the user is me or the app that posted the entry has an active class counterpart or not. with your suggestion we would have to write filtering code for each class (twitter, facebook, g+, app.net, …).

[danito]

In this case, your reclaim TL will variate in time, I'm not so in favour and if someone follows your rss feed will get the duplicates in any case.

And what about when I RT a tweet with a link to an instagram with an extra comment and later I like the Instagram on Instagram itself, my tweet has maybe more value than the like but would then get hidden from my stream...

What with bookmarks, I FB a link and later I bookmark it. Is this a case to filter it or not ?

Not so easy...

And please ad photo recognition so that when I publish the same photo on Instragram, Twitter, gplus and FB on the same time from my mobilephone it gets filtered and put into a media post... :D

2014-02-26 14:05 GMT+01:00 diplix notifications@github.com:

you're right but my suggestion seems more effective. it would work for all services right out of the box and does not have to differentiate whether the user is me or the app that posted the entry has an active class counterpart or not. with your suggestion we would have to write filtering code for each class (twitter, facebook, g+, app.net, ...).

Reply to this email directly or view it on GitHubhttps://github.com/espresto/reclaim-social-media/issues/127#issuecomment-36122491 .

Daniel Nix

[diplix]

jep. not so easy,