Discussions on Domains and Types

I added a Card Game type in my types... it it right to put it there?
It's supposed to be used for card games like poker and collectible card games...

Creating types in your personal namespace is a great place to start. If you'd like the type to be available for others to use you can check the box to 'publish this type' in type editor -- and it can also be submitted for consideration as a public type if it starts to be used widely.

When viewing a company... in this case Vodafone... the headquarters address type does not contain a country field. Could you please add this. Many thx. Iain

I'm quite new to freebase - trying to figure how to create the Mobile Country Code (MCC) & Mobile Network Code types for Mobile Network Operators... any help appreciated!

Totally new and somewhat bewildered - noticed that the Music domain does not contain 'music scores'. I want to eventually upload a bunch of info / links related to various music scores, specifically choral music scores held by our library. At this point, not at all sure how to accomplish this but a good start would be finding the correct topic. Many thanks for help in adding: music scores, choral music scores, musical scores or let me know where these are.

Infobuddha—Thanks for your comment. The music domain is still growing and evolving, and feedback is always helpful. In what sense is a score, for your purposes, different from compositions and arrangements? Do you specially need to describe attributes of the physical objects, or the publications of multiple (nearly-)identical physical objects, or the music described by the objects?

A good starting point would be to list the properties that you want a score to have, and a good way to do that is to prototype a type in your own domain. Let us know if you have any questions about that; the music domain would be a good place to continue the discussion.

Welcome!

infobuddha, It seems like the "Arrangement" type would work. I think the term "score" and "arrangement" could be used inter-changeably. Currently, "Arrangement" has only one property, "Arrangement of" which points to the composition (type "Composition"). I do think "Arrangement" could use some other properties: one comes to mind: "Arranged by", expected (new) type "Arranger" There are different kinds of arrangements. Rather than have that be a property, perhaps having additional types that impart the relevant properties of arrangements would make more sense: "Choral Arrangement" "Big Band Arrangement" "Orchestral Arrangement" As always, there are many ways to slice this pie. I'd love to hear your thoughts on this...

iain_herd, the country for headquarters is derived from the postal code and the city and doesn't need to be input manually from the address field. This isn't clear in the user interface though - we hope to fix this in an upcoming release.

Hi there - I'm missing a lot of abstract types. Such types that may be be used to classify a knowledge domain from a more abstract perspective. Types like: Idea, Process, Task, etc. and relationships like: Assigned_To, Interested_In, etc. In this way Freebase could also be used as a kind of project management tool. So firstly how about a project management domain? Later on this could be expanded to a more general process management domain (business, knowledge, whatever needs to be managed).

You're always welcome to create types of your own and if you think they'd be generally useful and suggest that they be promoted for public use.

We're still defining guidelines for what kinds of content will be most useful to store on Freebase - you should bear in mind that any data input will be visible to everyone, which could be a non-starter for certain kinds of project management activities. This isn't to say Freebase couldn't be used in that way, but there are definitely many other platforms that are designed more specifically for that kind of purpose.

Thanks.

Yeah and that was my next question. Does Freebase consider to make it possible to use Freebase for closed circuits with restricted access. This would be a key feature for commercial application of Freebase. Companies would the be able to control collaboration. I know it's kind of contradictory to the Freebase name but it's a shame to force commercial ontology development to proprietary systems (and a waste of ressources).

Concerning project management. Yes there are a lot of systems that handle this like there's a lot of systems handling CRM, finance, HRM, CMS, whatever. But all these systems could (and should) be handled within an ontology because they are all just specialized instances of a more general database.

We have considered the idea of restricted access content areas - and might provide that kind of functionality in the future, perhaps for a fee, perhaps on a related site, but that isn't our focus right now.

Feel free to try modeling a project management schema if you're interested. Might be instructive to see how that works.

I'm interested in a domain and set of types that focus specifically on the future. This goes beyond simply having dates that are at some point in the future. There is a broad range of types (trends, events, forecast, scenarios, etc.) that I need to work with as a professional futurist. About 6-7 years ago I identified a series of these types and their interrelationships as part of an effort to create an XML-based Futures Studies Markup Language. From the quick look that I've taken at Freebase, it appears that some of that old work may finally prove to be useful.

The challenge/opportunity with future-oriented data domains and types is that it can make use of much of the work done on basic types (people, locations, etc.). There is a significant amount of overlap in some areas with a history domain and associated types.

Hey, pvonstackleberg.

Types about the future sound interesting. If you want to take a shot at modeling some and you hit any snags, drop a line in the discussions to get some help.

Best,
Sarah

Sarah

Thanks for the offer of help via the discussions. I've started experimenting with a few futures-related types (trend, forecast, event, etc.) and am working my way through documentation on Freebase to figure out what I'm doing. At some point I'd like to get feedback from some of you more experienced folks to ensure that I'm on the right track with what I'm doing.

Peter

Terrific! When you're ready for feedback, post a note with any questions, and we'll take a look.

Hi Sarah:

I've been mulling over the Trend, Event, and Forecast types I've created. I have a couple of questions and am wondering if you and some other folks can take a quick look at what I've got done.

1. Trends and Events are linked. A Trend can contain several Events, set up as either a Supporting Event or Countervailing Event. I want to make sure that users can move back and forth easily between the events and the trends.

2. Trends can be linked to other trends (Supporting Trend and Countervailing Trend). Again, I want users to be able to move back and forth easily so they can look at all trends that associated.

3. Events can contain other Events. For example, the Event "World War 2" has several other events (for example, individual battles) that are included within it. It is possible that these events will have further events embedded within them. Movement back and forth from these events is important.

Have I set up the properties, etc. correctly for these different types in order to have a structure that is easy to navigate up and down in?

Peter

Hi Peter. On a Person page such as Danny Hillis, under Education the entries are each events, similar to you Event type. For example, the event where Danny earns a PhD. This is very similar to your Event type. Have you thought about general support for events in Freebase?

Sorry, I screwed up the link. (Freebase does a pretty good job of trying to hid the raw events.) Here is the event for Danny earns a PhD.

I do think we need broad support on Freebase for an event type.

First, let me make a distinction between what I mean by "Event" and the thread a couple items down on organized events (for example, a rock concert, a party, etc.)

The kind of event I need for futures research purposes is much more akin to the "Historical Events" type that I think you refer to. I added my own "Event" type simply because the "Historical Events" type was not complete enough for my particular needs.

However, I think current and historical events will be a major element of any data collection. As such, it would make a lot of sense to have broad, general support for an "Event" type. Having a bunch of slightly different "Event" types would be a significant pain in the behind and ends up defeating the purpose of Freebase.

If you have any suggestions on how we can pull together a common "Event" type, I'm certainly willing to work with folks on it.

Peter

In response to Peter's questions to Sarah (above):

1. In the trend type, "countervailing event" and "supporting event" have expected types of topic. If you make the expected type event, you'll be able to link events and trends directly, if you also add a reverse property on event.

2. This looks good to me. I'd consider reordering the properties so the supporting/countervailing pair is nearer the supported/opposed pair, and possibly renaming "countervailing trend" to "opposing trend" to make the relationship between those two properties clearer, but these are just cosmetic suggestions.

3. The included/higher-level event properties also look good to me; you'll find exactly parallel structures throughout Freebase. If you look at the military domain, there's a parallel structure to this that's explicitly for military conflicts.

Thanks for your comments Jeff. I've gone in an attempted to implement all of them on the "Trend" and "Event" types. If you get a chance, I'd appreciate if someone might check what I've done actually accomplishes what was suggested. (I'm still having a little trouble with the concept and implementation of the reverse properties.)

Peter

Also, I have several questions about more detailed aspects of setting up types. Comments would be much appreciated.

1. I have a number of "Trend" properties (Trend Timing, Trend Scope, etc.) that have values set up as integers. These values are intended for use in calculating an overall Trend Impact value. Should I have a "Trend Impact" property that somehow calculates a value based on the other properties, or should I set up the "Trend Impact" using a calculation in a Javascript or other function when the data is displayed on a page. It is important to be able to search and sort trends based on their impact (for example, "Show me all trends with an impact greater than 20" or "Sort all the trends from highest to lowest impact." I'm assuming that this kind of search and sort points towards having a "Trend Impact" property. That then leads to the questions of how I would do such a calculation for that property.

2. Assume that I want to include the development of a new technology as part of an event (for example, the World War 2 event includes another subordinate event that is "The first operational use of a jet fighter." I want information on the Me-262 jet fighter developed and flown by Germany towards the end of WW2, and that information is available via the "Aircraft" type. When I link in that content, I get a name that says "Me-262". I need a name that is more appropriate for my Event type (i.e., "The first operational use of...") How can I use the Me-262 content from "Aircraft" and still get the name/title I need for my event without changing the name in the "Aircraft" type?

That looks good to me. Nicely done.

I've got another question for those who you have more expertize with Freebase than I do...

I've created the "Issue" type so emerging problems/challenges can be entered into the system. I have two properties -- Issue Opponent and Issue Proponent -- to identify those who say there is no issue and those who say there is one. The opponents and proponents can be either people or organizations. How do I set it up so that these properties can be either, instead of having a separate pair of properties for people and organizations.

The way to do this is to create a new type, something like (but with a better name than) "Issue Opinion-Haver" that would link to issue, and could be applied as a co-type to the person, organization, agency, racehorse, etc. that supports or opposes the issue. I made up "issue opinion-haver" because I think it makes more sense to have a single type for this with two properties (e.g., issues opposes; issues supports) rather than two types "Issue Opponent" and "Issue Proponent" with one property each.

Thanks for the pointers, Jeff. They were very helpful.

Do you have any suggestions for the questions I asked 4 items up?

1) I would think that something like "Trend Impact" should be calculated from the other properties when its displayed (I'm assuming that you are talking about displaying it in an external application of some kind, since Freebase doesn't have this capability.)

2) I think you might need a property on event along the lines of "topics related to this event" which should have a new type as the expected property (I can't think of any good type names for this off-hand -- naming often turns out the be the hardest part of schema design!). Since the event "first operational use of a jet aircraft" is not the same thing as the "Me-252" (the aircraft topic covers a lot of things about the aircraft model, from its design to its final disposition, while the event topic is about a specific instance of its use), I don't think using the topic of the aircraft for the event makes sense.

The ability to add domains is now available...cool! Now how do I copy all my types from my personal domain to another domain (Futures Studies & Forecasting) that I've created so that I don't have to re-do all of them.

Also, is it possible to have one variation of a type in the personal domain and a slightly different type in the other domain?

We don't currently have a feature for moving types between domains via the UI. If this is a big hardship for a specific case, let us know and we can probably assist.

Per your second question, there is nothing to prevent you from having even identical types within two different domains. Where conflicts happen is if two types have the same key in the same domain -- which, coming full circle here, is one reason why we don't yet have easy tools for moving types in the UI.

Would it be possible to get someone to move all of the types currently under my personal domain over to the "Futures Studies & Forecasting" domain, as doing it manually will involve a significant amount of work. I am also wondering if it would be possible to move the content associated with all of those types as well to avoid having to reenter all of it.

It looks like they've been moved. Apologies if you had to do it manually.

I would like to see a Commercial Real Estate domain/type on here. By the way this is a great site!

I'm getting ready to build a database of political contributions for a community: office, candidate, year, contributor, etc. This would be for multiple boards and the idea would be to make some connections about who's supporting whom.

Is freebase a suitable tool, and has anyone done anything like this yet? I can find anything.

Barry, have you looked at Robert's "Congress" domain (http://www.freebase.com/view/user/robert/us_congress)? Is that the sort of thing you're looking for? I believe he got his data from the Campaign for Responsible Politics (I think that's the acronym expansion for CRP) which tracks who donates how much to which politicians.

Nice and easy to locate types, looks great.

See also eventful / EVDB.

Yes, I wish that Eventful.com would put their/our events in Freebase. They/we do a good job of figuring out how events are done.

When you see something has been typed incorrectly, feel free to correct it! Simply hover your mouse over the incorrect type and then select 'remove' from the menu provided.

I´d like to have these domains, too. The types which you suggest look like good choices for getting started :-)
As for where to place these domains... there was a related discussion recently here .

> Is there such a thing as a meta-domain, i.e. Academic Disciplines?

Is Field of Study close to what you´re looking for?

Hi, users are encouraged to add data and schema to Freebase. If you see needs for new types that are currently absent in Freebase, feel free to take a shot at putting together a schema in your own private domain (try it on sandbox.freebase.com first if you're experimenting). You can publish your types when they're ready for public use, and let us know if you'd like them considered to be promoted to public types. Some of the domains you see in Freebase today are developed from user types or results of collaboration between Metaweb staff and Freebase users.

Philosophy definitely should be a type in the Humanities domain. This domain could include classics,English literature, critical theory, philosophy, history, etc.

I strongly agree with the the mathematics domain.

i agree with phree. we need a humanities domain. we need it to be separate from mathematics and science.

We have certainly considered separate Mathematics and Philosophy domains. Typically, domains are fairly fine-grained and "humanities" seems a bit broad. History is spread across many domains (see the discussion below). English literature could be a domain in itself, although much of it is currently subsumed by "publishing". Critical theory could also be its own domain.

Here's my question: What types in the humanities would you specifically like to see that aren't currently in Freebase?

"Humanities" is an American point of view on scholarly domains, that has no translation in some other systems. I would not group them.

I've tagged as many relevant things as I could find as "Philosophy", "Epistemology", "Metaphysics", "Ethics", etc. and made these types public. I don't think they're complete, but it's a start.

Philosophy is a "tough" category. There is a philosophy that can be related to almost all topics. I think the notion "study of" is the place where it belongs as a specific type. There must be a way however to include it as a property for other types. I'm not sure what the structure would look like....I think we're creating a neural net of sorts....

In contribution to the community,

PMM

@PMM: I don't think that that's sufficient tho. "Study of..." might work well for as a stand-in for "Philosophy of...", in, say, "Mind" but not "Language", because the "Study of Language", aka "Linguistics", is not "Philosophy of Language". Furthermore, "Study of..." doesn't work well when you're dealing with things that _aren't_ studies of something, but still are philosophies, e.g. Existentialism, which is of the type "philosophy", but not the "study of" anything in particular.

No problem with your names, but trying something different: I just created a type "Algebraic structure" (following Wikipedia's definition) and classified "Abelian group" as an algebraic structure. I guess all the other types of groups would also fall into this category ...

Good catch, we'll fix it.

The automotive domain can be found here: http://www.freebase.com/view/domain?id=%2Fautomotive

I've been doing some work on the Automotive domain over the last week and have made it accessible under the special interests domain category.

There's a Physicists type in the Science domain and Einstein is correctly typed as such. There's also a Field of Study type in the Education domain, and a Profession type in the People domain, and Physics is typed as both. A profession has specializations, and I think Optics, Magnetism in my book are specialized areas in Physics, which you can fill out here:

http://www.freebase.com/view/?id=%239202a8c04000641f800000000002da19

In terms of adding more Physics-related types, we'd love to have more types and more data! I encourage you to try to model the schema in your private domain and publish it when you're ready for feedback. Good luck!

Personally recomment the types: * Unit ** SI-unit ** CGS-Unit ** etc. * Physical Dimension * Physical Dimension * Physical Constant

Johann, check out the measurement_unit domain, which though still in flux is where this stuff is going to live.

I've revisited "International System of Units"; I think that can work as example. I would check the physical dimensions first, because the units are defined on top of the dimensions.

You should be able to add values to the “Also known as” property for any topic. Note that currently, all such additions are marked as English, even if they don’t use Latin script; multilingual tools are being developed, but aren’t ready yet.

Ok, that's what I was wondering...if we had the ability to flag it as a different language. Thanks.

It's a topic. That doesn't mean that there couldn't also be a type for it, but the latter would only be useful if there were lots of things with QTVR-related properties.

I believe the expectation is that any former player in a given sport would be entered as an instance of Pro Athlete (http://www.freebase.com/view/schema?id=/sports/pro_athlete) or a sub-type of it, in the relevant sport domain. The career end date property for that instance would then be the primary indicator of whether the athlete was still active or not.

The best way to get the ball rolling on this is to model the types in your private domain and start adding some data. Once you're happy with them, drop a note to an admin in one of the related domains so they can take a look at what you're doing.

Good luck - and have fun!

* You should have got a warning popup window when initially logging in using Safari 2.x browser (unsure if Safari 3.0 beta would be any better) that would advise to use Firefox 2.0 and IE 6 & 7 at this time. Let us know (with details of OS/browser version, etc.) if you are still not seeing it.

* Sandbox data is not replicated/migrated from Freebase daily, which is why it may take a few days (maybe up to a week at the most, usually). Your user data while be updated there and you'll be able to login then on Sandbox.

* I was able to add Tuff Gong (an existing Recording Label), is it creating a new topic called "Tuff Gong Worldwide" that was giving you trouble?

(Is it Worldwide or International? I couldn't tell from the Rita Marley and Ziggy Marley websites what was correct and there seems to be no record of a T. G. Worldwide)

* All users are assumed to be humans and can have a topic that is themselves...Not favorite cheese or movie. So create a topic that is you and fill in as much bio data you wish.

Wanted to check out the sandbox to play, but was unable to log in...it is the same as here? Or do I need to register again for there?

Sandbox gets refreshed with data from the main site every Monday. I'm guessing there hasn't been a refresh since you got your account and tried to log in. So you should be able to log in to Sandbox once the refresh happens.

Thanks GM & Patrick for the quick reply. I'll try the sandbox laters.

I saw where "Tuff Gong" was added as ZM's label. His current disc is with "Tuff Gong Worldwide" a newer of many variants of this name. We're hoping to get the site up in August. When I double clicked on add, I entered the name, then saw the link to "add new record label" but when I clicked it, nothing happened.

I opened Freebase in Safari again, and this time, did see content. Maybe it was timing thing? Didn't see a warning. I'm using safari 3.0.2(522.12) and OS X 10.4.10

btw, seems I posted these questions in the wrong place, sorry, I didn't notice until afterwards - I thought this was a general dboard, but realized it was actually for Domains and Types. Can you let me know the better place?

Thanks

I'll jump in with a couple of additional answers.

Your user profile and the Topic about you are separate things. Your user page is the place to list things about yourself as a Freebaser; your topic page can include much of the same information, but it could be relatively different, reflecting your general public roles. And, as you note, anyone can edit the Topic about you, but only you can edit your profile page.

If you create a topic page about yourself, you can link it to your user page. If you're interested, here's the trick: on your user page, in the "Topic About Me" property, type your name until autocomplete brings up the topic about you. When you save that, you'll not only have the "Topic About Me" listed on your user page, but you'll also have a new topic-page property called "Users who say they are this person"--and your username will be an entry.

On the documentation front, I'm working on more civilian-level write-ups, and we're looking into helping users contribute, too. Is there something specific you want to see and/or work on?

Best,
Sarah

Thanks for the extra info Sarah.

"On the documentation front, I'm working on more civilian-level write-ups, and we're looking into helping users contribute, too. Is there something specific you want to see and/or work on? "

As yer already on it, maybe you could tell me where I might be able to assist? My interest is trying to speak human amidst all the tech stuff these days.

guess not then

Hey, Daneane. I'm trying to get a place set up on the site where we can talk about the Help files (there currently isn't a discussion associated with the Help generally). I thought it would be up by Friday, but since it's taking longer, I'll drop a note to your user page as soon as we have it together.

Best,
Sarah

ah, I see, k.

I wouldn't do so. Everything with a date in the past is History.

I think this is right. Domains should contain their own history, and, indeed, some domains will have a large proportion of historical data (such as "Visual Art" and "Military".) Also, domains are typically more focused -- humanities or even history are not fine-grained enough for a well-defined community. Domains really are communities, after all.

Interesting intuitions, robert. Not sure (as a historian conscious of belonging to a community of historians) that I quite agree. There are specific questions which arise in relation to types for historical topics which may not arise for contemporary topics. Take "Person": where their dates of birth and death are not known, the dates in which a person flourished is standardly used to disambiguate them. Take "Country", "Administrative division" & other similar within location: where these only existed between particular dates we need the dates where they did so (and don't want to confuse them with currently existing locations). Ditto for Government, etc. Is it sufficient to have 'Historical X' as a subtype of X, with some additional properties? In some cases this may be enough (though we don't want cases of merely historical X to be thought of as cases of presently existing X). In other cases, the present-day ontologies may not apply well. For present day purposes the predominant form of the sovereign nation-state makes it natural and relatively unproblematic to talk of countries, yet at least before the Peace of Westphalia empires and other sorts of overlapping jurisdictions are much more evident. To the extent that the past is a foreign country, it needs its own map and does have its own community of mapmakers.

We've seen some of this problem already. Many 'person' topics have nationalities from the "Soviet Union" and "East Germany". For now, these (along with other now vanished historic nation-states, loosely defined) are countries on par with current ones. "China" is even more of a problem since it would be hard to argue that it has been the same country though its several thousand year history. Beyond this, there are modeling problems -- the nationality property in the person schema 'expects' a topic of type 'country' on the other end. It can't be either 'country' or 'historical country' -- it can be only one type. Were we to make a 'historic person' and a 'historic country' that were compatible, we will have lost the ability to query people throughout time. Also, this cutoff is arbitrary since there are many living people who were born in historic countries (my wife's family is from Czechoslovakia, for instance.) The better solution would be to qualify assertions with time. Although we currently don't have this capability in a global sense, we can do it where needed. I can easily imagine that the "is a" relationship for a country instance to the type could have a date range that defines when it was valid. So the "is a" relationship for Soviet Union to Country would be from 1922 to 1991.

This is a problem that Robert and I have begun discussing on my weblog. One problem here is that the Type 'Country' is not static nor singular. Not only are countries, their names, constitutions and boundaries, dynamic, which is obvious, but the concept itself is dynamic and diverse, even at any given time. This point can be extended to all entities even people, though it is conceptually a bit harder to understand with people. My point, which I may be able to make clearer now, is that it is not enough to say that Czechoslovakia is a country, but we also must say when and how it was a country. The same holds for a person. It is not enough to say that a person is a father, but we also need to say when and how he was a father. This is partially accommodated through the link to the child, but what if we say that the person was a doctor? When were they a doctor, how did they become a doctor, what sort of a doctor, when did they stop, why? People, countries, objects, ideas all have biographies. What Robert has suggested, the "is a" relationship, which has a time attached, goes a long way towards a clever solution. However, you still need some way of accommodating, and associating, the many biographical accounts which make up the identity of these topics. If Freebase could find a way to represent these biographies, they have cracked one of the big on-line knowledge problems.

A couple of weeks ago, I proposed something that may help solve this. Every assertion, 'is-a' or otherwise should have the capability to have a date range when the assertion was valid. Most assertions won't, but where it is important, this meta-information could be added and queried. It also sounds like you might want a meta-meta assertion on the dates that suggest why the particular state started or ended. Technically, such a model can exist in Metaweb, but the UI would be very complex. Am I missing something?

I like your proposal of the dates, as this gives the potential of a timeline for topics and, hence, the beginning of a biography for topics. I won't pretend that I have your competence in data modeling, I am but a philosopher and historian, though I do build web-apps as well. My point was a conceptual one, which, I think you will agree, is the ultimate basis of the data model. Apologies aside, my point was that it is important to know why we assert a topic is a topic, why we assert it persisted when it did, etc. However, I don't think you need to go so far as to create a meta-meta assertion, which I agree would over-complicate things. Since the why, more so than the when, what, where and who, is always 'in dispute', why couldn't the associated discussions accommodate this?

robert, your proposal of allowing the general capability to date-range assertions would offer a radical flexibility, & I think it would definitely be useful to historians. If you think it can be done reasonably unobtrusively - i.e. without making either the UI or the API too confusing or forbidding to users uninterested in dates (of the non-edible variety) - that would be great. It won't of course short-circuit concrete debates on how to type particular historical topics, but it would provide a conceptual vocabulary in which to work these debates through.

This thread is of interest in that there is some overlap with work that I do in the field of futures studies. I would encourage you to look at setting up types that could be linked with a set of types that look to the future (i.e. trends, forecasts, etc.)

Hi Robert. You say "Every assertion ... should have the capability to have a date range when the assertion was valid." I'm excited that you are considering this. If the assertion also has the capability to have the context in which the assertion is made, it solves several problems for citing sources, assertions made about real people and places in a fictional context, etc. Permit me to copy my comments from the other post:

I can easily come up with lots of examples of why we need a context for each assertion because:
A. Every assertion has a source and set of assumptions. (Even well-meaning people will come to different conclusions)
B. Every assertion is true at some point in time, and can change later (because the world changes).
So, every assertion is in the context of a source and a time. Freebase can only skirt around this by assuming every assertion has only one universally greed source, and that circumstances will never change. This hampers the knowledge that Freebase can capture. Examples:

1. What is the population of India? Instead of making a special composite type for population, the simple assertion of 1.12 billion should be marked in the context of the source of the information and the time at which it was true.

2. In what viral group is this virus? That depends on what you assume is its genetic function. Reasonable people can come to different conclusions. Allow assertions for both group A and group B, but make the context of each assertion visible.

3. Who are the children of Thomas Jefferson? There are different claims of paternity. Instead of making a different topic for each "fictional" version of Thomas Jefferson, allow the assertion on the main Thomas Jefferson topic and cite the source of the claim on the assertion. (Currently, Freebase tragically obliterates all the footnotes in Wikipedia articles that cite sources, because a citation is a property of an assertion, not a topic. Freebase needs this mechanism.)

4. Who are graduates of Georgetown University? Apparently, one of them is Zoey Bartlet. Why? Because in the explosion of parallel types for Fictional Universes, no one has created a "Fictional Education" type, so the fictional character is made to be a real Person, so that it can be asserted that she went to the real Georgetown University. Is the solution to make a "fictional" Georgetown University and Education type to go with it? No. The assertion that Zoey Bartlet went to Georgetown University is in the context of the West Wing TV show.

5. How much does Oprah Winfrey weigh? Right now, the Weight property is a single value with no source or timestamp. But like all properties (height, religion, even gender nowadays), they change. Even the property Place of Birth can often legitimately be disputed. What does it mean if every month people go in and change the Weight property? Are they correcting a previously incorrect assertion, or are they correctly asserting that at the moment, according to this source this is true? (The Page History on a topic is not the place to track changes over time because it doesn't distinguish between actual changes to a value, and people just trying to correct mistakes.)

6. Who wrote the book Venus on the Half Shell? According to amazon.com, it was Kilgore Trout. But this is an attribution to a fictional character created by Kurt Vonnegut. It was really written by Philip José Farmer. How to handle this? Do we need a 'Fictional Book' type so that there can be a topic for the fictional version of the book? No. It's easy if we put the assertion that Kilgore Trout wrote the book in the context of Kurt Vonnegut's fictional universe. This requires Freebase to show properties on assertions, not just on topics.

I'll stop there, but the list goes on.

Right now, the Country Population property is unique value, but the City Population property allows multiple values so that you can. From the "track changes over time" page. I tried this on the population of San Francisco. Since population is a dated integer, you see the change of population over time. Is that the right approach, and if so, why not do it with all dated integers including Country Population?

Try again (should have used the Preview button)...
Right now, the Country Population property is unique value, but the City/Town Population property allows multiple values so that you can "track changes over time" as it says on the discussion page. I tried this on the population of San Francisco. Since population is a dated integer, you see the change of population over time. Is that the right approach, and if so, why not do it with all dated integers including Country Population?

jefft0: I completely agree with you that most assertions should have a timestamp attached to them. Most data changes over time, but it's valuable to have a history of the progression of that data. Adding a temporal element to assertion would make the data multi-dimensional in scope. I saw another post under the Data Modeling Guide that talks about adding time-series data to fb. I would imagine these two initiatives are related:

http://www.freebase.com/view/discuss/%239202a8c04000641f80000000045637f3

Also, in regards to your Zoey Bartlett inquiry, wouldn't you be able to define your query such that you search for all persons who are not fictional persons? Essentially, there is a implicit type of "Real Person" which is defined as [is "Person" and is not "Fictional Person"]. This type is useful, but I wouldn't suggest having someone go around and add a "Real Person" type to all persons not listed as "Fictional Persons". This manual task would not be sustainable as the data set grows. If there was a way to create implicit rule-based types by defining their relationship to existing sets, we could avoid this problem.

I could see an application for this in many areas. For instance, it makes sense to make the "Deceased Person" type implicit, defined by the rule [is "Person" and "Date of Death" not null]. Of course, there could be a case where we know a person is deceased but the date of death is unknown, so maybe we should make the implicit types "soft", serving as a default type to records that match their rules while allowing users to add the type manually as well.

As it happens, I'm working on this already. Expect to see something in the next couple weeks. :-)

And what about Art Galleries ? Galleries = Not Museum + Exhibit Art + Old Stuff + New Stuff. Hmm... I'll work on my equation later, since it currently doesn't add up. (grin)

I'm working on this now. I have a basic Museum Object type defined, but we really need a domain for "Museum Collections", as well as a type for the Institution. As there are so many different types of museum collections, we could define different types under the Domain, such as Archaeology, Anthropology, Natural Science, Social History, Art, etc. I don't think this should get too complicated, otherwise museums will be hesitant to upload their data.

Robin, this sounds great. I agree that it should not be too complicated for the same reasons. Just as a brainstorm, I'm thinking how the new Smithsonian - National Museum of African American History and Culture would define themselves. They will cover topics as varied as slavery, post-Civil War Reconstruction, the Harlem Renaissance and the civil rights movement. Perhaps having a property for governing bodies, board of regents, or foundations that provide funding or maintain their directives and policies ?

Thadguidry, this is a perpetual problem. Museums as institutions are just about as varied as their collections. You could ask the same question of the A:shiwi A:wan Museum & Heritage Center in Zuni, New Mexico, or the Alert Bay Museum in British Columbia, just two Indigenous museums, who would use quite different categories for describing themselves and their collections. You also have the Te Papa, National Museum of New Zealand, who have two communities who define the institution, and its collections, completely differently. Both are part of the governance of the museum. Not easy. I have been approaching the Museum Object Type cautiously, and I could use some suggestions here. I have uploaded a diagram of the type, and two new assumed types onto the "Museum Type Structure" topic. This matches the types I have built in my personal domain. They are designed to accommodate the idea of freebase, but also make it easy(ish) for museums to upload their data. I have used to dominant museum documentation standards (SPECTRUM and CIDOC) to ensure that these properties match, to a sufficient degree, what most museums do already at a basic level. However, then your problem arises quickly. What about all the difference below this point? I think that the solution, or something like a solution, can be found in Community Domains. That under a general Museum Domain, there are other domains that accommodate different types of museums. These can also be different community's approaches to museum collections, not just "official" categories. Perhaps we should start another discussion about this?

I am interested in what you all set up for Museums and their Collections. I am pondering similar issues for Archives and their Collections. Often there is a lot of data out there about individual archival collections - and I hope to make it easy to import data into Freeweb. The main archival descriptive standard is EAD - Encoded Archival Description. It permits XML encoding of data against a standard DTD (Document Type Definition).

I agree with robinboast about ensuring that we pick a good set of basic metadata for archives and their collections -- but I also want to make sure Freeweb has enough detailed data do interesting things with once it is captured.

I have already had a go at a basic Museum record type, which I will make public today. If you have a look and let me know what you think, it would be much appreciated. The problem I have, which is the same one that CIDOC and SPECTRUM have, is after a very basic set of properties, how do we accommodate the vast diversity of different types of collections (art, anthropology, archaeology, social history, natural science, etc.)? Any ideas would be welcome.

I think the database of oil companies idea is great! You can create your own types if you start with "Types" -> "Types you've created" - from there you can start to model what an oil company is, attributes that describe where their oil fields are, etc.

How about starting with an "Oil Field" type? (There's a list of oil fields e.g. in Wikipedia.) Then an "Oil Company" type could have a list of which oil fields are used by that company.

Is there a cross-reference of which oil companies use which oil fields on the Web somewhere?

I've taken a crack at this here, see the Wilmington Oil Field. It could definitely use some further refinement, but its a start at modeling 'oil field' and 'oil company'.

Looks good to me: I've added in Shell Nigeria and Bonga Field.

Looks like the ownership structure could get complicated: Shell Nigeria is a joint venture "composed of Nigerian National Petroleum Corporation (55 percent), Shell (30 percent), TotalFinaElf (10 percent) and Agip (5 percent)" (Wikipedia). So should we say that all these companies own Bonga Field? (Maybe this is more of a question about joint ventures ...)

hmm, I thought I clicked reply, try again...

Just and idea for ownership:

"Oil Field Ownership" (a "compound value type")
Properties:
* Company (expected type "Company" or maybe a new type for an oil/energy company
* Percent stake (et: float)
* Start Date (et: datetime)
* End Date (et: datetime)

or scratch the last three and have a property "Stake" that points to yet another CVT "Oil Field Stake" which would contain the Percent Stake, Start and End date.

Not sure if the complexity of the latter solution is really necessary.

i actually went more abstract than that, see: Asset Ownership

You may be right to put in a start and end date as well.

Should Asset Ownership also be used for joint ventures?

I just found a guide to ownership structures for at least natural gas companies at http://www.epis.com/EnergyLinks/natural_gas.htm

Yes - that was the intent, that it could be used to model joint ventures. FYI, we're likely going to migrate these types from my personal domain into the public 'company' domain shortly.

Cool source. Wonder if there are any rights issues with using that information?

Great - will try it out.

There's a statement on the parent page http://www.epis.com/EnergyLinks/Default.htm
as follows:

'This collection of links by category was developed to support market research across the wholesale energy market. These links are provided as a service to our EPIS/AURORA customers and website visitors. However, EPIS does not necessarily agree with positions profiled in these sites and does not guarantee or review the content for completeness or accuracy. This catalog of links is actively used for research so these links and the structure of the list will change without notice to reflect both changes in the source pages profiled and the structure of the industry. The information displayed here is continually "in development" as relationships in the industry come to be understood or the industry evolves. '

For a Company, won't there be a potential conflict between the Parent Company and the Owner(s)? See e.g. Shell Nigeria. I guess any state-owned company should have the state as the Owner; should the state also be the Parent Company? Do we need a "State-owned Company" as a "Company type"?

The taxonomy on this page is not intended to have much semantic value -- it's just here to make our types more browsable. In general, most taxonomies are not widely used except by experts in a particular field. That said, I think it would be great if we had LoC and others integrated into Freebase.

Robert thanks for reply. I agree with what you say although taxonomies are used more and more by 'normal' users as well and evidenced in the discussions here on freebase. Look at France for instance where this is much more common than in the US, also in Germany, Sweden, and the UK on websites. Most people like it when they come across it. I still think it is a bit too haphazard today on freebase but perhaps it is early days yet..

Torsten -

With the Freebase UI, we have shied away from promoting a central organizing taxonomy because it is so difficult to create one that is useful to most people -- there are often strong personal or cultural biases. Indeed, this is such a problem in general that there is a proliferation of these large-scale taxonomies of ideas.

That said, taxonomies are quite a natural structure in Freebase. They exist everywhere -- any place there is containment such as locations or genres for films and books.

It should be a trivial operation to map well established taxonomies into Freebase. Doing so would make it possible to "pivot" among those taxonomies. That is, you would be able to see where the topic "Japanese Poetry" sits in both the EB and LOC hierarchies and you could navigate using either tree.

If you have a particular taxonomy you would like to map into Freebase, let me know and I'll help you figure out how to import and reconcile it with existing topics.

Robert-
This is very good. Again I agree with what you say. And yes the taxonomies tend to have a cultural bias to a certain extent, but perhaps more so in the social sciences than natural. For instance, the area studies or history taxonomies usually fan out from one's own country or continent. But that is pretty easy to "rectify" in a more "objective" way.
In natural sciences the problems of this kind are probably much less.
It would be great to have a pivoting function. I have a liking for the EB Propaedia taxonomy, which is only a bit biased towards America/UK/West.

Thosten,

My guess is that a taxonomy like Propaedia is copyrighted. If you know of a taxonomy that has useful parts that in the public domain (such as something from EB long ago), then we could load that.

Hi Judith -- We've spoken with them and we're working out how to share data.

Yes, the religion domain is at:
http://www.freebase.com/api/metaweb/view?id=%2Freligion&typeId=%2Ftype%2Fdomain

All registered users are able to add or modify data, and that includes you. :)

No, there isn't. At this point, most sectors, including telecomm, are industries in the business domain. (I see you've been adding topics and data there, which is great!) Btw,the Types link at the top of the page takes you to a page where you can see all of the domains--in the center, there's a blue box with category names; click a link to get a list of the domains in that category. Best, Sarah

Hello.

I've been a Freebase user for about 10 hours now. I've been a telecommunications consultant for about 20 years...

I would like to suggest that a Telecommunications domain would be very valuable. In addition to providers / operators / businesses, I would like to suggest that the technologies involved should have their own topics. For example, what is a DS0? A DS1? A DS3? An OC3? How do these related to T1s? E1s? T3s? E3s? How many of each can be carried within each? How might a 100mb/s internet link be carried within? What might the interfaces look like?

A database of LERG entries would be good too (just how does the legacy PSTN operators exchange their traffic? Who (particularly in the Monopoly dominated developing nations) are not following the rules?).

Exciting possibilities.

-Chris

Hey, Chris. Welcome to Freebase! If you're interested in expanding the schema for telecomm--and it sounds like you'd be well-qualified--I'd suggest working on the model by creating some types/co-types/properties, and then adding some data to see how it's working. Other people will be able to see your types as you're working on them, but they won't be able to see the properties or data unless you publish the types. The Help page (link at the top of most Freebase pages) has a section on data modeling, and we're happy to answer questions, too.



Best, Sarah

It looks like Gilad has started work on telecom companies.

Robert, Chris,

I played a litle bit with companies and technologies. Its huge subject of-course and I have no idea how to organize it, so I just play with types and topics. Hope it makes sense.
I added some very basic info about DS0, ... You are welcome to add more meet on this.

Cheers,
Gilad

Hello Robert, Gilad and Sarah.

Despite some system problems earlier in the day, I've also been working. Have also added some TDM types (DSO, DS1, DS3, etc.). Please review -- I'm also new at this.

I would argue that "Telecommunications" requires a domain of its own. Companies -- Circuits -- Interfaces -- Cables, etc. (Oh My!)

Thanks.

-Chris

Hi Chris,

1) I fully agree that "Telecommunications" requires a domain of its own.

2) How we can we merge your definition for DS1 and DS0 and mine (which is just place holder)? I guess there is some mechanism to do it. Sarah, can you help here?

Best,
Gilad

Hi Gilad.

We appear to have a "definition space" collision here. If this hasn't happened in the past, I'm sure it's going to happen again in the future.

Sarah -- I echo Gilad -- how can we merge our Instances and Types? Can you help? How should this be handled in the future?

I've also defined (although not made public) a few additional types, including submarine cable systems. Tricky stuff.

Please let us know how to proceed -- duplication of efforts make little sense.

Thanks, and best regards to all.

-Chris

Merging instances is simple, just click on the pull down flag on the right of the title of the instance (e.g. there are two DS1 topics) and then select the Merge option. The instance/topic is flagged and soon the data gardening process will eventually merge both into one instance/topic.

It's great to see both of you (Chalsall & GiladGoren) so enthusiastically contributing, I have a suggestion though:

What would be best is if you modeled all of this out on Sandbox.Freebase.com (next time if you have it all worked already on Freebase.com), which would allow for mistakes and errors to be simply wiped away each week. Sometimes the changes for mergers, splits and deletions take more than a week or so to take effect. Be advised that new members may have to wait until after the following Monday (PST) for their user info and access privileges to be migrated upstream from Freebase.com to Sandbox.Freebase.com.

Hi Chris,

I solved the collision manually, by removing my definitions and replace them by yours. Next time I will use the Merging option explained by gmackenz.

I guess that collisions of definitions are easy to solve. Let see what happen when we will have collision of relations between topics. Right now I added my 'Networking Technology' type to your 'TDM circuit' topics. Its quite easy to do, but I guess It worth discussion (i.e. any comments are welcomed).

It seems their is no tool to merge Types (rather than Topics). I defined type for "telecom standards", and I use it within my topics. I guess that if someone else developed similar Type for telecom standards, it will be hard/impossible to merge the definitions. Comments?

I am less interested in describing specific items, and more on creating the structure of the telecom domain i.e. creating relevant types and interconnect them. As this become more complex over time, it seems that we lack GOOD visualization tool. Comments anyone?

Cheers,
Gilad

Gordon,
Tx for the comment. I guess we will have to wait until tomorrow.
Che