(Cross-posting in case there are people not on the public list, but on the private list)
Hey all,
As you probably know, I've been tasked with building a data visualisation platform that works. I opted to look at third-party software rather than wrestle with Limn, for maintenance and speed reasons.
Still a long way to go - mostly on the back end, hooking up connectors to get the data sucked into Labs from our EventLogging schemas - but yesterday's work has produced something that looks a bit like https://upload.wikimedia.org/wikipedia/commons/0/06/Dashboard_example.png
Key takeaways from this:
1. The text, as it says, is in Markdown. If there are problems, it's incredibly simple to fix. We can list outages and inaccuracies, explain where the data comes from, all the nice stuff 2. The visualisation is embedded JavaScript and can be resized, zoomed in- and out-of and highlighted easily. 3. Key statistics are called out in highlighted boxes using common iconographic elements. 4. It's reactive. IOW, if a new dataset is loaded on the server side with more up-to-date numbers while you're reading, no problem: the figures and graphic will, too. 5. The entire thing is 60 lines of code ;)
As the dropdown menu on the left suggests, there will be a lot of different panels and panes covering data from our different platforms, and subsets of that data. If you have suggestions for things from the EventLogging schemas you'd specifically like, let me know and I'll build them in!
(Moiz: you are absolutely welcome to have at the CSS and colour scheme, too :D)
Can you define search sessions, result sets, and clickthroughs?
I'm just guessing, but are we really not clicking on a search result on 90% of the result pages? I suspect that's https://phabricator.wikimedia.org/T97310 though.
Nik
On Fri, May 8, 2015 at 7:26 AM, Oliver Keyes okeyes@wikimedia.org wrote:
(Cross-posting in case there are people not on the public list, but on the private list)
Hey all,
As you probably know, I've been tasked with building a data visualisation platform that works. I opted to look at third-party software rather than wrestle with Limn, for maintenance and speed reasons.
Still a long way to go - mostly on the back end, hooking up connectors to get the data sucked into Labs from our EventLogging schemas - but yesterday's work has produced something that looks a bit like https://upload.wikimedia.org/wikipedia/commons/0/06/Dashboard_example.png
Key takeaways from this:
- The text, as it says, is in Markdown. If there are problems, it's
incredibly simple to fix. We can list outages and inaccuracies, explain where the data comes from, all the nice stuff 2. The visualisation is embedded JavaScript and can be resized, zoomed in- and out-of and highlighted easily. 3. Key statistics are called out in highlighted boxes using common iconographic elements. 4. It's reactive. IOW, if a new dataset is loaded on the server side with more up-to-date numbers while you're reading, no problem: the figures and graphic will, too. 5. The entire thing is 60 lines of code ;)
As the dropdown menu on the left suggests, there will be a lot of different panels and panes covering data from our different platforms, and subsets of that data. If you have suggestions for things from the EventLogging schemas you'd specifically like, let me know and I'll build them in!
(Moiz: you are absolutely welcome to have at the CSS and colour scheme, too :D)
-- Oliver Keyes Research Analyst Wikimedia Foundation
Wikimedia-search mailing list Wikimedia-search@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikimedia-search
That's the plan for the descriptive section below! At the moment it's blank because I know precisely zero about the internal workings of the EL schema and what it's actually tracking, beyond the schema docs. Once I've got the framework finished I plan to throw up something genuinely dynamic and hack on the descriptive text as much as possible.
On 8 May 2015 at 11:53, Nikolas Everett neverett@wikimedia.org wrote:
Can you define search sessions, result sets, and clickthroughs?
I'm just guessing, but are we really not clicking on a search result on 90% of the result pages? I suspect that's https://phabricator.wikimedia.org/T97310 though.
Nik
On Fri, May 8, 2015 at 7:26 AM, Oliver Keyes okeyes@wikimedia.org wrote:
(Cross-posting in case there are people not on the public list, but on the private list)
Hey all,
As you probably know, I've been tasked with building a data visualisation platform that works. I opted to look at third-party software rather than wrestle with Limn, for maintenance and speed reasons.
Still a long way to go - mostly on the back end, hooking up connectors to get the data sucked into Labs from our EventLogging schemas - but yesterday's work has produced something that looks a bit like https://upload.wikimedia.org/wikipedia/commons/0/06/Dashboard_example.png
Key takeaways from this:
- The text, as it says, is in Markdown. If there are problems, it's
incredibly simple to fix. We can list outages and inaccuracies, explain where the data comes from, all the nice stuff 2. The visualisation is embedded JavaScript and can be resized, zoomed in- and out-of and highlighted easily. 3. Key statistics are called out in highlighted boxes using common iconographic elements. 4. It's reactive. IOW, if a new dataset is loaded on the server side with more up-to-date numbers while you're reading, no problem: the figures and graphic will, too. 5. The entire thing is 60 lines of code ;)
As the dropdown menu on the left suggests, there will be a lot of different panels and panes covering data from our different platforms, and subsets of that data. If you have suggestions for things from the EventLogging schemas you'd specifically like, let me know and I'll build them in!
(Moiz: you are absolutely welcome to have at the CSS and colour scheme, too :D)
-- Oliver Keyes Research Analyst Wikimedia Foundation
Wikimedia-search mailing list Wikimedia-search@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikimedia-search
Wikimedia-search mailing list Wikimedia-search@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikimedia-search
Slightly off topic, but of related interest, is Alexandre Bertails' work on visualizing RDF data.
Alex is the mastermind behind banana-rdf, the functional Scala library for working with RDF/SPARQL/etc., and has spent the better part of the last year building out Scala.js support for it:
https://github.com/w3c/banana-rdf
Check out his Scala Days presentation this year, in which he demos a 3D visualization of some RDF data:
https://www.parleys.com/tutorial/banana-rdf-interacting-web-data-scala (free signup required, sorry)
Also of interest: there is some work in progress to build a Blazegraph implementation of the banana-rdf type classes.
On Fri, May 8, 2015 at 8:55 AM, Oliver Keyes okeyes@wikimedia.org wrote:
That's the plan for the descriptive section below! At the moment it's blank because I know precisely zero about the internal workings of the EL schema and what it's actually tracking, beyond the schema docs. Once I've got the framework finished I plan to throw up something genuinely dynamic and hack on the descriptive text as much as possible.
On 8 May 2015 at 11:53, Nikolas Everett neverett@wikimedia.org wrote:
Can you define search sessions, result sets, and clickthroughs?
I'm just guessing, but are we really not clicking on a search result on
90%
of the result pages? I suspect that's https://phabricator.wikimedia.org/T97310 though.
Nik
On Fri, May 8, 2015 at 7:26 AM, Oliver Keyes okeyes@wikimedia.org
wrote:
(Cross-posting in case there are people not on the public list, but on the private list)
Hey all,
As you probably know, I've been tasked with building a data visualisation platform that works. I opted to look at third-party software rather than wrestle with Limn, for maintenance and speed reasons.
Still a long way to go - mostly on the back end, hooking up connectors to get the data sucked into Labs from our EventLogging schemas - but yesterday's work has produced something that looks a bit like
https://upload.wikimedia.org/wikipedia/commons/0/06/Dashboard_example.png
Key takeaways from this:
- The text, as it says, is in Markdown. If there are problems, it's
incredibly simple to fix. We can list outages and inaccuracies, explain where the data comes from, all the nice stuff 2. The visualisation is embedded JavaScript and can be resized, zoomed in- and out-of and highlighted easily. 3. Key statistics are called out in highlighted boxes using common iconographic elements. 4. It's reactive. IOW, if a new dataset is loaded on the server side with more up-to-date numbers while you're reading, no problem: the figures and graphic will, too. 5. The entire thing is 60 lines of code ;)
As the dropdown menu on the left suggests, there will be a lot of different panels and panes covering data from our different platforms, and subsets of that data. If you have suggestions for things from the EventLogging schemas you'd specifically like, let me know and I'll build them in!
(Moiz: you are absolutely welcome to have at the CSS and colour scheme, too :D)
-- Oliver Keyes Research Analyst Wikimedia Foundation
Wikimedia-search mailing list Wikimedia-search@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikimedia-search
Wikimedia-search mailing list Wikimedia-search@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikimedia-search
-- Oliver Keyes Research Analyst Wikimedia Foundation
Wikimedia-search mailing list Wikimedia-search@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikimedia-search
Off the top of my head, I would guess that about 20% of my searches on the office wiki give me results worth clicking on. 10% feels low.
Anecdotes == data, right? :P
Kevin Smith Agile Coach Wikimedia Foundation
Imagine a world in which every single human being can freely share in the sum of all knowledge. That's our commitment. Help us make it a reality.
On Fri, May 8, 2015 at 8:53 AM, Nikolas Everett neverett@wikimedia.org wrote:
Can you define search sessions, result sets, and clickthroughs?
I'm just guessing, but are we really not clicking on a search result on 90% of the result pages? I suspect that's https://phabricator.wikimedia.org/T97310 though.
Nik
On Fri, May 8, 2015 at 7:26 AM, Oliver Keyes okeyes@wikimedia.org wrote:
(Cross-posting in case there are people not on the public list, but on the private list)
Hey all,
As you probably know, I've been tasked with building a data visualisation platform that works. I opted to look at third-party software rather than wrestle with Limn, for maintenance and speed reasons.
Still a long way to go - mostly on the back end, hooking up connectors to get the data sucked into Labs from our EventLogging schemas - but yesterday's work has produced something that looks a bit like https://upload.wikimedia.org/wikipedia/commons/0/06/Dashboard_example.png
Key takeaways from this:
- The text, as it says, is in Markdown. If there are problems, it's
incredibly simple to fix. We can list outages and inaccuracies, explain where the data comes from, all the nice stuff 2. The visualisation is embedded JavaScript and can be resized, zoomed in- and out-of and highlighted easily. 3. Key statistics are called out in highlighted boxes using common iconographic elements. 4. It's reactive. IOW, if a new dataset is loaded on the server side with more up-to-date numbers while you're reading, no problem: the figures and graphic will, too. 5. The entire thing is 60 lines of code ;)
As the dropdown menu on the left suggests, there will be a lot of different panels and panes covering data from our different platforms, and subsets of that data. If you have suggestions for things from the EventLogging schemas you'd specifically like, let me know and I'll build them in!
(Moiz: you are absolutely welcome to have at the CSS and colour scheme, too :D)
-- Oliver Keyes Research Analyst Wikimedia Foundation
Wikimedia-search mailing list Wikimedia-search@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikimedia-search
Wikimedia-search mailing list Wikimedia-search@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikimedia-search
Yep; Nik's phab link seems relevant :)
On 8 May 2015 at 12:07, Kevin Smith ksmith@wikimedia.org wrote:
Off the top of my head, I would guess that about 20% of my searches on the office wiki give me results worth clicking on. 10% feels low.
Anecdotes == data, right? :P
Kevin Smith Agile Coach Wikimedia Foundation
Imagine a world in which every single human being can freely share in the sum of all knowledge. That's our commitment. Help us make it a reality.
On Fri, May 8, 2015 at 8:53 AM, Nikolas Everett neverett@wikimedia.org wrote:
Can you define search sessions, result sets, and clickthroughs?
I'm just guessing, but are we really not clicking on a search result on 90% of the result pages? I suspect that's https://phabricator.wikimedia.org/T97310 though.
Nik
On Fri, May 8, 2015 at 7:26 AM, Oliver Keyes okeyes@wikimedia.org wrote:
(Cross-posting in case there are people not on the public list, but on the private list)
Hey all,
As you probably know, I've been tasked with building a data visualisation platform that works. I opted to look at third-party software rather than wrestle with Limn, for maintenance and speed reasons.
Still a long way to go - mostly on the back end, hooking up connectors to get the data sucked into Labs from our EventLogging schemas - but yesterday's work has produced something that looks a bit like https://upload.wikimedia.org/wikipedia/commons/0/06/Dashboard_example.png
Key takeaways from this:
- The text, as it says, is in Markdown. If there are problems, it's
incredibly simple to fix. We can list outages and inaccuracies, explain where the data comes from, all the nice stuff 2. The visualisation is embedded JavaScript and can be resized, zoomed in- and out-of and highlighted easily. 3. Key statistics are called out in highlighted boxes using common iconographic elements. 4. It's reactive. IOW, if a new dataset is loaded on the server side with more up-to-date numbers while you're reading, no problem: the figures and graphic will, too. 5. The entire thing is 60 lines of code ;)
As the dropdown menu on the left suggests, there will be a lot of different panels and panes covering data from our different platforms, and subsets of that data. If you have suggestions for things from the EventLogging schemas you'd specifically like, let me know and I'll build them in!
(Moiz: you are absolutely welcome to have at the CSS and colour scheme, too :D)
-- Oliver Keyes Research Analyst Wikimedia Foundation
Wikimedia-search mailing list Wikimedia-search@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikimedia-search
Wikimedia-search mailing list Wikimedia-search@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikimedia-search
Wikimedia-search mailing list Wikimedia-search@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikimedia-search
That was primarily blocked on me getting the logging running on my machine so I can audit whether the events it's sending match user behaviour, and file Phab tasks to fix any discrepancies. I got it up and running yesterday, so I plan to dive in to that soon. Getting it running on your machine may be a worthwhile task for you too, Oliver.
Process note: me saying things like "It might be worth you doing X" should be treated just like anyone else saying that; your response should be "File a Phab task and get Dan to prioritise it"! :-)
Dan
On Friday, May 8, 2015, Oliver Keyes okeyes@wikimedia.org wrote:
Yep; Nik's phab link seems relevant :)
On 8 May 2015 at 12:07, Kevin Smith ksmith@wikimedia.org wrote:
Off the top of my head, I would guess that about 20% of my searches on the office wiki give me results worth clicking on. 10% feels low.
Anecdotes == data, right? :P
Kevin Smith Agile Coach Wikimedia Foundation
Imagine a world in which every single human being can freely share in the sum of all knowledge. That's our commitment. Help us make it a reality.
On Fri, May 8, 2015 at 8:53 AM, Nikolas Everett neverett@wikimedia.org
wrote:
Can you define search sessions, result sets, and clickthroughs?
I'm just guessing, but are we really not clicking on a search result
on 90%
of the result pages? I suspect that's https://phabricator.wikimedia.org/T97310 though.
Nik
On Fri, May 8, 2015 at 7:26 AM, Oliver Keyes okeyes@wikimedia.org
wrote:
(Cross-posting in case there are people not on the public list, but on the private list)
Hey all,
As you probably know, I've been tasked with building a data visualisation platform that works. I opted to look at third-party software rather than wrestle with Limn, for maintenance and speed reasons.
Still a long way to go - mostly on the back end, hooking up connectors to get the data sucked into Labs from our EventLogging schemas - but yesterday's work has produced something that looks a bit like
https://upload.wikimedia.org/wikipedia/commons/0/06/Dashboard_example.png
Key takeaways from this:
- The text, as it says, is in Markdown. If there are problems, it's
incredibly simple to fix. We can list outages and inaccuracies, explain where the data comes from, all the nice stuff 2. The visualisation is embedded JavaScript and can be resized, zoomed in- and out-of and highlighted easily. 3. Key statistics are called out in highlighted boxes using common iconographic elements. 4. It's reactive. IOW, if a new dataset is loaded on the server side with more up-to-date numbers while you're reading, no problem: the figures and graphic will, too. 5. The entire thing is 60 lines of code ;)
As the dropdown menu on the left suggests, there will be a lot of different panels and panes covering data from our different platforms, and subsets of that data. If you have suggestions for things from the EventLogging schemas you'd specifically like, let me know and I'll build them in!
(Moiz: you are absolutely welcome to have at the CSS and colour
scheme,
too :D)
-- Oliver Keyes Research Analyst Wikimedia Foundation
Wikimedia-search mailing list Wikimedia-search@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikimedia-search
Wikimedia-search mailing list Wikimedia-search@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikimedia-search
Wikimedia-search mailing list Wikimedia-search@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikimedia-search
-- Research Analyst Wikimedia Foundation
Wikimedia-search mailing list Wikimedia-search@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikimedia-search
https://lists.wikimedia.org/mailman/listinfo/wikimedia-search
Hahaha. Noted!
On 8 May 2015 at 15:05, Dan Garry dgarry@wikimedia.org wrote:
That was primarily blocked on me getting the logging running on my machine so I can audit whether the events it's sending match user behaviour, and file Phab tasks to fix any discrepancies. I got it up and running yesterday, so I plan to dive in to that soon. Getting it running on your machine may be a worthwhile task for you too, Oliver.
Process note: me saying things like "It might be worth you doing X" should be treated just like anyone else saying that; your response should be "File a Phab task and get Dan to prioritise it"! :-)
Dan
On Friday, May 8, 2015, Oliver Keyes okeyes@wikimedia.org wrote:
Yep; Nik's phab link seems relevant :)
On 8 May 2015 at 12:07, Kevin Smith ksmith@wikimedia.org wrote:
Off the top of my head, I would guess that about 20% of my searches on the office wiki give me results worth clicking on. 10% feels low.
Anecdotes == data, right? :P
Kevin Smith Agile Coach Wikimedia Foundation
Imagine a world in which every single human being can freely share in the sum of all knowledge. That's our commitment. Help us make it a reality.
On Fri, May 8, 2015 at 8:53 AM, Nikolas Everett neverett@wikimedia.org wrote:
Can you define search sessions, result sets, and clickthroughs?
I'm just guessing, but are we really not clicking on a search result on 90% of the result pages? I suspect that's https://phabricator.wikimedia.org/T97310 though.
Nik
On Fri, May 8, 2015 at 7:26 AM, Oliver Keyes okeyes@wikimedia.org wrote:
(Cross-posting in case there are people not on the public list, but on the private list)
Hey all,
As you probably know, I've been tasked with building a data visualisation platform that works. I opted to look at third-party software rather than wrestle with Limn, for maintenance and speed reasons.
Still a long way to go - mostly on the back end, hooking up connectors to get the data sucked into Labs from our EventLogging schemas - but yesterday's work has produced something that looks a bit like
https://upload.wikimedia.org/wikipedia/commons/0/06/Dashboard_example.png
Key takeaways from this:
- The text, as it says, is in Markdown. If there are problems, it's
incredibly simple to fix. We can list outages and inaccuracies, explain where the data comes from, all the nice stuff 2. The visualisation is embedded JavaScript and can be resized, zoomed in- and out-of and highlighted easily. 3. Key statistics are called out in highlighted boxes using common iconographic elements. 4. It's reactive. IOW, if a new dataset is loaded on the server side with more up-to-date numbers while you're reading, no problem: the figures and graphic will, too. 5. The entire thing is 60 lines of code ;)
As the dropdown menu on the left suggests, there will be a lot of different panels and panes covering data from our different platforms, and subsets of that data. If you have suggestions for things from the EventLogging schemas you'd specifically like, let me know and I'll build them in!
(Moiz: you are absolutely welcome to have at the CSS and colour scheme, too :D)
-- Oliver Keyes Research Analyst Wikimedia Foundation
Wikimedia-search mailing list Wikimedia-search@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikimedia-search
Wikimedia-search mailing list Wikimedia-search@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikimedia-search
Wikimedia-search mailing list Wikimedia-search@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikimedia-search
-- Research Analyst Wikimedia Foundation
Wikimedia-search mailing list Wikimedia-search@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikimedia-search
Wikimedia-search mailing list Wikimedia-search@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikimedia-search
We now have a live instance at http://searchdata.wmflabs.org/ running off static data (it'll have dynamic data when Ops gets back to me). Moiz, tinker! Dan, tell me what datasets you'd like to see! Everyone else, feel free to do either or both of those things.
On 8 May 2015 at 18:30, Oliver Keyes okeyes@wikimedia.org wrote:
Hahaha. Noted!
On 8 May 2015 at 15:05, Dan Garry dgarry@wikimedia.org wrote:
That was primarily blocked on me getting the logging running on my machine so I can audit whether the events it's sending match user behaviour, and file Phab tasks to fix any discrepancies. I got it up and running yesterday, so I plan to dive in to that soon. Getting it running on your machine may be a worthwhile task for you too, Oliver.
Process note: me saying things like "It might be worth you doing X" should be treated just like anyone else saying that; your response should be "File a Phab task and get Dan to prioritise it"! :-)
Dan
On Friday, May 8, 2015, Oliver Keyes okeyes@wikimedia.org wrote:
Yep; Nik's phab link seems relevant :)
On 8 May 2015 at 12:07, Kevin Smith ksmith@wikimedia.org wrote:
Off the top of my head, I would guess that about 20% of my searches on the office wiki give me results worth clicking on. 10% feels low.
Anecdotes == data, right? :P
Kevin Smith Agile Coach Wikimedia Foundation
Imagine a world in which every single human being can freely share in the sum of all knowledge. That's our commitment. Help us make it a reality.
On Fri, May 8, 2015 at 8:53 AM, Nikolas Everett neverett@wikimedia.org wrote:
Can you define search sessions, result sets, and clickthroughs?
I'm just guessing, but are we really not clicking on a search result on 90% of the result pages? I suspect that's https://phabricator.wikimedia.org/T97310 though.
Nik
On Fri, May 8, 2015 at 7:26 AM, Oliver Keyes okeyes@wikimedia.org wrote:
(Cross-posting in case there are people not on the public list, but on the private list)
Hey all,
As you probably know, I've been tasked with building a data visualisation platform that works. I opted to look at third-party software rather than wrestle with Limn, for maintenance and speed reasons.
Still a long way to go - mostly on the back end, hooking up connectors to get the data sucked into Labs from our EventLogging schemas - but yesterday's work has produced something that looks a bit like
https://upload.wikimedia.org/wikipedia/commons/0/06/Dashboard_example.png
Key takeaways from this:
- The text, as it says, is in Markdown. If there are problems, it's
incredibly simple to fix. We can list outages and inaccuracies, explain where the data comes from, all the nice stuff 2. The visualisation is embedded JavaScript and can be resized, zoomed in- and out-of and highlighted easily. 3. Key statistics are called out in highlighted boxes using common iconographic elements. 4. It's reactive. IOW, if a new dataset is loaded on the server side with more up-to-date numbers while you're reading, no problem: the figures and graphic will, too. 5. The entire thing is 60 lines of code ;)
As the dropdown menu on the left suggests, there will be a lot of different panels and panes covering data from our different platforms, and subsets of that data. If you have suggestions for things from the EventLogging schemas you'd specifically like, let me know and I'll build them in!
(Moiz: you are absolutely welcome to have at the CSS and colour scheme, too :D)
-- Oliver Keyes Research Analyst Wikimedia Foundation
Wikimedia-search mailing list Wikimedia-search@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikimedia-search
Wikimedia-search mailing list Wikimedia-search@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikimedia-search
Wikimedia-search mailing list Wikimedia-search@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikimedia-search
-- Research Analyst Wikimedia Foundation
Wikimedia-search mailing list Wikimedia-search@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikimedia-search
Wikimedia-search mailing list Wikimedia-search@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikimedia-search
-- Oliver Keyes Research Analyst Wikimedia Foundation
\o/
On Mon, May 11, 2015 at 10:47 AM, Oliver Keyes okeyes@wikimedia.org wrote:
We now have a live instance at http://searchdata.wmflabs.org/ running off static data (it'll have dynamic data when Ops gets back to me). Moiz, tinker! Dan, tell me what datasets you'd like to see! Everyone else, feel free to do either or both of those things.
On 8 May 2015 at 18:30, Oliver Keyes okeyes@wikimedia.org wrote:
Hahaha. Noted!
On 8 May 2015 at 15:05, Dan Garry dgarry@wikimedia.org wrote:
That was primarily blocked on me getting the logging running on my
machine
so I can audit whether the events it's sending match user behaviour, and file Phab tasks to fix any discrepancies. I got it up and running
yesterday,
so I plan to dive in to that soon. Getting it running on your machine
may be
a worthwhile task for you too, Oliver.
Process note: me saying things like "It might be worth you doing X"
should
be treated just like anyone else saying that; your response should be
"File
a Phab task and get Dan to prioritise it"! :-)
Dan
On Friday, May 8, 2015, Oliver Keyes okeyes@wikimedia.org wrote:
Yep; Nik's phab link seems relevant :)
On 8 May 2015 at 12:07, Kevin Smith ksmith@wikimedia.org wrote:
Off the top of my head, I would guess that about 20% of my searches
on
the office wiki give me results worth clicking on. 10% feels low.
Anecdotes == data, right? :P
Kevin Smith Agile Coach Wikimedia Foundation
Imagine a world in which every single human being can freely share in the sum of all knowledge. That's our commitment. Help us make it a reality.
On Fri, May 8, 2015 at 8:53 AM, Nikolas Everett <
neverett@wikimedia.org>
wrote:
Can you define search sessions, result sets, and clickthroughs?
I'm just guessing, but are we really not clicking on a search
result on
90% of the result pages? I suspect that's https://phabricator.wikimedia.org/T97310 though.
Nik
On Fri, May 8, 2015 at 7:26 AM, Oliver Keyes okeyes@wikimedia.org wrote: > > (Cross-posting in case there are people not on the public list,
but on
> the private list) > > Hey all, > > As you probably know, I've been tasked with building a data > visualisation platform that works. I opted to look at third-party > software rather than wrestle with Limn, for maintenance and speed > reasons. > > Still a long way to go - mostly on the back end, hooking up
connectors
> to get the data sucked into Labs from our EventLogging schemas -
but
> yesterday's work has produced something that looks a bit like > >
https://upload.wikimedia.org/wikipedia/commons/0/06/Dashboard_example.png
> > Key takeaways from this: > > 1. The text, as it says, is in Markdown. If there are problems,
it's
> incredibly simple to fix. We can list outages and inaccuracies, > explain where the data comes from, all the nice stuff > 2. The visualisation is embedded JavaScript and can be resized,
zoomed
> in- and out-of and highlighted easily. > 3. Key statistics are called out in highlighted boxes using common > iconographic elements. > 4. It's reactive. IOW, if a new dataset is loaded on the server
side
> with more up-to-date numbers while you're reading, no problem: the > figures and graphic will, too. > 5. The entire thing is 60 lines of code ;) > > As the dropdown menu on the left suggests, there will be a lot of > different panels and panes covering data from our different
platforms,
> and subsets of that data. If you have suggestions for things from
the
> EventLogging schemas you'd specifically like, let me know and I'll > build them in! > > (Moiz: you are absolutely welcome to have at the CSS and colour > scheme, > too :D) > > -- > Oliver Keyes > Research Analyst > Wikimedia Foundation > > _______________________________________________ > Wikimedia-search mailing list > Wikimedia-search@lists.wikimedia.org > https://lists.wikimedia.org/mailman/listinfo/wikimedia-search
Wikimedia-search mailing list Wikimedia-search@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikimedia-search
Wikimedia-search mailing list Wikimedia-search@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikimedia-search
-- Research Analyst Wikimedia Foundation
Wikimedia-search mailing list Wikimedia-search@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikimedia-search
Wikimedia-search mailing list Wikimedia-search@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikimedia-search
-- Oliver Keyes Research Analyst Wikimedia Foundation
-- Oliver Keyes Research Analyst Wikimedia Foundation
Wikimedia-search mailing list Wikimedia-search@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikimedia-search
Just a wee signal boost to this for (Moiz|Wes|Dan if he's no longer ill) :)
On 11 May 2015 at 13:57, James Douglas jdouglas@wikimedia.org wrote:
\o/
On Mon, May 11, 2015 at 10:47 AM, Oliver Keyes okeyes@wikimedia.org wrote:
We now have a live instance at http://searchdata.wmflabs.org/ running off static data (it'll have dynamic data when Ops gets back to me). Moiz, tinker! Dan, tell me what datasets you'd like to see! Everyone else, feel free to do either or both of those things.
On 8 May 2015 at 18:30, Oliver Keyes okeyes@wikimedia.org wrote:
Hahaha. Noted!
On 8 May 2015 at 15:05, Dan Garry dgarry@wikimedia.org wrote:
That was primarily blocked on me getting the logging running on my machine so I can audit whether the events it's sending match user behaviour, and file Phab tasks to fix any discrepancies. I got it up and running yesterday, so I plan to dive in to that soon. Getting it running on your machine may be a worthwhile task for you too, Oliver.
Process note: me saying things like "It might be worth you doing X" should be treated just like anyone else saying that; your response should be "File a Phab task and get Dan to prioritise it"! :-)
Dan
On Friday, May 8, 2015, Oliver Keyes okeyes@wikimedia.org wrote:
Yep; Nik's phab link seems relevant :)
On 8 May 2015 at 12:07, Kevin Smith ksmith@wikimedia.org wrote:
Off the top of my head, I would guess that about 20% of my searches on the office wiki give me results worth clicking on. 10% feels low.
Anecdotes == data, right? :P
Kevin Smith Agile Coach Wikimedia Foundation
Imagine a world in which every single human being can freely share in the sum of all knowledge. That's our commitment. Help us make it a reality.
On Fri, May 8, 2015 at 8:53 AM, Nikolas Everett neverett@wikimedia.org wrote: > Can you define search sessions, result sets, and clickthroughs? > > I'm just guessing, but are we really not clicking on a search > result on > 90% > of the result pages? I suspect that's > https://phabricator.wikimedia.org/T97310 though. > > Nik > > On Fri, May 8, 2015 at 7:26 AM, Oliver Keyes okeyes@wikimedia.org > wrote: >> >> (Cross-posting in case there are people not on the public list, >> but on >> the private list) >> >> Hey all, >> >> As you probably know, I've been tasked with building a data >> visualisation platform that works. I opted to look at third-party >> software rather than wrestle with Limn, for maintenance and speed >> reasons. >> >> Still a long way to go - mostly on the back end, hooking up >> connectors >> to get the data sucked into Labs from our EventLogging schemas - >> but >> yesterday's work has produced something that looks a bit like >> >> >> https://upload.wikimedia.org/wikipedia/commons/0/06/Dashboard_example.png >> >> Key takeaways from this: >> >> 1. The text, as it says, is in Markdown. If there are problems, >> it's >> incredibly simple to fix. We can list outages and inaccuracies, >> explain where the data comes from, all the nice stuff >> 2. The visualisation is embedded JavaScript and can be resized, >> zoomed >> in- and out-of and highlighted easily. >> 3. Key statistics are called out in highlighted boxes using common >> iconographic elements. >> 4. It's reactive. IOW, if a new dataset is loaded on the server >> side >> with more up-to-date numbers while you're reading, no problem: the >> figures and graphic will, too. >> 5. The entire thing is 60 lines of code ;) >> >> As the dropdown menu on the left suggests, there will be a lot of >> different panels and panes covering data from our different >> platforms, >> and subsets of that data. If you have suggestions for things from >> the >> EventLogging schemas you'd specifically like, let me know and I'll >> build them in! >> >> (Moiz: you are absolutely welcome to have at the CSS and colour >> scheme, >> too :D) >> >> -- >> Oliver Keyes >> Research Analyst >> Wikimedia Foundation >> >> _______________________________________________ >> Wikimedia-search mailing list >> Wikimedia-search@lists.wikimedia.org >> https://lists.wikimedia.org/mailman/listinfo/wikimedia-search > > > > _______________________________________________ > Wikimedia-search mailing list > Wikimedia-search@lists.wikimedia.org > https://lists.wikimedia.org/mailman/listinfo/wikimedia-search >
Wikimedia-search mailing list Wikimedia-search@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikimedia-search
-- Research Analyst Wikimedia Foundation
Wikimedia-search mailing list Wikimedia-search@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikimedia-search
Wikimedia-search mailing list Wikimedia-search@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikimedia-search
-- Oliver Keyes Research Analyst Wikimedia Foundation
-- Oliver Keyes Research Analyst Wikimedia Foundation
Wikimedia-search mailing list Wikimedia-search@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikimedia-search
Wikimedia-search mailing list Wikimedia-search@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikimedia-search
wikimedia-search@lists.wikimedia.org