Re: [Wikidata-l] Cooperation between Wikidata and DBpedia and Wikipedia

List overview All Threads
Download

newer

older

Re: [Wikidata-l] Cooperation...

[Wikidata-l] Counting sitelinks -...

Gerard Meijssen

23 Aug 2013 23 Aug '13

5:53 a.m.

Hoi, I have been blogging a lot the last two days with DBpedia in mind. My understanding is that at DBpedia a lot of effort went into making something of a cohesive model of properties. Now that the "main type GND" is about to be deleted, it makes sense to adopt much of the work that has been done at DBpedia. The benefits are: - we will get access to academically reviewed data structures - we do not have to wait and ponder and get in to thebusiness enriching the data content of DBpedia - we can easily compare the data in DBpedia and Wikidata - more importantly, DBpedia has spend effort in connecting to other resources Yes, we can import data from DBpedia and we can import data from Wikipedia. Actually we can do both. The one thing that needs to be considered is that we need data before we can curate it. With more data available it becomes more relevant to invest time in tools that compare data. We can start doing this now and, over time this will become more relevant. But now we need more properties and the associated date. What do you think? Thanks, GerardM http://ultimategerard.blogspot.com

Attachments:

attachment.htm (text/html — 1.4 KB)

Show replies by date

Luca Martinelli

23 Aug 23 Aug

5:59 a.m.

New subject: [Wikidata-l] Cooperation between Wikidata and DBpedia and Wikipedia

Il giorno 23/ago/2013 11:54, "Gerard Meijssen" <gerard.meijssen(a)gmail.com

...

ha scritto:

...

Gerard Meijssen

6:28 a.m.

New subject: [Wikidata-l] Cooperation between Wikidata and DBpedia and Wikipedia

Hoi, With the "main type (GND)" we were imposing a classification that was developed by the Deutsche National Bibliothek. It may work for them as they have a specific application for their data but it does not necessarily work elsewhere. In my blogpost I refer to the "infoboxes task force". Information that exists in many infoboxes cannot be expressed in Wikidata. DBpedia has collected data from Wikipedias (more than hundred languages). It will follow that they have worked hard on a system that brings all the data together in a working set of properties. I am convinced that there will be issues with what they currently have. But unlike with the GND classification we can move away from it by making it our own. (it would no longer be the GND classification) Thanks, GerardM http://ultimategerardm.blogspot.nl/2013/08/wikidata-infoboxes-task-force.ht… On 23 August 2013 11:59, Luca Martinelli <martinelliluca(a)gmail.com> wrote:

...

Il giorno 23/ago/2013 11:54, "Gerard Meijssen" <gerard.meijssen(a)gmail.com> ha scritto:

[...] Now that the "main type GND" is about to be deleted, it makes sense to adopt much of the work that has been done at DBpedia. [...] May you explain better what we may want to adopt and which properties are missing? L. _______________________________________________ Wikidata-l mailing list Wikidata-l(a)lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikidata-l

Tom Morris

10:15 a.m.

New subject: [Wikidata-l] Cooperation between Wikidata and DBpedia and Wikipedia

On Fri, Aug 23, 2013 at 5:53 AM, Gerard Meijssen <gerard.meijssen(a)gmail.com>wrote;wrote:

...

I have been blogging a lot the last two days with DBpedia in mind. My understanding is that at DBpedia a lot of effort went into making something of a cohesive model of properties. Now that the "main type GND" is about to be deleted, it makes sense to adopt much of the work that has been done at DBpedia. The benefits are: - we will get access to academically reviewed data structures - we do not have to wait and ponder and get in to thebusiness enriching the data content of DBpedia - we can easily compare the data in DBpedia and Wikidata - more importantly, DBpedia has spend effort in connecting to other resources Yes, we can import data from DBpedia and we can import data from Wikipedia. Actually we can do both. The one thing that needs to be considered is that we need data before we can curate it. With more data available it becomes more relevant to invest time in tools that compare data. We can start doing this now and, over time this will become more relevant. But now we need more properties and the associated date.

I think reviewing existing ontologies/schemas like DBpedia (or Freebase) with an eye towards reusing them or incorporating pieces of them makes a lot of sense. I wouldn't take them wholesale without review though. Importing data from DBpedia, I'd be much more wary of. It can vary greatly in quality depending on how it was generated. I'd much rather see WikiData take Freebase's approach of quality over quantity and let coverage improve over time. Tom

Gerard Meijssen

10:58 a.m.

New subject: [Wikidata-l] Cooperation between Wikidata and DBpedia and Wikipedia

Hoi, The questions are: - would we advance a lot when we adopt the DBpedia schema as it is? - Would we be open to include substantially more data? - When we adopt the schema, can we tinker with it to suit our needs? If the answers to these questions are yes, what is the point in procrastinating??? One advantage of DBpedia over Freebase is that there is a real close connection between it and the many languages it retrieves data from (more than a hundred). If anything, I would really appreciate it when Wikidata is much richer in data.. I just created a person who was head of state of Uruguay, there is clear information in the es.wikipedia and we could just have it when we look for the inclusion of DBpedia data.. Another thing to mention is that a new version based on the April dump is about to be released (I have been told). One other big thing of DBpedia is that it is connected to many external resources. This will make it possible to verify our data against these other sources. This is imho the more important thing to do with the time of our volunteers. Doing the things that have already been done is a waste of time. Thanks, Gerard On 23 August 2013 16:15, Tom Morris <tfmorris(a)gmail.com> wrote:

...

On Fri, Aug 23, 2013 at 5:53 AM, Gerard Meijssen < gerard.meijssen(a)gmail.com> wrote:

David Cuenca

11:18 a.m.

New subject: [Wikidata-l] Cooperation between Wikidata and DBpedia and Wikipedia

Hi, I will answer with questions with more questions... On Fri, Aug 23, 2013 at 10:58 AM, Gerard Meijssen <gerard.meijssen(a)gmail.com

...

wrote:

...

Hoi, The questions are: - would we advance a lot when we adopt the DBpedia schema as it is? Which schema? All of them? Some? Article classification? Infobox

extraction? Wikidata is going to be linked to the infoboxes in Wikipedia, so the priority is to support those needs, not to replicate any schema.

...

- Would we be open to include substantially more data? Which data? All of it? What is the reliability?

...

- When we adopt the schema, can we tinker with it to suit our needs? Again, could you please give some example of what to import and how should

it be adapted?

...

If the answers to these questions are yes, what is the point in procrastinating???

Do we have already all the datatypes that would be needed? Most of the properties that are missing is because of the lack of "value" or others.

...

One other big thing of DBpedia is that it is connected to many external resources. This will make it possible to verify our data against these other sources. This is imho the more important thing to do with the time of our volunteers. Doing the things that have already been done is a waste of time.

Dimitris Kontokostas

26 Aug 26 Aug

3:16 a.m.

New subject: [Wikidata-l] Cooperation between Wikidata and DBpedia and Wikipedia

Speaking from DBpedia (not on behalf), we have been always trying to find ways to contribute data back to wikipedia and If licencing is the only issue here I am sure we can make any necessary arrangements. imho the main scepticism so far was "trust" from the WIkipedia community to load data in bulk. However, there are datasets of very high quality in DBpedia that could be used for that purpose. Recently we are experimenting in an alternative where people can manually import single facts in Wikidata from the DBpedia resource interface. Here's a recent talk about this [1] on the tech list. Any comments on that are also welcome. Best, Dimitris [1] http://lists.wikimedia.org/pipermail/wikidata-tech/2013-August/000189.html On Fri, Aug 23, 2013 at 6:18 PM, David Cuenca <dacuetu(a)gmail.com> wrote:

...

Hi, I will answer with questions with more questions... On Fri, Aug 23, 2013 at 10:58 AM, Gerard Meijssen < gerard.meijssen(a)gmail.com> wrote:

Hoi, The questions are: - would we advance a lot when we adopt the DBpedia schema as it is? Which schema? All of them? Some? Article classification? Infobox

extraction? Wikidata is going to be linked to the infoboxes in Wikipedia, so the priority is to support those needs, not to replicate any schema.

- Would we be open to include substantially more data? Which data? All of it? What is the reliability?

- When we adopt the schema, can we tinker with it to suit our needs? Again, could you please give some example of what to import and how

should it be adapted?

If the answers to these questions are yes, what is the point in procrastinating???

Do we have already all the datatypes that would be needed? Most of the properties that are missing is because of the lack of "value" or others.

The thing is that if those resources already are in dbpedia, we can just use dbpedia as a bridge, that is how linked data is supposed to be... no need to replicate everything, but of course, if it is worth replicating, we can go through case by case. Micru _______________________________________________ Wikidata-l mailing list Wikidata-l(a)lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikidata-l

-- Dimitris Kontokostas Department of Computer Science, University of Leipzig Research Group: http://aksw.org Homepage:http://aksw.org/DimitrisKontokostas

Gerard Meijssen

4:34 a.m.

New subject: [Wikidata-l] Cooperation between Wikidata and DBpedia and Wikipedia

Hoi, Two things to consider; the Wikipedia community and the Wikidata community are two separate entities. As far as I am concerned, Wikidata needs more data to get to the tipping point where it becomes useful to users. I am really vocal about both. The license has traditionally been a sticking point and the "not invented here" aspect of DBpedia is as well. You are effectively outside the Wikipedia community.. The other part is that the data of Wikidata is continually updated while DBpedia is not. So in my opinion the best thing that can happen is when DBpedia DOES update with Wikidata. The point is not to absorb it without thinking. What DBpedia has is a set of properties that work on the data that it has gleaned from the many Wikipedias. There is undoubtedly a lot of documentation on it and, it would be good when this is taken into consideration when accepting and proposing new properties for Wikidata. Obviously only the properties that are currently supported can be proposed at this time. I am quite willing to propose properties based on DBpedia (but so can you). With the properties in place, we can import data. We can import it from both DBpedia and from Wikipedia. Sanity checks are needed for both sources and as far as I am aware we do not have sanity checks at Wikidata. What we do have is the possibility to compare data with other sources and that is where the Wikidata community needs to grow and that is why we need much data in the first place in order to get into this. However, we can start building the tools to do this. I hope the DBpedia community can help us with that. Thanks, GerardM On 26 August 2013 09:16, Dimitris Kontokostas < kontokostas(a)informatik.uni-leipzig.de> wrote:

...

Hi, I will answer with questions with more questions... On Fri, Aug 23, 2013 at 10:58 AM, Gerard Meijssen < gerard.meijssen(a)gmail.com> wrote:

Hoi, The questions are: - would we advance a lot when we adopt the DBpedia schema as it is? Which schema? All of them? Some? Article classification? Infobox

extraction? Wikidata is going to be linked to the infoboxes in Wikipedia, so the priority is to support those needs, not to replicate any schema.

- Would we be open to include substantially more data? Which data? All of it? What is the reliability?

- When we adopt the schema, can we tinker with it to suit our needs? Again, could you please give some example of what to import and how

should it be adapted?

If the answers to these questions are yes, what is the point in procrastinating???

Do we have already all the datatypes that would be needed? Most of the properties that are missing is because of the lack of "value" or others.

-- Dimitris Kontokostas Department of Computer Science, University of Leipzig Research Group: http://aksw.org Homepage:http://aksw.org/DimitrisKontokostas _______________________________________________ Wikidata-l mailing list Wikidata-l(a)lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikidata-l

Kingsley Idehen

9:27 a.m.

New subject: [Wikidata-l] Cooperation between Wikidata and DBpedia and Wikipedia

On 8/26/13 4:34 AM, Gerard Meijssen wrote:

...

What is this "not invented here" aspect of DBpedia?

...

You are effectively outside the Wikipedia community.. The other part is that the data of Wikidata is continually updated while DBpedia is not.

What do you mean by that? Are you referring to schema/ontology/vocabulary evolution or instance data evolution? Remember, there are live editions of DBpedia.

...

So in my opinion the best thing that can happen is when DBpedia DOES update with Wikidata. The point is not to absorb it without thinking.

Yes, the first point of clarity would be when Wikidata produces a dump that can be ingested by DBpedia and any other data space in the LOD cloud. All that's required is data publication in Linked Data form.

...

What DBpedia has is a set of properties that work on the data that it has gleaned from the many Wikipedias. There is undoubtedly a lot of documentation on it and, it would be good when this is taken into consideration when accepting and proposing new properties for Wikidata. Obviously only the properties that are currently supported can be proposed at this time. I am quite willing to propose properties based on DBpedia (but so can you). With the properties in place, we can import data. We can import it from both DBpedia and from Wikipedia. Sanity checks are needed for both sources and as far as I am aware we do not have sanity checks at Wikidata. What we do have is the possibility to compare data with other sources and that is where the Wikidata community needs to grow and that is why we need much data in the first place in order to get into this. However, we can start building the tools to do this. I hope the DBpedia community can help us with that. Thanks, GerardM

As far as I know, DBpedia has always been interested in collaboration with Wikidata. Kingsley

...

On 26 August 2013 09:16, Dimitris Kontokostas <kontokostas(a)informatik.uni-leipzig.de <mailto:kontokostas@informatik.uni-leipzig.de>> wrote: Speaking from DBpedia (not on behalf), we have been always trying to find ways to contribute data back to wikipedia and If licencing is the only issue here I am sure we can make any necessary arrangements. imho the main scepticism so far was "trust" from the WIkipedia community to load data in bulk. However, there are datasets of very high quality in DBpedia that could be used for that purpose. Recently we are experimenting in an alternative where people can manually import single facts in Wikidata from the DBpedia resource interface. Here's a recent talk about this [1] on the tech list. Any comments on that are also welcome. Best, Dimitris [1] http://lists.wikimedia.org/pipermail/wikidata-tech/2013-August/000189.html On Fri, Aug 23, 2013 at 6:18 PM, David Cuenca <dacuetu(a)gmail.com <mailto:dacuetu@gmail.com>> wrote: Hi, I will answer with questions with more questions... On Fri, Aug 23, 2013 at 10:58 AM, Gerard Meijssen <gerard.meijssen(a)gmail.com <mailto:gerard.meijssen@gmail.com>> wrote: Hoi, The questions are: * would we advance a lot when we adopt the DBpedia schema as it is? Which schema? All of them? Some? Article classification? Infobox extraction? Wikidata is going to be linked to the infoboxes in Wikipedia, so the priority is to support those needs, not to replicate any schema. * Would we be open to include substantially more data? Which data? All of it? What is the reliability? * When we adopt the schema, can we tinker with it to suit our needs? Again, could you please give some example of what to import and how should it be adapted? If the answers to these questions are yes, what is the point in procrastinating??? Do we have already all the datatypes that would be needed? Most of the properties that are missing is because of the lack of "value" or others. One other big thing of DBpedia is that it is connected to many external resources. This will make it possible to verify our data against these other sources. This is imho the more important thing to do with the time of our volunteers. Doing the things that have already been done is a waste of time. The thing is that if those resources already are in dbpedia, we can just use dbpedia as a bridge, that is how linked data is supposed to be... no need to replicate everything, but of course, if it is worth replicating, we can go through case by case. Micru _______________________________________________ Wikidata-l mailing list Wikidata-l(a)lists.wikimedia.org <mailto:Wikidata-l@lists.wikimedia.org> https://lists.wikimedia.org/mailman/listinfo/wikidata-l -- Dimitris Kontokostas Department of Computer Science, University of Leipzig Research Group: http://aksw.org Homepage:http://aksw.org/DimitrisKontokostas _______________________________________________ Wikidata-l mailing list Wikidata-l(a)lists.wikimedia.org <mailto:Wikidata-l@lists.wikimedia.org> https://lists.wikimedia.org/mailman/listinfo/wikidata-l _______________________________________________ Wikidata-l mailing list Wikidata-l(a)lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikidata-l

-- Regards, Kingsley Idehen Founder & CEO OpenLink Software Company Web: http://www.openlinksw.com Personal Weblog: http://www.openlinksw.com/blog/~kidehen Twitter/Identi.ca handle: @kidehen Google+ Profile: https://plus.google.com/112399767740508618350/about LinkedIn Profile: http://www.linkedin.com/in/kidehen

Gerard Meijssen

10:44 a.m.

New subject: [Wikidata-l] Cooperation between Wikidata and DBpedia and Wikipedia

Hoi, As you may have noticed there are other people on the CC.. The "not invented here" refers to much of the Wiki world.. things get done but much is done over and over again EVEN when cooperation is easy and obvious. I have largely resigned myself to it, it is probably part of our persona. However, it is not in the best interest of our users. The people I originally wrote to are more DBpedia ... and very much Wikimedia as well. I do not remember that DBpedia has a live component, love to learn more.. I do know how much the DBpedia people want to reach out and connect in any positive way with both Wikipedia and Wikidata. It is something that would work wonders because it can speed up the addition of new properties, the import of wholesale data AND it may have us add the processes they have build to curate the data. <grin> and additional point would be that we include some academia at the same time </grin> Thanks, GerardM On 26 August 2013 15:27, Kingsley Idehen <kidehen(a)openlinksw.com> wrote:

...

On 8/26/13 4:34 AM, Gerard Meijssen wrote: Hoi, Two things to consider; the Wikipedia community and the Wikidata community are two separate entities. As far as I am concerned, Wikidata needs more data to get to the tipping point where it becomes useful to users. I am really vocal about both. The license has traditionally been a sticking point and the "not invented here" aspect of DBpedia is as well. What is this "not invented here" aspect of DBpedia? You are effectively outside the Wikipedia community.. The other part is that the data of Wikidata is continually updated while DBpedia is not. What do you mean by that? Are you referring to schema/ontology/vocabulary evolution or instance data evolution? Remember, there are live editions of DBpedia. So in my opinion the best thing that can happen is when DBpedia DOES update with Wikidata. The point is not to absorb it without thinking. Yes, the first point of clarity would be when Wikidata produces a dump that can be ingested by DBpedia and any other data space in the LOD cloud. All that's required is data publication in Linked Data form. What DBpedia has is a set of properties that work on the data that it has gleaned from the many Wikipedias. There is undoubtedly a lot of documentation on it and, it would be good when this is taken into consideration when accepting and proposing new properties for Wikidata. Obviously only the properties that are currently supported can be proposed at this time. I am quite willing to propose properties based on DBpedia (but so can you). With the properties in place, we can import data. We can import it from both DBpedia and from Wikipedia. Sanity checks are needed for both sources and as far as I am aware we do not have sanity checks at Wikidata. What we do have is the possibility to compare data with other sources and that is where the Wikidata community needs to grow and that is why we need much data in the first place in order to get into this. However, we can start building the tools to do this. I hope the DBpedia community can help us with that. Thanks, GerardM As far as I know, DBpedia has always been interested in collaboration with Wikidata. Kingsley On 26 August 2013 09:16, Dimitris Kontokostas < kontokostas(a)informatik.uni-leipzig.de> wrote:

Hi, I will answer with questions with more questions... On Fri, Aug 23, 2013 at 10:58 AM, Gerard Meijssen < gerard.meijssen(a)gmail.com> wrote:

Hoi, The questions are: - would we advance a lot when we adopt the DBpedia schema as it is? Which schema? All of them? Some? Article classification? Infobox

extraction? Wikidata is going to be linked to the infoboxes in Wikipedia, so the priority is to support those needs, not to replicate any schema.

- Would we be open to include substantially more data? Which data? All of it? What is the reliability?

- When we adopt the schema, can we tinker with it to suit our needs? Again, could you please give some example of what to import and how

should it be adapted?

If the answers to these questions are yes, what is the point in procrastinating???

Do we have already all the datatypes that would be needed? Most of the properties that are missing is because of the lack of "value" or others.

_______________________________________________ Wikidata-l mailing listWikidata-l@lists.wikimedia.orghttps://lists.wikimedia.org/mailman/listinfo/wikidata-l -- Regards, Kingsley Idehen Founder & CEO OpenLink Software Company Web: http://www.openlinksw.com Personal Weblog: http://www.openlinksw.com/blog/~kidehen Twitter/Identi.ca handle: @kidehen Google+ Profile: https://plus.google.com/112399767740508618350/about LinkedIn Profile: http://www.linkedin.com/in/kidehen _______________________________________________ Wikidata-l mailing list Wikidata-l(a)lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikidata-l

Kingsley Idehen

11:36 a.m.

New subject: [Wikidata-l] Cooperation between Wikidata and DBpedia and Wikipedia

On 8/26/13 10:44 AM, Gerard Meijssen wrote:

...

I do know how much the DBpedia people want to reach out and connect in any positive way with both Wikipedia and Wikidata.

I have no knowledge (on the DBpedia side) of any resistance to collaborate with Wikidata. We've always seen this effort (like other structured data efforts of this kind e.g., Freebase, YAGO etc..) as being mutually beneficial. -- Regards, Kingsley Idehen Founder & CEO OpenLink Software Company Web: http://www.openlinksw.com Personal Weblog: http://www.openlinksw.com/blog/~kidehen Twitter/Identi.ca handle: @kidehen Google+ Profile: https://plus.google.com/112399767740508618350/about LinkedIn Profile: http://www.linkedin.com/in/kidehen

Gerard Meijssen

25 Sep 25 Sep

4:26 a.m.

New subject: [Wikidata-l] Cooperation between Wikidata and DBpedia and Wikipedia

Hoi, Congratulations on the new version of DBpedia :) .. This makes it an auspicious occasion to talk about future collaboration. At this time several Wikimedians are busy harvesting data from Wikipedia and loading it into Wikidata. You are harvesting data from Wikipedia and loading it into DBpedia. As we are including data into Wikidata, it goes into DBpedia as well... We might as well work together on this. One of the best parts (as far as I am concerned) is your knowledge of fields used in infoboxes and knowing how they are the same / related to fields in other infoboxes. This expertise should be easy to absorb into Wikidata In Wikidata we use qualifiers, will you adopt qualifiers in DBpedia? At this time qualifiers are not handled by the harvesting software I am familiar with. To me these are the two issues that determine how easy it will be to effectively collaborate on great content for Wikidata. To me the most important aspect of Wikidata is that it is actually used. Information added becomes available in many places. Consequently more data for Wikidata, data that fits in well and is closely related to the 100+ wikipedias you are harvesting will ensure a rich experience in so many places. Thanks, GerardM On 26 August 2013 17:36, Kingsley Idehen <kidehen(a)openlinksw.com> wrote:

...

On 8/26/13 10:44 AM, Gerard Meijssen wrote:

I do know how much the DBpedia people want to reach out and connect in any positive way with both Wikipedia and Wikidata.

I have no knowledge (on the DBpedia side) of any resistance to collaborate with Wikidata. We've always seen this effort (like other structured data efforts of this kind e.g., Freebase, YAGO etc..) as being mutually beneficial. -- Regards, Kingsley Idehen Founder & CEO OpenLink Software Company Web: http://www.openlinksw.com Personal Weblog: http://www.openlinksw.com/**blog/~kidehen<http://www.openlinksw.com/blog… Twitter/Identi.ca handle: @kidehen Google+ Profile: https://plus.google.com/**112399767740508618350/about<https://plus.googl… LinkedIn Profile: http://www.linkedin.com/in/**kidehen<http://www.linkedin.com/in/kidehen&…

3860

days inactive

3893

days old

wikidata@lists.wikimedia.org

Manage subscription

11 comments

6 participants

tags (0)

participants (6)

David Cuenca
Dimitris Kontokostas
Gerard Meijssen
Kingsley Idehen
Luca Martinelli
Tom Morris