On toxic communities

List overview All Threads
Download

newer

older

Community Wishlist Survey -...

Re: [Wikimedia-l] [Wikimedia...

Denny Vrandečić

13 Nov 2015 13 Nov '15

5:12 p.m.

Very interesting read (via Brandon Harris): http://recode.net/2015/07/07/doing-something-about-the-impossible-problem-o… "the vast majority of negative behavior ... did not originate from the persistently negative online citizens; in fact, 87 percent of online toxicity came from the neutral and positive citizens just having a bad day here or there." "... incidences of homophobia, sexism and racism ... have fallen to a combined 2 percent of all games. Verbal abuse has dropped by more than 40 percent, and 91.6 percent of negative players change their act and never commit another offense after just one reported penalty." I have plenty of ideas how to apply this to Wikipedia, but I am sure Dario and his team as well :) - and some opportunity for the communities to use such results. Cheers, Denny

Show replies by date

Chris Keating

13 Nov 13 Nov

5:26 p.m.

Really interesting - thanks for sharing! On Fri, Nov 13, 2015 at 10:12 PM, Denny Vrandečić <vrandecic(a)gmail.com> wrote:

...

Pine W

5:32 p.m.

New subject: [Analytics] On toxic communities

We're discussing this on the Research mailing list, among others. (: Pine On Fri, Nov 13, 2015 at 2:12 PM, Denny Vrandečić <vrandecic(a)gmail.com> wrote:

...

Ryan Kaldari

14 Nov 14 Nov

2:25 a.m.

New subject: [Analytics] On toxic communities

...

rupert THURNER

7:30 a.m.

New subject: [Analytics] On toxic communities

interesting. i read 90% male, 85% between 16 and 30 years of age, 12 mio players a day, 1 bio hours played a month (2012). they had a tribunal which is switched off since a year. the market is 54 mio usd a month for multiplayer online battle arena (moba) in the united states. league of legends earns 120 mio usd per month, out of a monthly player base of over 60 mio, which is 3 times more player than dota2, and 6 times more income than dota2 (beginning of 2015). * http://www.ign.com/articles/2012/10/15/riot-games-releases-awesome-league-o… * http://www.kitguru.net/gaming/development/jon-martindale/league-of-legends-… * http://www.polygon.com/2014/5/27/5723446/women-in-esports-professional-gami… * http://venturebeat.com/2015/03/24/dota-2-makes-18m-per-month-for-valve-but-… On Sat, Nov 14, 2015 at 8:25 AM, Ryan Kaldari <rkaldari(a)wikimedia.org> wrote:

...

I was skeptical of even reading this article, but it actually seems pretty insightful. It also seems more relevant to Wikipedia than I was expecting: "The answer had to be community-wide reform of cultural norms. We had to change how people thought about online society and change their expectations of what was acceptable.... How do you introduce structure and governance into a society that didn’t have one before?" It has some interesting ideas about using science to change the social dynamics of online communities and leveraging the work of academics who want to work on these problems. Some of the techniques they used remind me of Aaron's revision scoring. I wonder if there's any chance we could talk with them or some of their researchers. On Fri, Nov 13, 2015 at 3:12 PM, Denny Vrandečić <vrandecic(a)gmail.com> wrote: > Very interesting read (via Brandon Harris): > > > http://recode.net/2015/07/07/doing-something-about-the-impossible-problem-o… > > "the vast majority of negative behavior ... did not originate from the > persistently negative online citizens; in fact, 87 percent of online > toxicity came from the neutral and positive citizens just having a bad day > here or there." > > "... incidences of homophobia, sexism and racism ... have fallen to a > combined 2 percent of all games. Verbal abuse has dropped by more than 40 > percent, and 91.6 percent of negative players change their act and never > commit another offense after just one reported penalty." > > I have plenty of ideas how to apply this to Wikipedia, but I am sure Dario > and his team as well :) - and some opportunity for the communities to use > such results.

Andreas Kolbe

11:10 a.m.

New subject: [Analytics] On toxic communities

Thanks for the article, Denny. A couple of weeks ago, I posted a proposal to use machine learning to identify problematic talk page behaviour at the English Wikipedia's Village Pump.[1] The ideas seem roughly equivalent: the main aim is to make people aware of it when they are about to engage in counterproductive behaviour, and to ensure there is more timely feedback and outside input. Personally, I think the community needs a push like this in order to make that cultural shift. It is encouraging to learn that such an effort can yield tangible results in practice (something a few of the commenters at the Village Pump were doubtful about). Please review the linked Village Pump discussion, and provide input on how and whether this could be made to work in Wikipedia. [1] https://en.wikipedia.org/wiki/Wikipedia:Village_pump_(proposals)#Proposed:_… Permalink: https://en.wikipedia.org/w/index.php?title=Wikipedia:Village_pump_(proposal…

...

On Fri, Nov 13, 2015 at 3:12 PM, Denny Vrandečić <vrandecic(a)gmail.com> wrote: > Very interesting read (via Brandon Harris): > > >

http://recode.net/2015/07/07/doing-something-about-the-impossible-problem-o…

> > "the vast majority of negative behavior ... did not originate from the > persistently negative online citizens; in fact, 87 percent of online > toxicity came from the neutral and positive citizens just having a bad

day

> here or there." > > "... incidences of homophobia, sexism and racism ... have fallen to a > combined 2 percent of all games. Verbal abuse has dropped by more than

> percent, and 91.6 percent of negative players change their act and never > commit another offense after just one reported penalty." > > I have plenty of ideas how to apply this to Wikipedia, but I am sure

Dario

> and his team as well :) - and some opportunity for the communities to

use

> such results.

_______________________________________________ Wikimedia-l mailing list, guidelines at: https://meta.wikimedia.org/wiki/Mailing_lists/Guidelines Wikimedia-l(a)lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l, <mailto:wikimedia-l-request@lists.wikimedia.org?subject=unsubscribe>

Benjamin Lees

4:13 p.m.

This article highlights the happier side of things, but it appears that Lin's approach also involved completely removing bad actors: "Some players have also asked why we've taken such an aggressive stance when we've been focused on reform; well, the key here is that for most players, reform approaches are quite effective. But, for a number of players, reform attempts have been very unsuccessful which forces us to remove some of these players from League entirely."[0] A little context about League of Legends (I haven't played in a couple years, so my apologies if anything I say is out of date): * In an average game you are thrust onto a team with 4 complete strangers you will probably never meet again, and must work together to defeat the other team. * Individual player mistakes hurt the team, often a lot. Think making an error at the World Series in baseball. * A typical game lasts 20-50 minutes. If you leave the game before it finishes, you will be punished. (After 20 minutes your team can surrender if 4 of your players agree to do so.) * Some games affect your global ranking relative to all other players. These game mechanics promote a form of tension which is part of the excitement of the game but which is also sometimes stressful (if, say, you're doing really badly and your team doesn't want to quit). By and large, Wikipedia's mechanics seem very different from this, there are a few areas where users are pushed into a more hostile role with one another. In those narrow cases, like the village pump, I could maybe see benefits from trying to re-engineer interactions, but I'm skeptical that this will somehow engineer a cultural shift. P.S. As Rupert noted, over 90% of LoL players are male (how much over 90%?).[1] It would be interesting to know whether this percentage has changed along with the improvements described in the article. P.P.S. In League you have to pay if you want to transfer your account from one region to another. I'm sure we could resolve all ENGVAR disputes once and for all by adding some region locking. :-) [0] http://www.polygon.com/2014/7/21/5924203/league-of-legends-ban-code-2500-to… [1] http://majorleagueoflegends.s3.amazonaws.com/lol_infographic.png On Fri, Nov 13, 2015 at 5:12 PM, Denny Vrandečić <vrandecic(a)gmail.com> wrote:

...

Andreas Kolbe

15 Nov 15 Nov

7:56 a.m.

On Sat, Nov 14, 2015 at 9:13 PM, Benjamin Lees <emufarmers(a)gmail.com> wrote:

...

Thanks for the added context, Benjamin. Of course, banning bad actors that they consider unreformable is something Wikipedia admins have always done as well. The League of Legends team began by building a dataset of interactions that the community considered unacceptable, and then applied machine-learning to that dataset. It occurs to me that the English Wikipedia has ready access to such a dataset: it's the totality of revision-deleted and oversighted talk page posts. The League of Legends team collaborated with outside scientists to analyse their dataset. I would love to see the Wikimedia Foundation engage in a similar research project. I've added this point to the community wishlist survey: https://meta.wikimedia.org/wiki/2015_Community_Wishlist_Survey#Machine-lear…

...

P.S. As Rupert noted, over 90% of LoL players are male (how much over 90%?).[1] It would be interesting to know whether this percentage has changed along with the improvements described in the article.

Indeed.

Aaron Halfaker

11:32 a.m.

...

The League of Legends team collaborated with outside scientists to analyse their dataset. I would love to see the Wikimedia Foundation engage in a similar research project.

Oh! We are! :) When we have time. :\ One of the projects that I'd like to see done, but I've struggled to find the time for is a common talk page parser[1] that could produce a dataset of talk page interactions. I'd like this dataset to be easy to join to editor outcome measures. E.g. there might be "aggressive" talk that we don't know is problematic until we see the kind of effect that it has on other conversation participants. Anyway, I want some powerful utilities and datasets out there to help academics look into this problem more easily. For revscoring, I'd like to be able to take a set of talk page diffs, have them classified in Wiki labels[2] as "aggressive" and the build a model for ORES[3] to be used however people see fit. You could then use ORES to do offline analysis of discussions for research. You could use ORES to interrupt the a user before saving a change. I'm sure there are other clever ideas that people have for what to do with such a model that I'm happy to enable it via the service. The hard part is getting a good dataset labeled. If someone wants to invest some time and energy into this, I'm happy to work with you. We'll need more than programming help. We'll need a lot of help to figure out what dimensions we'll label talk page postings by and to do the actual labeling. 1. https://github.com/Ironholds/talk-parser 2. https://meta.wikimedia.org/wiki/Wiki_labels 3. https://meta.wikimedia.org/wiki/ORES On Sun, Nov 15, 2015 at 6:56 AM, Andreas Kolbe <jayen466(a)gmail.com> wrote:

...

On Sat, Nov 14, 2015 at 9:13 PM, Benjamin Lees <emufarmers(a)gmail.com> wrote:

Indeed. _______________________________________________ Wikimedia-l mailing list, guidelines at: https://meta.wikimedia.org/wiki/Mailing_lists/Guidelines Wikimedia-l(a)lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l, <mailto:wikimedia-l-request@lists.wikimedia.org?subject=unsubscribe>

Katherine Casey

11:40 a.m.

I'd be happy to offer my admin/oversighter experience and knowledge to help you develop the labeling and such, Aaron! I just commented on Andreas's proposal on the Community Wishlist, but to summarize here: I see a lot of potential pitfalls in trying to handle/generalize this with machine learning, but I also see a lot of potential value, and I think it's something we should be investigating. -Fluffernutter On Sun, Nov 15, 2015 at 11:32 AM, Aaron Halfaker <ahalfaker(a)wikimedia.org> wrote:

...

The League of Legends team collaborated with outside scientists to analyse their dataset. I would love to see the Wikimedia Foundation

engage

in a similar research project.

On Sat, Nov 14, 2015 at 9:13 PM, Benjamin Lees <emufarmers(a)gmail.com> wrote:

Thanks for the added context, Benjamin. Of course, banning bad actors

that

they consider unreformable is something Wikipedia admins have always done as well. The League of Legends team began by building a dataset of interactions

that

the community considered unacceptable, and then applied machine-learning

that dataset. It occurs to me that the English Wikipedia has ready access to such a dataset: it's the totality of revision-deleted and oversighted talk page posts. The League of Legends team collaborated with outside scientists to analyse their dataset. I would love to see the Wikimedia Foundation

engage

in a similar research project.

> > I've added this point to the community wishlist survey: > > > https://meta.wikimedia.org/wiki/2015_Community_Wishlist_Survey#Machine-lear…

-- Karen Brown user:Fluffernutter *Unless otherwise specified, any email sent from this address is in my volunteer capacity and does not represent the views or wishes of the Wikimedia Foundation*

Ziko van Dijk

1:21 p.m.

Hello, Just yesterday I had a long talk with a researcher about how to define and detect trolls on Wikipedia. E.g., whether "unintentional trolling" should be included or not. In my opinion, it is not possible to detect by machine trollism, unkindness, harassment, mobbing etc., maybe with the exception of swear words. A lot of turntaking, deviation from the topic and other phenomena can be experienced by the participants as positive or as negative. You might need to ask them, and even then they might not be aware of a problem that works through in subtlety. Also, third persons not involved in the conversation can be effected negatively (look at ... page X... and you know why you don't like to contribute there). Kind regards Ziko 2015-11-15 17:40 GMT+01:00 Katherine Casey <fluffernutter.wiki(a)gmail.com>om>:

...

The League of Legends team collaborated with outside scientists to analyse their dataset. I would love to see the Wikimedia Foundation

engage

in a similar research project.

On Sat, Nov 14, 2015 at 9:13 PM, Benjamin Lees <emufarmers(a)gmail.com> wrote:

Thanks for the added context, Benjamin. Of course, banning bad actors

that

they consider unreformable is something Wikipedia admins have always done as well. The League of Legends team began by building a dataset of interactions

that

the community considered unacceptable, and then applied machine-learning

engage

in a similar research project.

> > I've added this point to the community wishlist survey: > > > https://meta.wikimedia.org/wiki/2015_Community_Wishlist_Survey#Machine-lear…

-- Karen Brown user:Fluffernutter *Unless otherwise specified, any email sent from this address is in my volunteer capacity and does not represent the views or wishes of the Wikimedia Foundation* _______________________________________________ Wikimedia-l mailing list, guidelines at: https://meta.wikimedia.org/wiki/Mailing_lists/Guidelines Wikimedia-l(a)lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l, <mailto:wikimedia-l-request@lists.wikimedia.org?subject=unsubscribe>

Pharos

1:36 p.m.

The figure quoted is quite interesting, but do we have a comparable metric for the Wikimedia projects? : "... incidences of homophobia, sexism and racism ... have fallen to a combined 2 percent of all games" 2% sounds "low", but do we indeed know if this is better or worse than us? Would our comparable metric be the % of bigoted comments per article, per talk page discussion, per time that an editor spends at the project? I would think that encountering bigoted comments on 1 in 50 discussions would still be pretty significant. Thanks, Pharos On Sun, Nov 15, 2015 at 1:21 PM, Ziko van Dijk <zvandijk(a)gmail.com> wrote:

...

I'd be happy to offer my admin/oversighter experience and knowledge to

help

you develop the labeling and such, Aaron! I just commented on Andreas's proposal on the Community Wishlist, but to summarize here: I see a lot of potential pitfalls in trying to handle/generalize this with machine learning, but I also see a lot of potential value, and I think it's something we should be investigating. -Fluffernutter On Sun, Nov 15, 2015 at 11:32 AM, Aaron Halfaker <

ahalfaker(a)wikimedia.org>

wrote: > > > > The League of Legends team collaborated with outside scientists to > > analyse their dataset. I would love to see the Wikimedia Foundation > engage > > in a similar research project. > > > Oh! We are! :) When we have time. :\ One of the projects that I'd

like to

> see done, but I've struggled to find the time for is a common talk page > parser[1] that could produce a dataset of talk page interactions. I'd

> this dataset to be easy to join to editor outcome measures. E.g. there > might be "aggressive" talk that we don't know is problematic until we

see

> the kind of effect that it has on other conversation participants. > > Anyway, I want some powerful utilities and datasets out there to help > academics look into this problem more easily. For revscoring, I'd like

> be able to take a set of talk page diffs, have them classified in Wiki > labels[2] as "aggressive" and the build a model for ORES[3] to be used > however people see fit. You could then use ORES to do offline analysis

> discussions for research. You could use ORES to interrupt the a user > before saving a change. I'm sure there are other clever ideas that

people

> have for what to do with such a model that I'm happy to enable it via

the

> service. The hard part is getting a good dataset labeled. > > If someone wants to invest some time and energy into this, I'm happy to > work with you. We'll need more than programming help. We'll need a

lot of

> help to figure out what dimensions we'll label talk page postings by

and to

> do the actual labeling. > > 1. https://github.com/Ironholds/talk-parser > 2. https://meta.wikimedia.org/wiki/Wiki_labels > 3. https://meta.wikimedia.org/wiki/ORES > > On Sun, Nov 15, 2015 at 6:56 AM, Andreas Kolbe <jayen466(a)gmail.com>

wrote:

> > > On Sat, Nov 14, 2015 at 9:13 PM, Benjamin Lees <emufarmers(a)gmail.com> > > wrote: > > > > > This article highlights the happier side of things, but it appears > > > that Lin's approach also involved completely removing bad actors: > > > "Some players have also asked why we've taken such an aggressive > > > stance when we've been focused on reform; well, the key here is that > > > for most players, reform approaches are quite effective. But, for a > > > number of players, reform attempts have been very unsuccessful which > > > forces us to remove some of these players from League entirely."[0] > > > > > > > > > Thanks for the added context, Benjamin. Of course, banning bad actors > that > > they consider unreformable is something Wikipedia admins have always

done

> > as well. > > > > The League of Legends team began by building a dataset of interactions > that > > the community considered unacceptable, and then applied

machine-learning

> to > > that dataset. > > > > It occurs to me that the English Wikipedia has ready access to such a > > dataset: it's the totality of revision-deleted and oversighted talk

page

> > posts. The League of Legends team collaborated with outside

scientists to

> > analyse their dataset. I would love to see the Wikimedia Foundation > engage > > in a similar research project. > > > > I've added this point to the community wishlist survey: > > > > > > >

https://meta.wikimedia.org/wiki/2015_Community_Wishlist_Survey#Machine-lear…

> > > > > > > > > P.S. As Rupert noted, over 90% of LoL players are male (how much

over

> > > 90%?).[1] It would be interesting to know whether this percentage

has

> > > changed along with the improvements described in the article. > > > > > > > > > Indeed. > > _______________________________________________ > > Wikimedia-l mailing list, guidelines at: > > https://meta.wikimedia.org/wiki/Mailing_lists/Guidelines > > Wikimedia-l(a)lists.wikimedia.org > > Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l

<mailto:wikimedia-l-request@lists.wikimedia.org?subject=unsubscribe>

https://meta.wikimedia.org/wiki/Mailing_lists/Guidelines

Wikimedia-l(a)lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l,

<mailto:wikimedia-l-request@lists.wikimedia.org?subject=unsubscribe> _______________________________________________ Wikimedia-l mailing list, guidelines at: https://meta.wikimedia.org/wiki/Mailing_lists/Guidelines Wikimedia-l(a)lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l, <mailto:wikimedia-l-request@lists.wikimedia.org?subject=unsubscribe>

Risker

1:41 p.m.

I am going to quote Joseph Reagle, who responded to a similarly titled threat on Wiki-en-L: date:13 November 2015 at 13:48 It's been great that Riot games has had someone like Lin (an experimental psychologist) to think about issues of community and abuse. And I appreciate that Lin has been previously been so forthcoming about their experiences and findings. But the much trumpeted League of Legends Tribunal has been down "for maintenance" for months, even before this article was published, with much discussion by the community of how it was broken. On this, Riot and Lin have said nothing. Copying Joseph in case he wants to respond to some of the discussions here. Risker/Anne On 15 November 2015 at 10:36, Pharos <pharosofalexandria(a)gmail.com> wrote:

...

I'd be happy to offer my admin/oversighter experience and knowledge to

help > you develop the labeling and such, Aaron! I just commented on Andreas's > proposal on the Community Wishlist, but to summarize here: I see a lot

potential pitfalls in trying to handle/generalize this with machine learning, but I also see a lot of potential value, and I think it's something we should be investigating. -Fluffernutter On Sun, Nov 15, 2015 at 11:32 AM, Aaron Halfaker <

ahalfaker(a)wikimedia.org>

like to >> see done, but I've struggled to find the time for is a common talk

page

> parser[1] that could produce a dataset of talk page interactions. I'd

like >> this dataset to be easy to join to editor outcome measures. E.g.

there

> might be "aggressive" talk that we don't know is problematic until we

see >> the kind of effect that it has on other conversation participants. >> >> Anyway, I want some powerful utilities and datasets out there to help >> academics look into this problem more easily. For revscoring, I'd

to >> be able to take a set of talk page diffs, have them classified in Wiki >> labels[2] as "aggressive" and the build a model for ORES[3] to be used >> however people see fit. You could then use ORES to do offline

analysis

> discussions for research. You could use ORES to interrupt the a user > before saving a change. I'm sure there are other clever ideas that

people

> have for what to do with such a model that I'm happy to enable it via

the >> service. The hard part is getting a good dataset labeled. >> >> If someone wants to invest some time and energy into this, I'm happy

> work with you. We'll need more than programming help. We'll need a

lot of

> help to figure out what dimensions we'll label talk page postings by

and to

wrote: >> >> > On Sat, Nov 14, 2015 at 9:13 PM, Benjamin Lees <

emufarmers(a)gmail.com>

>> > wrote: >> > >> > > This article highlights the happier side of things, but it appears >> > > that Lin's approach also involved completely removing bad actors: >> > > "Some players have also asked why we've taken such an aggressive >> > > stance when we've been focused on reform; well, the key here is

that

>> > > for most players, reform approaches are quite effective. But, for

>> > > number of players, reform attempts have been very unsuccessful

which

>> > > forces us to remove some of these players from League

entirely."[0]

>> > > >> > >> > >> > Thanks for the added context, Benjamin. Of course, banning bad

actors

> that > > they consider unreformable is something Wikipedia admins have always

done >> > as well. >> > >> > The League of Legends team began by building a dataset of

interactions

> that > > the community considered unacceptable, and then applied

machine-learning >> to >> > that dataset. >> > >> > It occurs to me that the English Wikipedia has ready access to such

> > dataset: it's the totality of revision-deleted and oversighted talk

page

> > posts. The League of Legends team collaborated with outside

scientists to

> > analyse their dataset. I would love to see the Wikimedia Foundation > engage > > in a similar research project. > > > > I've added this point to the community wishlist survey: > > > > > > >

https://meta.wikimedia.org/wiki/2015_Community_Wishlist_Survey#Machine-lear…

> > > > > > > > > P.S. As Rupert noted, over 90% of LoL players are male (how much

over

> > > 90%?).[1] It would be interesting to know whether this percentage

has >> > > changed along with the improvements described in the article. >> > > >> > >> > >> > Indeed. >> > _______________________________________________ >> > Wikimedia-l mailing list, guidelines at: >> > https://meta.wikimedia.org/wiki/Mailing_lists/Guidelines >> > Wikimedia-l(a)lists.wikimedia.org >> > Unsubscribe:

https://lists.wikimedia.org/mailman/listinfo/wikimedia-l

, >> > <mailto:wikimedia-l-request@lists.wikimedia.org

?subject=unsubscribe>

>> > >> _______________________________________________ >> Wikimedia-l mailing list, guidelines at: >> https://meta.wikimedia.org/wiki/Mailing_lists/Guidelines >> Wikimedia-l(a)lists.wikimedia.org >> Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l

<mailto:wikimedia-l-request@lists.wikimedia.org?subject=unsubscribe>

https://meta.wikimedia.org/wiki/Mailing_lists/Guidelines

Wikimedia-l(a)lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l,

Aaron Halfaker

19 Nov 19 Nov

11:47 a.m.

I've started a thread on our "Revision scoring as a service" talk page regarding labeled conversation datasets & modeling work we could do. See https://meta.wikimedia.org/wiki/Research_talk:Revision_scoring_as_a_service… On Sun, Nov 15, 2015 at 12:41 PM, Risker <risker.wp(a)gmail.com> wrote:

...

The figure quoted is quite interesting, but do we have a comparable

metric

for the Wikimedia projects? : "... incidences of homophobia, sexism and racism ... have fallen to a combined 2 percent of all games" 2% sounds "low", but do we indeed know if this is better or worse than

us?

Would our comparable metric be the % of bigoted comments per article, per talk page discussion, per time that an editor spends at the project? I would think that encountering bigoted comments on 1 in 50 discussions

would

still be pretty significant. Thanks, Pharos On Sun, Nov 15, 2015 at 1:21 PM, Ziko van Dijk <zvandijk(a)gmail.com>

wrote:

> Hello, > > Just yesterday I had a long talk with a researcher about how to define > and detect trolls on Wikipedia. E.g., whether "unintentional trolling" > should be included or not. > > In my opinion, it is not possible to detect by machine trollism, > unkindness, harassment, mobbing etc., maybe with the exception of > swear words. A lot of turntaking, deviation from the topic and other > phenomena can be experienced by the participants as positive or as > negative. You might need to ask them, and even then they might not be > aware of a problem that works through in subtlety. Also, third persons > not involved in the conversation can be effected negatively (look at > ... page X... and you know why you don't like to contribute there). > > Kind regards > Ziko > > > 2015-11-15 17:40 GMT+01:00 Katherine Casey <

fluffernutter.wiki(a)gmail.com

>: > > I'd be happy to offer my admin/oversighter experience and knowledge

> help > > you develop the labeling and such, Aaron! I just commented on

Andreas's

> > proposal on the Community Wishlist, but to summarize here: I see a

lot

of > > potential pitfalls in trying to handle/generalize this with machine > > learning, but I also see a lot of potential value, and I think it's > > something we should be investigating. > > > > -Fluffernutter > > > > On Sun, Nov 15, 2015 at 11:32 AM, Aaron Halfaker < > ahalfaker(a)wikimedia.org> > > wrote: > > > >> > > >> > The League of Legends team collaborated with outside scientists to > >> > analyse their dataset. I would love to see the Wikimedia

Foundation

> engage > > in a similar research project. > > > Oh! We are! :) When we have time. :\ One of the projects that I'd

like to >> see done, but I've struggled to find the time for is a common talk

page > >> parser[1] that could produce a dataset of talk page interactions.

I'd

like >> this dataset to be easy to join to editor outcome measures. E.g.

there > >> might be "aggressive" talk that we don't know is problematic until

> see > >> the kind of effect that it has on other conversation participants. > >> > >> Anyway, I want some powerful utilities and datasets out there to

help

>> academics look into this problem more easily. For revscoring, I'd

like > to > >> be able to take a set of talk page diffs, have them classified in

Wiki

> >> labels[2] as "aggressive" and the build a model for ORES[3] to be

used

>> however people see fit. You could then use ORES to do offline

analysis > of > >> discussions for research. You could use ORES to interrupt the a

user

> >> before saving a change. I'm sure there are other clever ideas that > people > >> have for what to do with such a model that I'm happy to enable it

via

the >> service. The hard part is getting a good dataset labeled. >> >> If someone wants to invest some time and energy into this, I'm happy

> work with you. We'll need more than programming help. We'll need a

lot of

> help to figure out what dimensions we'll label talk page postings by

and to

wrote: >> >> > On Sat, Nov 14, 2015 at 9:13 PM, Benjamin Lees <

emufarmers(a)gmail.com> > >> > wrote: > >> > > >> > > This article highlights the happier side of things, but it

appears

> >> > > that Lin's approach also involved completely removing bad

actors:

>> > > "Some players have also asked why we've taken such an aggressive >> > > stance when we've been focused on reform; well, the key here is

that > >> > > for most players, reform approaches are quite effective. But,

for

>> > > number of players, reform attempts have been very unsuccessful

which

>> > > forces us to remove some of these players from League

entirely."[0]

>> > > >> > >> > >> > Thanks for the added context, Benjamin. Of course, banning bad

actors > >> that > >> > they consider unreformable is something Wikipedia admins have

always

done >> > as well. >> > >> > The League of Legends team began by building a dataset of

interactions > >> that > >> > the community considered unacceptable, and then applied > machine-learning > >> to > >> > that dataset. > >> > > >> > It occurs to me that the English Wikipedia has ready access to

such

a > >> > dataset: it's the totality of revision-deleted and oversighted

talk

> page > >> > posts. The League of Legends team collaborated with outside > scientists to > >> > analyse their dataset. I would love to see the Wikimedia

Foundation

> engage > > in a similar research project. > > > > I've added this point to the community wishlist survey: > > > > > > >

https://meta.wikimedia.org/wiki/2015_Community_Wishlist_Survey#Machine-lear…

> >> > > >> > > >> > > >> > > P.S. As Rupert noted, over 90% of LoL players are male (how much > over > >> > > 90%?).[1] It would be interesting to know whether this

percentage

https://lists.wikimedia.org/mailman/listinfo/wikimedia-l

, >> > <mailto:wikimedia-l-request@lists.wikimedia.org

?subject=unsubscribe> > >> > > >> _______________________________________________ > >> Wikimedia-l mailing list, guidelines at: > >> https://meta.wikimedia.org/wiki/Mailing_lists/Guidelines > >> Wikimedia-l(a)lists.wikimedia.org > >> Unsubscribe:

https://lists.wikimedia.org/mailman/listinfo/wikimedia-l

, > >> <mailto:wikimedia-l-request@lists.wikimedia.org

?subject=unsubscribe>

> >> > > > > > > > > -- > > Karen Brown > > user:Fluffernutter > > > > *Unless otherwise specified, any email sent from this address is in

> > volunteer capacity and does not represent the views or wishes of the > > Wikimedia Foundation* > > _______________________________________________ > > Wikimedia-l mailing list, guidelines at: > https://meta.wikimedia.org/wiki/Mailing_lists/Guidelines > > Wikimedia-l(a)lists.wikimedia.org > > Unsubscribe:

https://lists.wikimedia.org/mailman/listinfo/wikimedia-l,

3078

days inactive

3084

days old

wikimedia-l@lists.wikimedia.org

Manage subscription

13 comments

12 participants

tags (0)

participants (12)

Aaron Halfaker
Andreas Kolbe
Benjamin Lees
Chris Keating
Denny Vrandečić
Katherine Casey
Pharos
Pine W
Risker
rupert THURNER
Ryan Kaldari
Ziko van Dijk