There are times when a few keywords are often interchanged with each other in different languages and dialects.
It seems advantageous to somehow tell Wikidata Search that when someone types Harvard College to interchange and also look for Harvard University, and vice versa.
An interchange mapping table might suffice just for this use case, or something else, dunno...
How far in the future is this feature ? Roadblocks ?
Thad +ThadGuidry https://www.google.com/+ThadGuidry
College is not in general synonymous with university. This would be something that would have to be resolved on a case-by-case basis, although an alias should also suffice ("Harvard College" in this case).
On Sun, Jun 14, 2015 at 6:54 PM, Thad Guidry thadguidry@gmail.com wrote:
There are times when a few keywords are often interchanged with each other in different languages and dialects.
It seems advantageous to somehow tell Wikidata Search that when someone types Harvard College to interchange and also look for Harvard University, and vice versa.
An interchange mapping table might suffice just for this use case, or something else, dunno...
How far in the future is this feature ? Roadblocks ?
Thad +ThadGuidry https://www.google.com/+ThadGuidry
Wikidata mailing list Wikidata@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikidata
Per Jasper basically. In Canada, for example, "college" only refers to technical institutions or others which do not grant degrees.
Adrian Raddatz
On Sun, Jun 14, 2015 at 10:48 PM, Jasper Deng jasper@jasperswebsite.com wrote:
College is not in general synonymous with university. This would be something that would have to be resolved on a case-by-case basis, although an alias should also suffice ("Harvard College" in this case).
On Sun, Jun 14, 2015 at 6:54 PM, Thad Guidry thadguidry@gmail.com wrote:
There are times when a few keywords are often interchanged with each other in different languages and dialects.
It seems advantageous to somehow tell Wikidata Search that when someone types Harvard College to interchange and also look for Harvard University, and vice versa.
An interchange mapping table might suffice just for this use case, or something else, dunno...
How far in the future is this feature ? Roadblocks ?
Thad +ThadGuidry https://www.google.com/+ThadGuidry
Wikidata mailing list Wikidata@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikidata
Wikidata mailing list Wikidata@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikidata
On 15 June 2015 at 02:54, Thad Guidry thadguidry@gmail.com wrote:
It seems advantageous to somehow tell Wikidata Search that when someone types Harvard College to interchange and also look for Harvard University, and vice versa.
This is what the "alias" parameter is for.
Andy,
I know we have an alias parameter...but...
Do you want to set that alias on 24,000 Universities ? I don't. But perhaps a simple backend script could do it...sure.
My point is to all, that having a wiser Wikidata Search seems like a logical approach, and it doesn't change or skew the intent or meaning of the data as the rest of you have raised that concern. Its just a smarter Search, that is more helpful to folks finding entities and properties.
Thad +ThadGuidry https://www.google.com/+ThadGuidry
On Mon, Jun 15, 2015 at 6:14 AM, Andy Mabbett andy@pigsonthewing.org.uk wrote:
On 15 June 2015 at 02:54, Thad Guidry thadguidry@gmail.com wrote:
It seems advantageous to somehow tell Wikidata Search that when someone types Harvard College to interchange and also look for Harvard
University,
and vice versa.
This is what the "alias" parameter is for.
-- Andy Mabbett @pigsonthewing http://pigsonthewing.org.uk
Wikidata mailing list Wikidata@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikidata
The search is a kind of stupid dialogue system, and it only has a user model that is sensitive for language. A better dialogue system with an individual user model could use location as a hint for context. There are several heavy books about that topic!
If I search for "Oslo" and live in Norway it is highly likely that I want the article about the city in Norway. If I live in Marshall County, Minnesota, it is not so obvious that I want the city in Norway to be ranked first. But what if I live in Norway and have just searched for Marshall County? It is not easy to get these things right, and it is a lot more difficult than just adding some aliases. The aliases can solve the alternate label problem, but it can not solve the user context problem.
On Mon, Jun 15, 2015 at 4:41 PM, Thad Guidry thadguidry@gmail.com wrote:
Andy,
I know we have an alias parameter...but...
Do you want to set that alias on 24,000 Universities ? I don't. But perhaps a simple backend script could do it...sure.
My point is to all, that having a wiser Wikidata Search seems like a logical approach, and it doesn't change or skew the intent or meaning of the data as the rest of you have raised that concern. Its just a smarter Search, that is more helpful to folks finding entities and properties.
Thad +ThadGuidry
On Mon, Jun 15, 2015 at 6:14 AM, Andy Mabbett andy@pigsonthewing.org.uk wrote:
On 15 June 2015 at 02:54, Thad Guidry thadguidry@gmail.com wrote:
It seems advantageous to somehow tell Wikidata Search that when someone types Harvard College to interchange and also look for Harvard University, and vice versa.
This is what the "alias" parameter is for.
-- Andy Mabbett @pigsonthewing http://pigsonthewing.org.uk
Wikidata mailing list Wikidata@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikidata
Wikidata mailing list Wikidata@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikidata
Perhaps someone does not know, .. It is a city called Oslo in Marshall County, Minnesota. ;)
On Mon, Jun 15, 2015 at 5:00 PM, John Erling Blad jeblad@gmail.com wrote:
The search is a kind of stupid dialogue system, and it only has a user model that is sensitive for language. A better dialogue system with an individual user model could use location as a hint for context. There are several heavy books about that topic!
If I search for "Oslo" and live in Norway it is highly likely that I want the article about the city in Norway. If I live in Marshall County, Minnesota, it is not so obvious that I want the city in Norway to be ranked first. But what if I live in Norway and have just searched for Marshall County? It is not easy to get these things right, and it is a lot more difficult than just adding some aliases. The aliases can solve the alternate label problem, but it can not solve the user context problem.
On Mon, Jun 15, 2015 at 4:41 PM, Thad Guidry thadguidry@gmail.com wrote:
Andy,
I know we have an alias parameter...but...
Do you want to set that alias on 24,000 Universities ? I don't. But perhaps a simple backend script could do it...sure.
My point is to all, that having a wiser Wikidata Search seems like a logical approach, and it doesn't change or skew the intent or meaning of the data as the rest of you have raised that concern. Its just a smarter Search, that is more helpful to folks finding entities and properties.
Thad +ThadGuidry
On Mon, Jun 15, 2015 at 6:14 AM, Andy Mabbett andy@pigsonthewing.org.uk wrote:
On 15 June 2015 at 02:54, Thad Guidry thadguidry@gmail.com wrote:
It seems advantageous to somehow tell Wikidata Search that when someone types Harvard College to interchange and also look for Harvard University, and vice versa.
This is what the "alias" parameter is for.
-- Andy Mabbett @pigsonthewing http://pigsonthewing.org.uk
Wikidata mailing list Wikidata@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikidata
Wikidata mailing list Wikidata@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikidata
I don't want to solve problems that you describe John.
I just want to the Wikidata Search to be a bit smarter in regards to common labels are the sometimes interchanged in various languages. It's something we solved in Freebase and can be done. (We generated special Lucene indexes, back in the day).
Thad +ThadGuidry https://www.google.com/+ThadGuidry
On Mon, Jun 15, 2015 at 10:03 AM, John Erling Blad jeblad@gmail.com wrote:
Perhaps someone does not know, .. It is a city called Oslo in Marshall County, Minnesota. ;)
On Mon, Jun 15, 2015 at 5:00 PM, John Erling Blad jeblad@gmail.com wrote:
The search is a kind of stupid dialogue system, and it only has a user model that is sensitive for language. A better dialogue system with an individual user model could use location as a hint for context. There are several heavy books about that topic!
If I search for "Oslo" and live in Norway it is highly likely that I want the article about the city in Norway. If I live in Marshall County, Minnesota, it is not so obvious that I want the city in Norway to be ranked first. But what if I live in Norway and have just searched for Marshall County? It is not easy to get these things right, and it is a lot more difficult than just adding some aliases. The aliases can solve the alternate label problem, but it can not solve the user context problem.
On Mon, Jun 15, 2015 at 4:41 PM, Thad Guidry thadguidry@gmail.com
wrote:
Andy,
I know we have an alias parameter...but...
Do you want to set that alias on 24,000 Universities ? I don't. But perhaps a simple backend script could do it...sure.
My point is to all, that having a wiser Wikidata Search seems like a
logical
approach, and it doesn't change or skew the intent or meaning of the
data as
the rest of you have raised that concern. Its just a smarter Search,
that
is more helpful to folks finding entities and properties.
Thad +ThadGuidry
On Mon, Jun 15, 2015 at 6:14 AM, Andy Mabbett <
andy@pigsonthewing.org.uk>
wrote:
On 15 June 2015 at 02:54, Thad Guidry thadguidry@gmail.com wrote:
It seems advantageous to somehow tell Wikidata Search that when
someone
types Harvard College to interchange and also look for Harvard University, and vice versa.
This is what the "alias" parameter is for.
-- Andy Mabbett @pigsonthewing http://pigsonthewing.org.uk
Wikidata mailing list Wikidata@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikidata
Wikidata mailing list Wikidata@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikidata
Wikidata mailing list Wikidata@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikidata
John Erling Blad, 15/06/2015 17:00:
If I search for "Oslo" and live in Norway it is highly likely that I want the article about the city in Norway. If I live in Marshall County, Minnesota, it is not so obvious that I want the city in Norway to be ranked first.
If Chinese really create a city call "Parma" to sell more prosciutto, I want Chinese users to be given the real Parma first always. :)
Nemo
There are a lot of places called Parma, and it is not obvious which one should be listed first. Perhaps Parma in Tibet?
This is actually a problem that can't be easily solved. "Parma" is interpreted in a cultural context, and the Italian city is just one of several places called the same. It might be obvious for an Italian that "Parma" is an Italian city, but it is equally obvious for someone from Tibet? What about Parma, Ohio, more than 80.000 people live there?
The solution is to use a set of standardized user models, typically they will follow the language regions.
On Mon, Jun 15, 2015 at 5:45 PM, Federico Leva (Nemo) nemowiki@gmail.com wrote:
John Erling Blad, 15/06/2015 17:00:
If I search for "Oslo" and live in Norway it is highly likely that I want the article about the city in Norway. If I live in Marshall County, Minnesota, it is not so obvious that I want the city in Norway to be ranked first.
If Chinese really create a city call "Parma" to sell more prosciutto, I want Chinese users to be given the real Parma first always. :)
Nemo
Wikidata mailing list Wikidata@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikidata