Freebase: Discussion about lukeschubert

Military: confusion

2009-12-15T05:32:40.0006Z

Sorry for the slow reply. I was on the road.

Which type do you want to be promoted? This discussion is on my Military base in general.

140hours: Real time price dicovery for fine art

2009-12-15T04:42:53.0018Z

There are only a handful of websites that are looking into the problem of real price discovery for fine art. I can only think of two: www.140hours.com and my own site www.painttwits.com. Ironically, the huge number of art web sites hasn't changed this? Why is this?

jamespotter: Thanks and a tip

2009-12-15T01:20:28.0018Z

Hi,

I want to thank you for your contribution to Freebase and point out something tricky that's easy to miss: The Family type which you used to create the Potter Family topic is meant for real people (type /people/person) only, not fictional characters (type /fictional_universe/fictional_character). So as a result, when you asserted that Harry Potter is a member of the Potter Family, you also inadvertently added the type /people/person to the fictional character. I'm going to roll it back as I believe this is unintended. I understand that type names are not always clear -- I recommend taking a few seconds to look at any type description if there is any or feel free to raise any questions. Cheers.

Video Game: Games "based on" existing works (e.g. films)

2009-12-15T00:45:51.0005Z

Hi,

How should I use to link a video game that either exists within the realm of or extends that of an existing work, such as a film? For example, I would like to say that the Lego Star Wars game series is derived from the Star Wars films, and is a part of the Star Wars franchise. There are several Franchise types but none pertains to video games. There's also Consumer Product with property Product Line that can probably be used to link books and films and games and toys of the same theme together.

Another possibility is to use Fictional Universe to indicate that a game is defined within a fictional universe. However, some fandoms are quite sensitive about Canon vs. Expanded Universe. Video games can be problematic when it comes to continuity.

Suggestions?

Video Game: Games "based on" existing works (e.g. films)

2009-12-15T00:45:51.0005Z

Hi,

Suggestions?

General Support: how do i get do not have permission to add a return property on Composer.

2009-12-14T23:21:23.0018Z

Will using 'composer' in my catalog base pick up the other info about that composer from the other base(s)?

I don't really understand how to link my info. with the data already in Freebase! and what the path is from a stand-alone database to something integrated with everything else in Freebase.

I'd be grateful for any pointers.

Simon

General Support: how do i get do not have permission to add a return property on Composer.

2009-12-14T23:17:00.0005Z

Thanks for the reply - the question was barely coherent! ('cut and paste fever').

'Amending a commons schema' sounds like a big deal!

All I want to do is get my base linking to existing composer info. and showing up when people look at a composer. Am I going about this the right way?

I'm not insistent on keeping ownership of the data and would like to make it available for anyone to use/add to.

Simon

Degree: A.M. and other names for a Master of Arts...

2009-12-14T22:30:23.0013Z

I agree that the reference to the degree should be as it was awarded, but unfortunately there's no way to do that currently with Freebase. The link is symbolic and the name at the end of the link is used universally for all occurrences so you can't have Artium Magister one place and Master of Arts another. The are occasional places in the schema where the actual value is preserved (Book Edition "credited as" for authors comes to mind), but when a literal string is used like that you lose the power of the symbolic connection. There isn't yet a good way to have your cake and eat it too.

Awards: Submissions

2009-12-14T21:31:29.0013Z

I've created a property and types for long-lists now. I think it should work pretty well. Let me know if anything's missing or needs to be clearer.

Degree: A.M. and other names for a Master of Arts...

2009-12-14T21:24:29.0000Z

The standard practice in Freebase is not to create multiple topics for things that are identical except in name, so the practice with degrees has always been to use the same degree topic, no matter how called by the granting institution at the time. Thus degrees of "Artium Magister", "Master of Arts", and "M.A." are all represented by http://www.freebase.com/view/en/master_of_arts. Or, in other words, the name isn't as important as what it represents. Hence, there is also some variability on how the degree topics themselves are named, based on whatever the user who created (or renamed) the topic entered. (E.g, Bachelor of Arts vs M.A.) I'd personally support expanding the names, since it's less ambiguous that way, but as long as aliases are entered for variants, they'll still show up for autocompletion and reconciliation, which is the most important thing.

Degree: A.M. and other names for a Master of Arts...

2009-12-14T21:01:23.0013Z

Some institutions use, or have used "the Latin degree names, such as Artium Magister (AM) or Scientiæ Magister (SM)." Though the wikipedia gives examples of institutions that currently use these names, "Harvard University, the University of Chicago, MIT and the University of Pennsylvania use the degree abbreviations A.M. and S.M. for some of their master's degrees." it does not say what institutions have used them in the past.

I suspect that the use of such names was much more popular in the past and many bio's list degrees as such.

See, for example: http://www.freebase.com/edit/topic/en/john_erskine_1879

My question--yes, there is a question--is: how should degree information be entered?

As awarded, or as known today? I think it should be as awarded, but I'd like to see a mapping to the current nomenclature, which I suspect exists, but I can't see in action.

Additionally, I'd like to see more discussion of this in the help text, right now it just says to enter info about the degree: http://rookery5.aviary.com/storagev12/2678000/2678399_95c4_625x625.jpg

It should give instructions as to how to enter the degree: In the original form. In fully qualified format and not an abbreviation. ...and other stuff

Dublin Mid West: Politicial district not Governmental Jurisdiction?

2009-12-14T19:29:47.0013Z

Yes, that's correct -- districts aren't usually going to be jurisdictions, and definitely not in this case. Unfortunately, we can't set up an automatic rule to detect this sort of thing because some "districts" are also jurisdictions -- US states function as the "district" for US senators, for example.

General Support: Is it appropriate to edit out data integrity inconsistencies?

2009-12-14T19:03:36.0013Z

OK, Thanks. I've tweaked those two named examples. Hopefully I won't hit too many editorial decisions. I'd already removed a loop which had http://www.freebase.com/view/en/brown-eyed_soul as both a sub and super genre of http://www.freebase.com/view/en/rhythm_and_blues because that made no sense, but thought I'd check this sort of repair was OK and would have a constructive effect. There is another few in the import I've tried so far, I'll modify my system to make it easier to track and fix...

General Support: Is it appropriate to edit out data integrity inconsistencies?

2009-12-14T18:22:19.0013Z

A lot of stuff like this are just simple mistakes, so yes, cleaning it up would be a help. Cycles in this type of schema are clearly a logical inconsistency that should be fixed. For cycles of length 1, it's pretty easy to tell where to break the cycle. For longer cycles, if you can tell where to break them, go ahead and do it. If there's a question, or you need help, post a discussion note to the relevant topics and someone more familiar with the topic might be able to clarify where the cycle should be broken.

General Support: Is it appropriate to edit out data integrity inconsistencies?

2009-12-14T18:13:08.0000Z

I'm importing a set of statements about Music Genres as an exercise. I get some retrieval problems because of (among others) http://www.freebase.com/view/en/gothic_metal

This entry has been set as both a super-genre and a sub-genre of itself. Without trying to be a fact-checker, I'm pretty sure that is not actually possible under the concept of sub-genre. So I think I should remove that relationship by hand.

There are a few other, longer loops also. http://www.freebase.com/view/en/garage_rock is apparently a super- and sub- genre of http://www.freebase.com/view/en/psychobilly I'm not an expert in the field, but there is something intrinsically inconsistent with this.

I realize that this sort of situation will arise when combining disparate data sources, so it's basically inevitable. I guess also that there is no reasoning or constraints being applied to prevent it.

Therefore ... would any attempt at gardening these paradoxes out of the system be entirely futile?

My consuming app is going to have to sanitize this data itself, so that's fine. My question is should I use the warnings that I'm discovering to come back here and clean things up for everyone?

General Support: Incorrect ID

2009-12-14T15:11:19.0013Z

Thanks for the clear explanation!

I understand that each URI is really just a GUID of sorts, and the data works fine when requested by that name. It just looked erroneous to be using that one as the primary ID when I queried for http://rdf.freebase.com/rdf/music.genre . It was an odd one out. I was a bit concerned that something with a UserID in the path may be sandbox data rather than canonic. Is there a way to promote http://www.freebase.com/view/en/country_music to being the 'preferred' or primary ID? ... Or not ... I see that is actually meaning a specific album name :-( . Dang.

FYI, the issue is: I'm testing a facility to import and reference shared vocabularies, and was planning to use the supplied URI/GUIDs as a primary key in an internal RDF triplestore. (Maybe I shouldn't do that, and just internalize the sameAs relations)

The given URI will still work (thanks for clearing that up) but just looked out of place in the diagnostic logs I was seeing. Partially (my fault) because my code is using the key found at the end of an URL as a temporary name for placeholder topics that haven't been individually fully retrieved yet. This is an ugly short-cut, but was returning sane results most of the time. My data queue listed : [ ... pop_music, trance_music, 00b3-10cd, breakbeat, black_metal ... ] ... I'll eventually be able to resolve that as I get the preferred name I guess. Just something to know! Thanks again. I think I'll be able to work with this..

General Support: Incorrect ID

2009-12-14T14:08:44.0013Z

This is entirely expected. Each entity in Freebase can be referred to either by its GUID, or by an arbitrary number (possibly zero) of other IDs; there's then an algorithm which chooses the "primary" ID for each topic. Put simply, this algorithm chooses the shortest ID from the ones available, or uses the GUID if none exist.

In this case, the topic had not been assigned another ID (note that both /en/country and /en/country_music are used elsewhere), so when it was given an ID in Arthur's personal namespace, it started using that as its primary ID. However, nothing has changed here: you can still refer to this topic by its GUID or even by a number of Wikipedia keys, or even by its Wikipedia numeric article ID if you want. (The /wikipedia keys are never allowed to become the primary ID for a topic, which is why you weren't seeing those before). In the same way, you'll see raw GUIDs for any topics which don't have any other ID set.

Is this change of primary ID causing you a problem?

General Support: Incorrect ID

2009-12-14T13:07:16.0014Z

Something has happened to this topic (Country Music). When the ID /user/avh/ellerdale.00b3-10cd was added to it, it seems to have taken over the identity of this resource. Now listings that include the concept of 'Country Music" Now instead include the concept of this odd User-defined Id

eg http://rdf.freebase.com/rdf/music.genre http://rdf.freebase.com/ns/type.type.instance [ http://rdf.freebase.com/ns/en.pop_music. http://rdf.freebase.com/ns/en.trance_music. http://rdf.freebase.com/ns/user.avh.ellerdale.00b3-10cd. http://rdf.freebase.com/ns/en.breakbeat. http://rdf.freebase.com/ns/en.black_metal ]

I can see when the erroneous ID was added http://www.freebase.com/history/view/user/avh/ellerdale/00b3-10cd But I can't see how to undo it.

I'm spotting a few other oddities (raw GUDs) in instances under music.genre as well.

Dublin Mid West: Politicial district not Governmental Jurisdiction?

2009-12-14T10:19:16.0019Z

Shouldn't Dublin Mid West be just a political district rather than a governmental jurisdiction? For example, Joanna Tuffy should have "Dublin Mid West" as her political district if (as I assume) she was elected in this constituency; to have this topic as her jurisdiction implies that she can only pass laws applying to Dublin Mid West.

General Support: how do i get do not have permission to add a return property on Composer.

2009-12-14T06:48:25.0013Z

simonhill,

I think it's because you linked to the wrong Composer type from Thematic Catalog. You linked it to the Composer type in the Music commons (which as masouras says can't be easily changed). I see that you have created a Composer type in the Catalog base which you probably meant to link to. You should change where the property links to, and ensure you click the Composer in your Catalog base.

General Support: how do i get do not have permission to add a return property on Composer.

2009-12-14T04:58:20.0013Z

You have to get someone who is an admin of the Music domain (/music) to add it.

However, we don't usually link to base or user properties unless they move into the commons schema. If you want to propose amendments to any commons schema, you can always ask in the data-modeling@freebase.com mailing list.

Subscribe here: http://www.freebase.com/build

General Support: how do i get do not have permission to add a return property on Composer.

2009-12-14T01:00:10.0000Z

How can i get iy shiow up when looking at composer topic?

General Support: how do i get do not have permission to add a return property on Composer.

2009-12-14T00:59:28.0020Z

I've made a composer's thematic catalog database .

Military: confusion

2009-12-09T19:57:30.0013Z

ya, or atleast beside them in suggest. i really like this schema. /military/military posts are vague. but i did, yesterday, spend a fair amount of time writing data to the wrong type. maybe we can push it for promotion?

Military: confusion

2009-12-09T05:41:25.0013Z

How will it steal writes? Are types in this base appearing higher than types in the military commons?

Military: confusion

2009-12-09T05:36:27.0013Z

i'm worried this type will steal writes intended for the military commons, rename it?

Particle spin: Particle type?

2009-12-07T22:12:10.0022Z

Phil: Ahh, I see. There is no way to make a model like that which will work in the client. MQL doesn't care about nested CVTs, but the client balks at them.

Particle spin: Particle type?

2009-12-06T18:40:32.0008Z

Jeff: composition is already non-unique. Ideally, what I'd like to end up with is something like the structure pictured here, but none of the "xxx" nodes make sense to be topics themselves.

Particle spin: Particle type?

2009-12-05T11:00:08.0020Z

I've deleted particle statistical family and group - the particle type is looking nice and clean now.

For π⁰, I guess we could have a formula (i.e. text) similar to a chemical compound formula, but then we lose the linkage to the quarks it makes up. Otherwise I think we'll have to leave composition of π⁰ etc and self antiparticles as open questions for now.

Another question: why doesn't the Browse view for an Electron display its particle properties?

Particle spin: Particle type?

2009-12-04T18:37:21.0013Z

Composition of π⁰: Could composition be non-unique, and defined as "possible compositions"?

Particle spin: Particle type?

2009-12-04T12:05:57.0011Z

I've added particle family to the particle type so I think we can now drop particle statistical family and group. I've also engaged in some property reciprocation.

Self antiparticles: ugh. The double links solution is pretty horrible. The only other solution I can think of would be to have a "self antiparticle" boolean, but that's probably even worse.

Composition of π⁰: in slight handwaving terms, the π⁰ can be thought of as alternating between being composed of (an up quark and an up antiquark) and (a down quark and a down antiquark), but never (an up quark and a down antiquark) or any of the other combinations. The only ways I can see to model this either end up with CVTs inside CVTs or creating a topic for something like "composition of the π⁰".

Spin: yeah, I screwed up the pion spins. Good catch. Now fixed, I think.

Particle spin: Particle type?

2009-12-04T00:49:03.0013Z

I like the hadron type!

I've changed spin to float and updated the spins that were already entered in (note: Wikipedia says pions have zero spin? also, should spins be negative?). I'll start moving types from my base to the new base soon.

Particle spin: Particle type?

2009-12-03T18:48:31.0031Z

Self antiparticles: There isn't a way to set a particle to be its own antiparticle using a sibling-relationship CVT -- the CVT has only one property to hold both particles, and you cannot enter the same topic as a value for a property more than one time. The only work-around would be to use something that resembled a phylogeny pattern, with two linked properties "antiparticle" and "antiparticle of". This is kind of ugly, but would handle the π⁰ case. The problem with this pattern is that you have to enter all the relationships in twice for all particles that aren't antiparticles of themselves -- once in the "antiparticle" property, and once in "antiparticle of".

Composition of π⁰: I'm not sure what "linear combination of (up-antiup) and (down-antidown)" means, but I can see in Wikipedia's list of mesons that there are several mesons with similarly complex compositions.

Uncertainties: we can do this -- see how we handled it for atomic masses.

Spin: Agreed -- let's make it just be a float for the particles. We can refactor the Particle Spin type separately.

Particle spin: Particle type?

2009-12-03T13:07:45.0013Z

More from me... it doesn't at the moment to seem possible to set a particle to be its own antiparticle (try it with the π⁰). Any thoughts on a way round this one?

Also, we need to find a way to model the composition of the π⁰, which is a linear combination of (up-antiup) and (down-antidown).

I'll shut up for a bit now and let other people get a word in edgeways :-)

Particle spin: Particle type?

2009-12-03T10:14:50.0012Z

Two more thoughts from the way to work: do we want to include uncertainties on properties like mass, etc? Also, I created a hadron type to model the composition of baryons and mesons.

Particle spin: Particle type?

2009-12-03T08:38:34.0006Z

Yes, it would allow people to enter the class as "fermion". If we really wanted to avoid that possibility, we could have "particle family" type (quark, gauge boson, meson, etc) which link to a "particle class" type (fermion, boson, hadron), but that seems to me to be to getting towards creating types for the sake of it.

Spin: just a floating point number, yes. The particle spin used in chemistry is the same concept, but it's typically expressed in a different way.

In other news, I've created the particle [physics bas](http://particlephysics.freebase.com/)e and added the particle family type. Luke, you're an admin so you can move your types over there if you want.

lukeschubert: Australian Parliament - historical

2009-12-03T01:57:17.0013Z

Link has moved here.

Chemistry: Why so many 0+ topics?

2009-12-01T19:43:41.0052Z

Argh. It looks like the loading script created new spin instances for each isotope. I'll create a bug, and see if this can be cleaned up without clogging the merge queue. DA-998

Chemistry: Why so many 0+ topics?

2009-12-01T10:53:12.0000Z

I think so and similarly for the other particle spins. Cross-posting to the bot's accounts and chemistry Commons for any admin comments.

Chemistry: Why so many 0+ topics?

2009-12-01T04:47:12.0013Z

Why are there so many "0+" topics of type Particle spin? Shouldn't they be merged?