New Google interface to Wikipedia

List overview All Threads
Download

newer

older

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

Office hour with WMF researchers

Anders Wennersten

29 Sep 2013 29 Sep '13

1:02 p.m.

The number of accesses to Wikipedia has jumped 25-40% since July [1] and I suspect it is related to the new interface Google has created where an extract from Wikipedia resides on the right part of search page. It is very neat feature and I have noticed you get it from the language version you are used to, meaning it is not only using enwp as Facebook does. It also uses some intelligent way of doing the extract as it is not 100% as the wikiepdiatext and also find illustrations from other sources than Wikipedia/commons. Does anyone know the background and technique/algorithms behind, or have they developed this all by themselves? Anders [1) http://stats.wikimedia.org/EN/TablesPageViewsMonthlyCombined.htm

Show replies by date

Fæ

29 Sep 29 Sep

1:10 p.m.

On 29 September 2013 11:02, Anders Wennersten <mail(a)anderswennersten.se> wrote:

...

Oliver Keyes

30 Sep 30 Sep

5:26 a.m.

We actually hosted a couple of people from the search team who work with Wikipedia at the office the other day. If people are interested in talking to them about this kinda thing, drop me an email and I'll put you in touch. On 29 September 2013 03:10, Fæ <faewik(a)gmail.com> wrote:

...

On 29 September 2013 11:02, Anders Wennersten <mail(a)anderswennersten.se> wrote:

The number of accesses to Wikipedia has jumped 25-40% since July [1] and

suspect it is related to the new interface Google has created where an extract from Wikipedia resides on the right part of search page. It is very neat feature and I have noticed you get it from the language version you are used to, meaning it is not only using enwp as Facebook

does.

It also uses some intelligent way of doing the extract as it is not 100%

the wikiepdiatext and also find illustrations from other sources than Wikipedia/commons. Does anyone know the background and technique/algorithms behind, or have they developed this all by themselves? Anders [1) http://stats.wikimedia.org/EN/TablesPageViewsMonthlyCombined.htm

I know that I'm rather impressed with my Nexus 7, which nicely speaks back to me aloud, explaining answers to my spoken questions using text from Wikipedia. My Mother, who is in her mid 70's, finds the voice interface intuitive and needed no training before using this as a way to access the 'sum of human knowledge'. :-D I can't fault the outcome of this development work, it would be interesting to find out how much of the code is reusable or if there are plans for an API so we can piggy-back some interesting apps on it. Fae -- faewik(a)gmail.com http://j.mp/faewm _______________________________________________ Wikimedia-l mailing list Wikimedia-l(a)lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l, <mailto:wikimedia-l-request@lists.wikimedia.org?subject=unsubscribe>

-- Oliver Keyes Product Analyst Wikimedia Foundation

James Alexander

5:42 a.m.

What Oliver said, They are really interested in finding ways to feed back into the ecosystem where there algorithm says that there is a mistake (interlanguage links pointing to two different subjects, contradictions between article and infobox etc). They are probably going to reach out to wikidata (and Lydia) first for the interlanguage links but it would be great to have some community members helping to shepherd them and find good uses for their data. Oliver's your man! <runs away> James Alexander Legal and Community Advocacy Wikimedia Foundation (415) 839-6885 x6716 @jamesofur On Sun, Sep 29, 2013 at 7:26 PM, Oliver Keyes <okeyes(a)wikimedia.org> wrote:

...

On 29 September 2013 11:02, Anders Wennersten <mail(a)anderswennersten.se> wrote: > The number of accesses to Wikipedia has jumped 25-40% since July [1]

and

does. > It also uses some intelligent way of doing the extract as it is not

100%

as > the wikiepdiatext and also find illustrations from other sources than > Wikipedia/commons. > > Does anyone know the background and technique/algorithms behind, or

have

they developed this all by themselves? Anders [1) http://stats.wikimedia.org/EN/TablesPageViewsMonthlyCombined.htm

-- Oliver Keyes Product Analyst Wikimedia Foundation _______________________________________________ Wikimedia-l mailing list Wikimedia-l(a)lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l, <mailto:wikimedia-l-request@lists.wikimedia.org?subject=unsubscribe>

David Gerard

10:29 a.m.

On 30 September 2013 03:42, James Alexander <jalexander(a)wikimedia.org> wrote:

...

They are really interested in finding ways to feed back into the ecosystem where there algorithm says that there is a mistake (interlanguage links pointing to two different subjects, contradictions between article and infobox etc). They are probably going to reach out to wikidata (and Lydia) first for the interlanguage links but it would be great to have some community members helping to shepherd them and find good uses for their data. Oliver's your man! <runs away>

Gerard Meijssen

11:57 a.m.

Hoi, Wikidata has a limit of one link of a Wikipedia article to Wikidata. There is no such thing as a compulsory minimum of two articles. As a matter a fact there are many Wikidata items that have no relations to Wikipedia articles. To make this even more obvious, Wikidata also links to Commons and Wikivoyage. This means that Wikidata is not Wikipedia centric. The main point of Wikidata is that its information can be used in our projects, that it is useful in bringing the sum of all knowledge to our readers. That it does. There are a number of Wikipedias who use the information of Wikidata for their stubs. Consequently when the information of the subjects of these articles are enriched in Wikidata, new information flows into these Wikipedias. Thanks, GerardM On 30 September 2013 09:29, David Gerard <dgerard(a)gmail.com> wrote:

...

On 30 September 2013 03:42, James Alexander <jalexander(a)wikimedia.org> wrote:

They are really interested in finding ways to feed back into the

ecosystem

where there algorithm says that there is a mistake (interlanguage links pointing to two different subjects, contradictions between article and infobox etc). They are probably going to reach out to wikidata (and

Lydia)

first for the interlanguage links but it would be great to have some community members helping to shepherd them and find good uses for their data. Oliver's your man! <runs away>

I notice with Wikidata-based interwikis, there's now an enforced 1-1 relationship between related articles in different wikis. Which is not *always* accurate, but is considerably easier to machine-analyse. - d. _______________________________________________ Wikimedia-l mailing list Wikimedia-l(a)lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l, <mailto:wikimedia-l-request@lists.wikimedia.org?subject=unsubscribe>

Federico Leva (Nemo)

7:15 a.m.

Anders Wennersten, 29/09/2013 12:02:

...

The number of accesses to Wikipedia has jumped 25-40% since July [1]

You mean positive +25 % don't you? I don't see this figure anywhere, where are you looking? There is a +25 % in August compared to July in sv.wiki (but single-month peaks are not to be trusted that much).

...

and I suspect it is related to the new interface Google has created where an extract from Wikipedia resides on the right part of search page.

https://en.wikipedia.org/wiki/Knowledge_Graph Usually it's been anectodically blamed for reducing rather than increasing visits. There is no clear timeline of the changes they did so it's hard to make correlations in a serious way (nobody even tried so far, I think?). Nemo

Anders Wennersten

7:45 a.m.

Federico Leva (Nemo) skrev 2013-09-30 06:15:

...

Anders Wennersten, 29/09/2013 12:02:

The number of accesses to Wikipedia has jumped 25-40% since July [1]

Look at September figures too. The total is over 25 000 up 25% from July Anders

Federico Leva (Nemo)

8:15 a.m.

Anders Wennersten, 30/09/2013 06:45:

...

Look at September figures too. The total is over 25 000 up 25% from July

Oh, sure. I tend to ignore projections but they're at the 28th already. I'm curious to see next months, indeed the table doesn't have many cases of two consecutive months with two-figures increase. Nemo

3859

days inactive

3860

days old

wikimedia-l@lists.wikimedia.org

Manage subscription

8 comments

7 participants

tags (0)

participants (7)

Anders Wennersten
David Gerard
Federico Leva (Nemo)
Fæ
Gerard Meijssen
James Alexander
Oliver Keyes