[08:31:57] DrTrigon: Saw the stats on newimages ? Want any more metrics ? [08:48:02] AbdealiJK: not recently - some days ago - leaf cat indicator (bool) was missing then... [08:48:35] True, I haven't added that - doing [08:48:52] but more important would be uncategorized run if possible [08:50:13] preferrably on a day you did newimages too - to compare [12:28:33] DrTrigon_: Hi [12:28:42] DrTrigon_: Is today Skype or IRC ? [12:29:14] I put skype due to the MVP and bad connection last week. [12:29:56] Alright [12:30:14] Hope connection is better - otherwise we have to fallback to IRC [12:30:39] (could not follow some part of last weeks discussion) [12:34:49] * DrTrigon back again ... router reboot to optimize connection ... [12:34:57] (y) [12:35:03] * DrTrigon believes in voodoo ;) [12:35:34] Let's talk until we have jayvdb here too. [12:35:55] Talk on irc or skype ? [12:36:14] Whatever you like :) [12:36:20] IRC it is [12:36:35] (just do not want to text in skype ;) [12:36:41] So, as discussed we wanted day wise stats [12:37:13] yes...? [12:37:14] I ran it on abot half the month, but as mentioned earlier - there were some issues like bugs in the script which ere not seen before [12:37:17] Was fixing those [12:37:37] good for comming beta-testing [12:37:42] nod [12:37:55] https://commons.wikimedia.org/wiki/User:AbdealiJKTravis/logs/newimages has top level stats on the files per day [12:38:11] https://commons.wikimedia.org/wiki/Special:PrefixIndex/User:AbdealiJKTravis/logs/newimages/ has day wise logs [12:38:25] * DrTrigon saw them - VERY nice! [12:38:32] Awesome :) [12:38:45] It's nice to see than 5% was acheived - and this was only with faces majorly [12:38:46] What your conclusion from this? [12:39:26] Overall, including downloading the file and analyzing, it takes 15 seconds on toolslab [12:39:34] 15 seconds per file * [12:39:54] And I think results are pretty neat. (But always scope for improvement) [12:40:17] I agree - actually I had this oppinion before already! [12:40:37] Regarding MVP and jayvdb, we have to come to a conclusion now... [12:40:45] I think these results should gain some momentum from others to contribute to the project to, as we've got a good working "prototype" [12:40:57] ...do you think you can give a realistic but still amitious MVP now? [12:41:37] So, let's see. [12:42:05] Based on previous discussions 5% was a benchmark, and this acheives that. [12:42:21] ^ In every bucket we had decided on during mid term meeting [12:42:36] I was hoping that we keep it as 5% and say the MVP is acheived [12:42:46] Cuz I'd rather work on UI and documentation for the next 3 weeks [12:42:54] * DrTrigon checks buckets... [12:42:56] PS: GSoC has only 3 more weeks before final review [12:43:23] * AbdealiJK still feels like making a nice web interface for this :P [12:44:11] DrTrigon: What is your opinion ? [12:44:17] * DrTrigon has other ideas too... ;) [12:44:40] I was initially under the impression (not sure why) that "re-evaluationg" the MVP was only required if 5% was not being acheived [12:45:22] The 3 weeks are the thing we I wanted to agree on the MVP a soon as possible... [12:45:35] What do you mean by "re-evaluating"? [12:46:21] By re-evaluation, I mean setting/changing the % number we want to be able to acheive [12:46:51] As I understood jayvdb the "5%" was just an example and he wanted a "sexy" number ... [12:46:59] ... whatever "sexy" means. [12:47:11] As I understood it should be as abitious as possible ... [12:47:16] Right, so this is a difference in interpretation [12:47:23] ... though I'm not sure whether it is an MVP then. [12:47:47] As I understood jayvdb, he though 5% was a sexy number which he was not sure we could acheive and wanted the stats to decide whether we need to reduce it [12:47:48] I was of the oppinion I and jayvdb have to define that. [12:48:02] that = MVP [12:48:21] But I guess now he wants you to do that - but not sure about that. [12:48:27] Nod. I think this is getting highly speculative on what we thought jayvdb meant :P Let's just hear it from him [12:48:51] Correct. [12:48:58] If he feels that we should set the % to 15% icould work on that. if he thinks 5% (We have ~8.5% right now iirc) is sexy enough, I can work on UI [12:48:59] So from the number I see now, we could even talk about 20% ... [12:49:21] ... 20% of content [12:49:36] ... but I would actually not want to do that. [12:49:52] Ah, 20% content could indeed be done if I get those maps identified correctly [12:50:01] SVG? [12:50:14] Would SVG help also? [12:50:50] I have not kept track of the SVG enough. It would probably help though - yes. [12:51:20] I have the impression it should be a low hanging fruit? Am I wrong? [12:51:41] SOrry, but could you repeat what detection in SVG ? [12:51:45] The text ? [12:52:09] No I mentioned the created with templates ... that also need validity info [12:52:17] Ah. yes that ! [12:52:26] But that wouldnt come in the "content" bucket. That would be type [12:53:20] I see - not used to these buckets yet ;) [12:53:29] :D [12:53:45] Basically "content" => it tells you what is the image showing. [12:54:08] Like "ball", "person", "cycle", "car", "plane", etc [12:54:40] Or even "Text", "Bar chart", etc. [12:54:42] (I was more thinking about leaf cats and templates are some... that was the route of thinking) [12:55:31] Yeah - leaf cats is another metric that can be used too [12:55:40] So can you write an MVP down - e.g. like the ones I proposed - just to have that finished and let jayvdb agree on it? [12:55:54] Well, anyway. I guess we need jayvdb's input for the MVP [12:55:56] may be using or including buckets [12:56:12] exatcly [12:56:25] nod [12:57:07] BTW: 6 days to go for Conference [12:57:13] I need you/us to also agree on how to future of this - very cool - projects looks like! [12:57:36] So, I'm going to be doing only documentation and testing this week - is that fine ? [12:57:50] I don't want to add any more new features this week because that increases chances of bugs. [12:58:11] I think a feture close for now until the copherence makes sense yes! [12:58:21] Phew - cool (Y) [12:58:34] * AbdealiJK will release 0.2.0 by monday then [12:58:53] And we need to agree on the future of the project, maintenace etc. anyways so why not do it now and have later time for coding fun... [12:59:22] Sure - what were you thinking about the future exactly ? [12:59:38] As mentioned in the agenda: https://phabricator.wikimedia.org/T141194 [12:59:59] I of course HOPE VERY MUCH you stay with us and continue your work! [13:00:18] ;)) [13:00:25] I do plan to :) [13:00:31] Ouh, John just pinged on skype. [13:00:43] perfect, so that makes that part very short ... [13:00:53] sorry [13:01:13] DrTrigon: But I'd probably take a break after August to get back on track with other stuff [13:07:28] https://phabricator.wikimedia.org/T141194 [13:12:47] lost you all [13:13:01] me too [13:13:13] lost you on 70% of categorizations... [13:15:48] I can hear you, go on [13:18:44] jayvdb: I was lookin into file names [13:19:22] ... thinking about using google search or learning by using natural languag kits [14:23:27] I guess I lost you guys - are we done? [14:24:25] AbdealiJK, jayvdb: ^^^ [14:24:27] Yeah, I ended the call because there was some really loud noise at the end [14:24:44] Should I restart the call ? Or shall we call it a day ? [14:24:48] That might have been my hone ringing - sorry! :) [14:26:05] Alright then - calling it a wrap [14:26:08] * AbdealiJK goes for dinner [14:26:29] Enjoy! [14:26:32] Thanks for the meeting guys! [14:26:42] Thanks for all the input :)