User talk:West.andrew.g/Archive 3
This is an archive of past discussions with User:West.andrew.g. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page. |
Archive 1 | Archive 2 | Archive 3 | Archive 4 | Archive 5 | → | Archive 10 |
Sorry
Sorry about that... I had been on Stiki for a while and got sort of tired of honestly looking for vandalism so my judgement got kind of hazy... won't happen again.... :) Cheers... Mìthrandir (talk) 21:43, 4 February 2011 (UTC)
Thanks
Hi Andrew, thanks for adding your paper to the anti-vandalism bot census, I have read it and it is very interesting (but I have to learn a bit about classifiers). I'm working in some Wikipedia charts, specially this one about reverts (updated every hour). You can see the performance of several anti-vandalism tools. Surprisingly, about a half of reverts are classify in other (probably they are handy reverts). If you have any suggestion, please, notice me. Regards. emijrp (talk) 14:06, 11 February 2011 (UTC)
- I guess my question is, "what are you trying to achieve?" You've done a quite thorough job over at your anti-vandalism census. Are you eventually hoping to eventually build a superior bot? I have lots of ideas about all types of things -- but I'd like to know what direction you are interested in. Thanks, West.andrew.g (talk) 17:15, 11 February 2011 (UTC)
- I'm researching how other bots revert vandalism. My AVBOT uses regular expresions and some metadata analysis to detect bad faith edits, but it is not enough. I want to learn about classifiers and machine learning and use Wikipedia like a big "automatic" corpus (for example analysing trusted users reverts to anonymous vandals). emijrp (talk) 18:09, 11 February 2011 (UTC)
An "otherwise needs attention" button
Hi,
I've found Stiki quite good fun. However, in my view it really needs an "otherwise needs attention" button, for edits that are not vandalism but are otherwise problematic. This button would leave a note of the edits somewhere for further review. Even better, there could be two "needs attention buttons" - one of which would revert the edits, one of which wouldn't. At present I'm ignoring most stuff as "innocent", while the more egregious things I'm sorting out manually... I don't really like calling either "innocent". Or maybe you could just rename the button to "not vandalism"?
Gosh, just realised that reads as a list of demands! It's not supposed to be... it's just my thoughts. Thanks so much for your good work!
Egg Centric (talk) 19:10, 13 February 2011 (UTC)
- Hi Egg Centric, and thanks for your use of my tool. There has been a lot of discussion along these lines in the past, with a lot of mixed feelings.
- So you are proposing two new buttons, a "good faith -- revert" and a "good faith -- tag"? The revert case is pretty straightforward (and something I'm not opposed to), but how do you think the "tag" case should be handled? Tagging the article directly might not be the greatest idea, and encourage tagging for minor issues (which is discouraged). Writing "tagged edits" to another location might generate little exposure -- and be the moral equivalent of ignoring them. I'd like to hear your thoughts on the matter. Thanks, West.andrew.g (talk) 01:49, 14 February 2011 (UTC)
- STiki can already do all this. For the case of a non-vandalism revert, change the edit summary on the bottom left as follows: "Reverted edit(s) by [[Special:Contributions/#u#|#u#]]
identified as test/vandalismusing [[WP:STiki|STiki]]", uncheck "Warn Offending Editor?", and click "Vandalism". This will preform the revert without prejudice, allow you to leave a manual warning for the editor using Twinkle (if needed), and most importantly, train the STiki classifier that this was a problematic edit. For the case of attention needed, either fix it yourself, tag it, leave the editor a note, or click "Pass" and hope that the next STiki user is better equipped to deal with the problem. —UncleDouggie (talk) 05:14, 14 February 2011 (UTC)
- STiki can already do all this. For the case of a non-vandalism revert, change the edit summary on the bottom left as follows: "Reverted edit(s) by [[Special:Contributions/#u#|#u#]]
- Perhaps it could tag the talk page? It would be reasonably easy to develop a few boilerplate messages (as well as a custom one) - indeed we already have ones like {{cleanup}}, {{confusing}} and so on... I'm pretty sure at this time writing to a log somewhere will indeed end up with a very large list, but then the wiki is never finished ;) Egg Centric (talk) 18:14, 14 February 2011 (UTC)
IP vandal 216.56.48.115
Cluebot sometimes screws up, but looking at that IP's contribution history, the AIV report was spot-on. --Alan the Roving Ambassador (talk) 18:31, 23 February 2011 (UTC)
Suggestion
Thanks for being on top of the further vandalism by User 146.243.4.157. FYI -- the IP had just received a final warning, a bit before your warning (at a much lower level). I would encourage you, when you see vandalism and the user has a very recent final warning, to seek to have them blocked -- rather then give them a reduced warning. If you have thoughts, you can respond here -- I will watch your page for a bit. Many thanks for your anti-vandal efforts.--Epeefleche (talk) 19:11, 3 March 2011 (UTC)
- This was a tool edit, by WP:STiki, of which I am the author. The tool also places warnings. The reason it reported a level-1-vandalism warning was because that there were no other recent *vandalism* warnings on the page. Had there been a recent level-4-vandalism (or 4im), it would have reported to AIV. However, "blanking" and "unsourced content" warnings could be somewhat orthogonal issues to vandalism (esp. when you consider the entire suite of escalating warnings) -- so the tool assumes good faith. This has been discussed before at User_talk:West.andrew.g/TalkArchive02#STiki. Thanks, West.andrew.g (talk) 19:20, 3 March 2011 (UTC)
STiki bug
I recently made this edit to AIV reporting a vandal. In the edit summary it appears that it tried to link [[Special:Contributions/:xxx.xx...]] instead of excluding the colon. Just thought you'd like to know about this. —GƒoleyFour— 04:16, 8 March 2011 (UTC)
- What is the release date on the version you are using? See Wikipedia_talk:STiki/TalkArchive02#Bug_report:_broken_link_in_AIV_summary -- a fix was rolled out on Jan. 31. If your version is more recent than that, let me know. Regardless, thank you for your use of my tool. Thanks, West.andrew.g (talk) 04:23, 8 March 2011 (UTC)
- Oh yeah, I'm probably on a older version. Time to update... —GƒoleyFour— 04:30, 8 March 2011 (UTC)
- What is the release date on the version you are using? See Wikipedia_talk:STiki/TalkArchive02#Bug_report:_broken_link_in_AIV_summary -- a fix was rolled out on Jan. 31. If your version is more recent than that, let me know. Regardless, thank you for your use of my tool. Thanks, West.andrew.g (talk) 04:23, 8 March 2011 (UTC)
Your message
Was your message automated? Because if it wasn't, it was polite and I commend you for it. I Know that may seem silly but when people troll the Mods sometimes have to be real squirmy about it. Thank you for being human :3 — Preceding unsigned comment added by LegoSaur (talk • contribs) 04:58, 8 March 2011 (UTC)
- To be honest, I used a tool to locate your unconstructive edit and the message I placed was according to a template format. Such warnings intend to start out "polite" so that newcomers are not turned away. However, subsequent vandalisms will earn one increasingly stern messages. I noticed you damaged a second article (which I have since fixed). However, since you seem to have taken my first warning to heart -- I won't ding you again. Hopefully you will become a constructive contributor, or at minimum, not introduce more damaging revisions. Thanks, West.andrew.g (talk) 05:08, 8 March 2011 (UTC)
I plan to do so, good sir. I am hoping I'll get into a Physics college course and contribute to science pages here :3 — Preceding unsigned comment added by LegoSaur (talk • contribs) 05:15, 8 March 2011 (UTC)
WikiMania
No idea if you're planning on it, but since STiki has come a long way in the past year, just wanted to pass on this call for WikiMania: http://wikimania2011.wikimedia.org/wiki/Call_for_Participation . Cheers, Ocaasi (talk) 09:27, 10 March 2011 (UTC)
- Yes, WikiMania is on my radar. I've applied for a scholarship to attend (and will present if received). WikiMania doesn't publish proceedings and is a bit "non-academic" -- so its a tough sell for those who usually reimburse my academic travels. An early version of STiki was presented at WikiMania last year in Gdansk (co-located with the more academic WikiSym): [1]. I've also got some novel work in progress which I am targeting for WikiSym 2011. Thanks West.andrew.g (talk) 17:15, 10 March 2011 (UTC)
- Ask Jimbo to write a letter to your funding committee. Maybe a nod from the man himself will turn it over. Heck, maybe the Foundation has funds. I still think widespread use of CBNG combined with STiki and perhaps integration of Pending Changes could be a remarkable combination. If it refines CBNG and your metadata, it will only get more powerful. Seems like a no-brainer why they would want to fund it. Vandalism is every Wiki's problem, and if there's a scalable solution here, it's big freakin news. Anyway, I'm a fan. Ocaasi (talk) 19:05, 10 March 2011 (UTC)
ClueBot
You can remove this notice at any time by removing the {{Talkback}} or {{Tb}} template.
STiki Public Service Announcement
Hello STiki users, and thanks for your continued use of my tool.
As you may have noticed, the "ClueBot NG" queue has not been updated in some time. Initially, this was the result of the ClueBot NG being completely offline. Bot service has since been restored. However, the IRC feeds on which STiki depends to enqueue scores have not yet been restarted. I have raised this as an issue with the developers (see User_talk:ClueBot_Commons#IRC_Feeds) -- but I am yet to receive a response. Perhaps some others can pile on over at that talk page to see if we can't make some traction.
In the meanwhile, I encourage STiki users to switch to the "metadata" queue, which is actively processing/en-queuing fresh edits. You are much more likely to find vandalism in this stack. I also realize, that due to to the CBNG-queue being down, that STiki may take a rather lengthy period of time to load/initialize. I apologize. Fixes in the near future will allow me to adjust the "default" queue on the server to circumvent such issues. Look for these changes in a couple weeks time -- I am quite bogged down at current trying to prepare some academic work for WikiSym. Thanks, West.andrew.g (talk) 22:17, 22 March 2011 (UTC)
- This feed came back up today! (Let's just hope it sticks around). Assuming that holds true, users can resume using the CBNG queue. Thanks, West.andrew.g (talk) 21:29, 26 March 2011 (UTC)
- Err, not so fast. ClueBot is down again. I just suggest users play things by ear. If the ClueBot queue is showing poor performance, the feeds are probably down, and you should transition over to the STiki metadata queue in the meantime. Thanks, West.andrew.g (talk) 20:44, 28 March 2011 (UTC)
Stiki Races
Hi Andrew, Been a wee while since I've used Stiki (sorry, not sure of CaSiNg) but I'm going to take it out for a wee whirl now, after all it is Saturday night (!) but something just came to my mind... how about a Stiki Race? First to revert 1000 vandalism cases wins, or something of that nature. And they all have to be genuine vandalism, to prevent damage and so on...Cheers, Egg Centric 21:41, 26 March 2011 (UTC)
- Hi there. First, the "correct" casing is "STiki" which originally stood for "(S)patio-(T)emporal analysis over W(iki)pedia." The software has expanded a good bit from those origins, though, so the acronymic nature is now downplayed. Second, I'm all for encouraging use of the tool -- but wouldn't want one to encourage anyone to be too aggressive in the hopes of winning some "race". Perhaps when real-life settles down in the near future I should award my most prolific users with barnstars as a token of appreciation. Thanks, West.andrew.g (talk) 20:49, 28 March 2011 (UTC)
Oh.... And what happened here?
I seem to have somehow given a stiki warning without a revert... see User_talk:Carmichael95#Toast.
I do recall something about toast, but I can't see that that user did anything deserving a revert... and even had they done so (or I thought that they had done so) then surely there ought to have been a revert done at the same time, not just a warning? Any ideas? This seems like more than user error to me. Egg Centric 00:43, 27 March 2011 (UTC)
- I see what you are seeing, but this is very difficult to explain. In almost 50,000 other reverts this behavior has never been reported -- and there is dedicated logic in my code to keep it from happening. I wonder if something very funky didn't hiccup on Wikipedia's side (race conditions come to mind -- or a weird edit conflict condition). I am not going to let this trouble me unless is happens again -- but if it does, report it and I'll start digging. Thanks, West.andrew.g (talk) 20:53, 28 March 2011 (UTC)
On-wiki Research
Thought you might be interested in this thread. Berkman at Harvard is trying to do a study and wants to post survey invites to 40,000 user talk pages. Needless to say, it's facing a bit of resistance. I think you might be able to offer some insight from an academic and technical perspective, particularly with the experience you've gained from working with the community over the last year. Cheers, Ocaasi c 10:09, 2 April 2011 (UTC)
STiki
Hello i just wanted to say I STiki is a great tool.TucsonDavid (talk) 06:56, 29 November 2010 (UTC)
- Thanks, TucsonDavid West.andrew.g (talk) 22:20, 7 December 2010 (UTC)
History TA proven wrong by STiki assisted vandal-fighter
Didn't know if you'd seen this: http://en.wikipedia.org/wiki/User_talk:Arthena#Freedmans_Bureau_edit
- Some history TA tried to prove Wikipedia was a minefield of misinformation by adding some. Editor Arthena caught it in under 30 minutes using STiki. Reddit picked up the story. Ocaasi (talk) 21:25, 16 December 2010 (UTC)
- Awesome story! Thanks for the pointer, West.andrew.g (talk) 21:41, 16 December 2010 (UTC)
Statistics
Not sure you are the right guy to ask, but I'll risk the shoulder-shrug and "WTF" response.
Is it currently possible to extract from existing Wikipedia systems (only English Wikipedia) statistics on, for example, the total number of pages, contributors, edits, reverts, admin requests, admin interventions by type, etc? If that is not possible, based on your experience, would it be an achievable project to make such figures accessible and auto-published, say once a month?
The context: there are periodic discussions about perceived trends and issues of concern at Wikipedia that suffer from being largely opinion-based. Without a 'dashboard' of thoughtful metrics, that discussion will remain opinion-based, and periodically divisive. But if a set of metrics can be agreed on and measured to offer a snapshot of what actually occurs here, such discussions would at least be grounded in cold, hard facts.
If you are the wrong guy to talk to about this, maybe you know where I could ask that question more usefully. Regards — Peter S Strempel | Talk 22:19, 7 April 2011 (UTC)
- See Wikipedia:Statistics. It seems that many of the counts you describe exist in some form on that page (I'd like to highlight [2] and [3] in particular). It might not be the convenient "dashboard" display you describe -- but the raw data is out there. Most of these tools build on the toolserver, which basically allows queries to be run against the live Wiki-* database. Is this what you were looking for, is your proposal in some way distinct from this? Thanks, West.andrew.g (talk) 23:38, 7 April 2011 (UTC)
- I have to have a look and play. I've never had reason to concern myself with this side of Wikipedia before. But thank you so much for your prompt and helpful reply. Regards — Peter S Strempel | Talk 02:39, 8 April 2011 (UTC)
- You might also be interested in the (somewhat limited) information that Special:Statistics yields, as well. Killiondude (talk) 07:32, 9 April 2011 (UTC)
- I have to have a look and play. I've never had reason to concern myself with this side of Wikipedia before. But thank you so much for your prompt and helpful reply. Regards — Peter S Strempel | Talk 02:39, 8 April 2011 (UTC)
Smart watchlist
Since you're not busy or anything... Jimbo asked about how he could a) see if he missed anything on his watchlist and b) know which watchlisted edits were more likely to be vandalism. The second part seems like an ideal project for CBNG or STiki, as part of a .js tool that I thought you might have an idea about or interest in. It could add the CBNG neural network score, or STiki metadata score (or both) next to each watchlist edit. Perhaps it would color code edits by reliability, or flag them after a certain threshold. Theoretically, users could even set their own threshold, either globally or for individual articles. I have no idea what it would take to implement this, but as long as there was full access to the CBNG/STiki servers (which might require moving them somewhere more robust) then every edit could get scored in this way. I mentioned you at the discussion, since this is your territory. Cheers! Ocaasi c 23:47, 7 April 2011 (UTC)
- Yes, I am a bit deluged right now. Two papers went out to WikiSym on Friday (more details coming soon!). I also got a partial scholarship to, and will be attending, WikiMania in Israel in the Fall. I plan to write up presentation proposals for that in the coming days (I anticipate one will focus on STiki and anti-vandalism progress over the past year: CBNG, the CLEF competition, etc.) -- I'll put a pointer to them here when they are done. Some email anti-spam work is keeping me busy right now for CEAS, before I turn to [4] in the next month.
- The latter should benefit STiki, as it will allow me to improve STiki's *algorithm*, and not just the GUI tool (which has received much of the recent attention). This seems particularly important. CBNG has been offline for 1 week+ now, and if Cobi continues to be inactive, I could even turn the STiki logic into a bot that handles the most egregious instances of vandalism.
- Yes, the suggestion that you/Jimbo put forth seems to be a quite reasonable one. If it can gain community consensus, it should be a quite simple interfacing matter with the devs. Score one for my support, as an opt-in extension, at minimum. STIki already writes its scores to an IRC feed, and it would be simple to transfer the Java code onto a Wikimedia server to give everyone some sense of reliability. Thanks, West.andrew.g (talk) 02:37, 8 April 2011 (UTC)
- Australia, Amsterdam, Israel. Surely the CS grad programs keep this little perk a secret. PAN should be exciting, is it a full competition again?
- I don't think its a secret. I could have a "real" job and be out-earning my graduate stipend considerably, but a little bit of travel now and then makes the long hours a bit more worth it (and the papers still have to get written, and accepted!). PAN is a competition again (I didn't enter last year), but has some fresh rules to keep things interesting. For example, there are some foreign language edits which must be scored, and you are allowed to use features derived from "future" data. Not all this is completely relevant to the detection of "fresh" vandalism on en.wikipedia -- but Martin Potthast also outsourced some fresh en.wikipedia corpus tags -- which can help refine everyone's techniques.
- Concerning about Cobi and CBNG. Have you tried Crispy? That bot does way too much work to be hosted and managed completely externally, even though those guys seem like two of the more competent coders I've run across on the entire project. Not that this is a matter at all, but I think they're on the younger side, and might be running into some real life changes that might make it harder to keep up. Pure speculation, (and a further nod to their remarkable aptitude anyway), but it might be good reason to try and integrate some of these projects somewhere more central, toolserver, the WMF, WikiProject Vandalism, etc. It's both a wonder and a terror that the primary anti-vandalism method on the world's largest encyclopedia and 5th most popular website is run by two college kids on their home computer! Crazy! (but also very cool). It's not like there aren't layers of redundancy between the other anti-vandal programs (Huggle and STiki) in particular, as well as regular recent changes patrollers, but if we're going to move away from that manual model and crack the vandalism nut for good, it has to be a consistent machine-learning tool.
- I have been poking my head in over at the CBNG talk page and CBNG IRC just with the hopes of getting their IRC feed restarted (now irrelevant seeing as how the whole bot is down). Not sure what's really going on -- but we know that there exist code which works. If the only problem here is a reliable hosting provider, that is trivially overcome.
- As for STiki replacing CBNG as the logic bot, I still don't understand the relative accuracy/efficiency between them. Crispy seemed to thing neural networks under proper tutelage and tinkerage were the way to go, but just hearing him describe CGNB's workings made it seem like the delicate construction of a genius, not something particularly simple to replicate or replace.
- I could crank up STiki's accuracy considerably given some time (remember, its original concept was to use only *metadata*). I just really hadn't worried about this since the CBNG folks had a fine general-purpose system and we were interfacing well. I would change STiki (the algorithm) considerably if it needed to do something like this. Absent that, CBNG is *trivial* to replicate if Cobi/Crispy disappear forever, I have their source code!!
- If you don't mind, I'll repost the relevant pieces on Jimbo's talk page, and maybe throw up a Village Pump Technical proposal or a note at WikiProject Userscripts.
- Please do
- Off-hand, if you ever want a reader or copy-editor or non-CS assist, I'd love to look at something you're working on. Also, i'm about 20 min outside of Penn, which would have been old stomping grounds for me, if I was a) old and b) went much into the city. Ocaasi c 02:56, 8 April 2011 (UTC)
- I'll consider it. Out of curiosity, what is your "real world" area of expertise? Thanks, West.andrew.g (talk) 03:55, 8 April 2011 (UTC)
- Frankly, my real world expertise is absorbing information and presenting arguments across different fields. I'm a bit of a generalist, or as they politely say in college, pre-law. But I didn't go to law school (at least not yet) and spent some years teaching (and learning) outside of my poli-sci-econ studies. But Wikipedia has brought be back to my roots as an editor and dilettante and idea networker. One of my great strengths (and perhaps weaknesses) is quickly acquiring a conversational level of intelligence with complex topics, without (necessarily) mastering them. But I'm getting slightly better at the mastery side over time. Anyway, I read and write and research and discourse, those are my areas of expertise. Ocaasi c 04:17, 8 April 2011 (UTC)
- Update: Uncle Douggie writes: "Sorting by the criteria requested by Jimbo, or by vandalism risk, are new ideas that I agree would be very useful. I've designed a rough framework to accomplish all this in JavaScript that would act as a fairly transparent replacement for the standard watchlist functionality. To get the vandalism risk, we would need to have West.andrew.g provide an API to query the STiki database, which stores the scores for all three rating engines, including User:ClueBot NG. While this sounds like a fun thing to do, I'm consumed at the moment rewriting Twinkle before HTML5 gets turned back on and kills it. Perhaps Andrew would be willing to work on the API in the meantime. If Andrew doesn't have time, we could setup our own API server somewhere that would query the STiki database on behalf of clients. If any other developers want to start tackling this, let me know and I'll send you my rather rough notes. I'm a bit concerned by how dependent we are becoming on using non-WMF servers and software to manage the project. When ClueBot NG went down for a weeks earlier this year, it was a major blow to vandalism prevention. If Cobi or Andrew aren't around when there's a problem, we can get into a real bind. We are also so dependent on user scripts like Twinkle that whenever something breaks them, productivity comes to a crashing halt. The MediaWiki developers don't test compatibility with user scripts before deploying changes, as we know all too well. However, we need to keep moving forward and a smart watchlist seems like the next logical step to improving article quality. I'm sure we will find another dozen features to add to it once something is in place." pasted by Ocaasi c 06:05, 8 April 2011 (UTC)
I have created a first cut at a smart watchlist. Please see User talk:UncleDouggie/smart watchlist.js. —UncleDouggie (talk) 08:46, 11 April 2011 (UTC)
- On (next) Monday I'll be in a position to comment on this and start thinking about API improvements. As a thread with Ocassi over at STiki:Talk describes, I am overwhelmed with work deadlines right now. Please don't interpret my silence between now and then as any sign of ill-will. Thanks, West.andrew.g (talk) 18:49, 11 April 2011 (UTC)
STiki API improvements
Regarding the implementation of a smart watchlist, the current STiki API is a good start, but it could use some more features. It will be more efficient to query multiple revids at the same time, much like the MediaWiki API supports with revids=4781398|5487332
. The result should be a JSON object such as: { revid_4781398: 0.0558824660557638, revid_5487332: 0.0518232344557638 }
. This format will make it fast and easy to lookup scores using JavaScript's native ability to hash the property names for an object. You might need to check the max JSON size supported on your server and either increase it or limit the number of revids in a single request.
Ideally, I would like a single number that is a combination of the STiki and CBNG scores. I haven't found WikiTrust to be terribly useful myself, but perhaps a really high score from them would be good to factor in. —UncleDouggie (talk) 07:14, 8 April 2011 (UTC)
- We would have a problem using the API hosted on your site for the smart watchlist due to browser security restrictions on accessing resources from other domains within a script. Our options are:
- Somehow host the API server on en.wikipedia.org/w/stiki_api.php
- Use Cross-Origin Resource Sharing, which requires your web server to return HTTP headers specifically permitting the cross-site sharing. This only works with certain newer browsers. Even supporting IE8 would be difficult because I don't think I could use jQuery to invoke the partial implementation. Nevertheless, most power watchlist users use Firefox or Chrome anyway.
- Use JSONP. This would work in all browsers and only requires a small PHP code change on your end. This method only supports HTTP GET requests, which limits the query to about 2 kB. It would be nice to use a longer POST request for a big watchlist, but we could just split up the requests into chunks for now. There is also a security risk that if your server is hacked, it could be used to run evil JavaScript directly on Wikipedia as any user who is using the smart watchlist feature. A fix has been proposed for this, but it's not implemented by any browsers currently, so forget that. Nevertheless, we could use this technique in the short-term to get the smart watchlist gadget developed. In parallel, we should figure out a long-term solution that is to everyone's satisfaction.
- —UncleDouggie (talk) 13:11, 8 April 2011 (UTC)
My research @ Wikimania 2011
Hello, everyone. As a bit of a PSA, I'd like to announce that recent Wikipedia research (and STiki in particular) figures heavily into my two submissions to Wikimania 2011:
- "Anti-Vandalism Research: The Year in Review" -- Looking at both practical and academic anti-vandalism progress over the last year. On the practical side, this includes the evolution of the STiki tool, the inclusion of third-party queues (such as ClueBot NG), and recent proposals to integrate anti-vandalism algorithms into "pending changes" and/or "smart watchlists."
- "Autonomous Detection of Collaborative Link Spam" -- This is a queue which is in development for STiki, targeting external link spam. The theoretical work is largely complete (and in submission to WikiSym 2011), and I'm currently working (with my co-authors) to leverage the technique in a live/online fashion.
I'd encourage my supporters to visit the Wikimania site (it can be done under unified login) and indicate your interest in these presentations (only if you feel so inclined). Though I realize few (if any) of you will be in Haifa in August -- you can still benefit. There are plans to tape/stream the presentations and make them available to the entire Wikipedia community. Thanks, West.andrew.g (talk) 04:03, 23 April 2011 (UTC)
STiki False Positive
Hello, I recently received a message from you regarding an edit of mine that came up in your STiki results as potential vandalism. I wanted to report this false positive as a portrayal in a feature film is certainly not vandalism. I will be undoing your edit but applaud your in your constructive effort to aid in Wikipedia's goal of "Good Faith Collaboration." Sincerely, Geoff76.25.18.197 (talk) 10:34, 11 May 2011 (UTC)
- Looking at your edit again, it does appear to be a good-faith one, and I apologize. Something about the connotation of the word "lame" and the fact the left-half of that HTML comment wasn't immediately apparent motivated me to pull the trigger. STiki does not have "false positives", but its users can make mistakes, which in this case was me. Thanks, West.andrew.g (talk) 14:20, 11 May 2011 (UTC)
A discussion about improving the help documentation inspired an idea--Wikipedia tutorials would be best if they were interactive and immersive. The thought of a learning-teaching game came up, one based on a real interface with realistic 'missions'. Would you be interested in providing some feedback or helping work on it, or know some editors who might? The idea is just getting started and any assistance with the help/policy side, the experienced-editor side, or the coding/game-making side would be great. Cheers, Ocaasi c 03:49, 28 May 2011 (UTC)
- Please put your responses at User_talk:Ocaasi/The_Wikipedia_Game to consolidate discussion. Dcoetzee 11:10, 29 May 2011 (UTC)
Idea
I had an idea, which in general terms may sound startlingly like Wikipedia but hear it out. Create a website with a catchy name, allow people to register, create a compelling narrative that they are 'guardians of the encyclopedia' or 'protectors of knowledge', present them with diffs, ask them to rate whether they are harmful (definitely, maybe, not sure). Then take 4/5 ratings and a) feed those to the top of the STiki queue; or b) revert them under some approved 'game' account. You could possibly have users go through a training mode before their entries are counted, and you could weight entries based on past performance (agreement with the group, or gold standard placements). It would be somewhat similar to STiki, except gamified, and web-hosted. The other half of any great code is getting people to use it. This could be a neat project, and it has an academic component in the vein of http://zooniverse.org and other crowdsourced, crowd-tasked projects (a la GWAP. Note, this is not related to the game I've been working at, but it is in the same arena. What do you think? Ocaasi c 06:21, 1 June 2011 (UTC)
- Is the Zooniverse link correct? Seems like some standard SOE content at that destination. I think the game is a fine idea, but there are some implementation concerns. Some administrators did not not like the idea of new/unregistered editors having access to STiki. I can only imagine what would happen if random "users" edits were streamlined into a single master "STiki" account -- seems to be a loss of accountability there. STiki could trivially be implemented as a web applet. But isn't your suggestion just basically lowering the threshold for becoming a STiki user by eliminating the "download" and "create a username on Wikipedia" steps? Why would anyone play this game (who wouldn't already register/download)? Wouldn't there need to be some kind of incentive? Of course, we could have users "climb the ladder" by ranking them based on revert-count -- but this seems to encourage over-zealous reverting and edit-count-itis (and would require an account to track persistently?). Plus, there would need to be advertising to get anyone to visit this game in the first place. Not meaning to beat down your idea -- but this seems to suffer from several challenges that don't seem to affect your other proposal. Thanks, West.andrew.g (talk) 23:22, 1 June 2011 (UTC)
- Sorry, .org not .com. The accountability issue is interesting and would involve a less xenophobic mindset. As a tradeoff for using registered editors, the appeal of the game would be its moderation approach. Instead of 1 qualified user making changes, It would have a tutorial to both teach users how to identify vandalism and then allow them to practice, and eventually move on to real diffs. (Also, any or all of this could be on Wikipedia). User ranking would be rated on their level of agreement with other editors and with gold standard diffs. Users who significantly deviated from either would go lower in the rankings. That algorithm is a harder problem beyond my background, but I'm sure game devs use them all the time. The 'incentive' to this general idea, would be the rich graphic element, the rush of getting to play judge about edits (informed judge), web-based convenience, the game/storyline element, and the hivemind element (3/4 or 4/5 agreement required). I agree this idea is less further along then the interactive tutorial. It's just another game-related application. Advertising for the game would hopefully happen from press. I don't know how much zooniverse advertises, but the game is wildly popular. Of course, people get to explore galaxies, but getting to play God/Editor is appealing to many (as we know). An alternate focus may be just on getting broader adoption for STiki. That's not as exciting in the new-project department, but it may be much more fruitful. Maybe that's the more pragmatic focus to have now. Ocaasi t | c 18:16, 4 June 2011 (UTC)
Talkback
Message added 13:13, 4 June 2011 (UTC). You can remove this notice at any time by removing the {{Talkback}} or {{Tb}} template.
- DamianZaremba (talk • contribs) 13:14, 4 June 2011 (UTC)
STicki Question
Hello, does STicki work for other MediaWiki sites such as Wikia or is it Wikipedia only?--184.56.245.182 (talk) —Preceding undated comment added 22:18, 7 June 2011 (UTC).
- At current it only actively processes English Wikipedia edits. However, it would be straightforward to port a version to any MediaWiki site, given that they share an API. Thanks, West.andrew.g (talk) 23:16, 7 June 2011 (UTC)
Dead links
I just started Budi County. Checking "What links here" I saw User:West.andrew.g/Dead links, with the link http://ops.unocha.org/ProjectSheet.aspx?appealid=908&projectid=38691&doctype=pdf. The link is not really dead, but launches download of the PDF document. I don't know if there is a better way to handle it. Aymatth2 (talk) 19:03, 14 July 2011 (UTC)
- See the intro text at WP:STiki/Dead_links, as it describes there, the reports are not perfectly accurate. Generally the reports are correct (at the time indicated on the report), but the server may have fixed the issue, or there could some anomaly. If it bothers you that it is showing up in "What links here", feel free to delete the report from the page. Thanks, West.andrew.g (talk) 19:30, 14 July 2011 (UTC)
- It doesn't bother me at all. I just checked the link out of curiosity. It is a slightly odd url because it does not open a page but starts a download, at least in Chrome. So I wondered a) is the script getting thrown off by the oddball link, and then b) should I have coded it some other way? Aymatth2 (talk) 19:44, 14 July 2011 (UTC)
- The link works fine on my personal machine. The system just uses the URLs the way the Wikipedia database stores them. The "start a download" bit doesn't bother the system at all: there are many many PDF, DOC, ZIP, etc. links on Wikipedia that are handled in the same way. No worries, and nothing about the link format is too unusual compared to others that are handled just fine. I'm writing this off as an a case where the link's server might have had a split-second hiccup. Thanks, West.andrew.g (talk) 19:57, 14 July 2011 (UTC)
This is an archive of past discussions with User:West.andrew.g. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page. |
Archive 1 | Archive 2 | Archive 3 | Archive 4 | Archive 5 | → | Archive 10 |