spatialed

Yale University: yale college and yale university?

2008-07-24T14:54:09.0000Z

Wikipedia tends to classify by the broadest institution, so there are a lot of actors associated with Yale University instead of the drama school. I think it's fine that we start with the lower-precision classification and as data becomes available, replace it with the more specific.

Apple Inc.: What will be the next bullet of Apple Inc.

2008-07-24T14:32:34.0012Z

this is the interesting questions, I also wondering what would be the next bullet of Apple

Apple Inc.: What will be the next bullet of Apple Inc.

2008-07-24T07:38:50.0017Z

Apple have bring 3G iphone to the market, any idea about what's the next bullet of Apple

United States: Removed "Official Language" and Liberty Dollar as form of Currency

2008-07-03T16:55:01.0005Z

Removed "Official Language" as per Wikipedia entry:

http://en.wikipedia.org/wiki/Languages_of_the_United_States

Removed "Liberty Dollar" as it is not a currency in itself, but rather a unit of currency.

spatialed: Properties co-typed as other types

2008-07-01T15:34:09.0005Z

Thanks, Ed.
I think Physiognomy and Satellite sensor status (the properties) should be separate, discrete objects than Physiognomy and Satellite sensor status (the topic and LC) and not conflated. You do have a separate Physiognomy type (http://www.freebase.com/tools/schema/user/spatialed/land_cover/physiognomy) - should that have included types of topic and LC?

spatialed: Properties co-typed as other types

2008-07-01T04:09:05.0012Z

Sure but they were purposeful typings. LC is a bucket type I created for all types that have something to do with land cover mapping even if they are best placed in another domain. They are usually topics too. No problem if you want to detype them. I've put that project on the shelf for a little bit and know how to pick it back up later.

spatialed: Properties co-typed as other types

2008-06-30T16:41:39.0012Z

Hi Ed,

I've noticed that you have 2 properties, Physiognomy and Satellite sensor status cotyped as LC and Topic. I'm guessing this might have been an errant mistyping due to autocomplete. I'd like to detype them - is that ok?

United States: Liberty Dollar

2008-06-27T16:52:19.0005Z

I don't know that this can be rightfully linked as a form of currency inside the US as it is not recognized by the Federal government.

By the same token, the United States has no official language, although English is the predominant language.

Organism observation: Are observations right for freebase?

2008-06-25T15:19:54.0018Z

Sorry I haven't had a chance to follow up. I've been on vacation. Lots of ideas, no time. I hope to follow up sometime next week.

Spaceflight: Spacecraft-Type?

2008-06-20T19:22:07.0000Z

Late to the party, but I'm hoping someone still reads this. Could I suggest adding some types to "Spacecraft"? Ideally these would include missions flown, date built, and date destroyed/decommissioned/lost. Thanks!

Kawasaki Heavy Industries Consumer Products and Machinery Company: Merge then split

2008-06-17T00:54:51.0007Z

Thanks Jeff. Very useful information in those links.

Kawasaki Heavy Industries Consumer Products and Machinery Company: Merge then split

2008-06-16T23:34:20.0007Z

This is confusing, but I think they shouldn't be merged. According to this org chart from KHI's webpage, the Consumer Products and Machinery Company (Wikipedia's current name for what is titled in Freebase "Kawasaki motorcycles") is a division of Kawasaki Heavy Industries, and is not equivalent to Kawasaki Motors Corp, USA. I'd create a new topic for that, and link it as a subsidiary of Kawasaki Heavy Industries. (More fun with KHI's organizational status can be had here: http://www.khi.co.jp/kanren/index_e.html.

Organism observation: Are observations right for freebase?

2008-06-12T07:00:03.0000Z

I changed the name of the "Organism observed" property to "Taxonomic concept". I also added a "Taxon description or key" property that is expected to be a written_work. We may be able to eventually drop this new property after organism_classification is linked to its published description(s).

Based on our discsussion, "Taxonomic concept" is probably a more appropriate property name than both "Observed organism" and "Observed organism classification". The name of this type, "Organism observation", is still probably appropriate.

To summarize the schema: The naturalist observed, at a specified location, an organism or part of an organism such as a fruiting body (not an organism classification) and then labled that organism on a specified date as a taxonomic concept based on a description or key that was usually written and published.

Do these modifications improve the utility of the type for you?

Should I add a Person property so that entries can be added for non-users? This would add denormalization between user and person but I can't think of a way to avoid that except by dropping the user name property. Some users may object to giving their names so that might not be a good strategy.

Organism observation: Are observations right for freebase?

2008-06-12T05:35:28.0007Z

A lot of the data in VegBank was collected through the Carolina Vegetation Survey "pulses". There is usually at least one and sometimes two pulses per year. A CVS pulse is organized by a small group of plant experts. The organizers rent a house or campground for a week and send out a request for volunteers. Everyone is welcome and over fifty volunteers typically show up for at least a day. The organizers and "honchos" (typically graduate students who are good botanists) lead crews of volunteers to sample various plant communities using standard protocols. I've volunteered two days each of the past three years. It's a lot of fun.

After a full day of "botanizing", the crews have a late dinner and then spend a couple hours with lenses, microscopes and published taxonomic keys to identify anything they couldn't ID in the field. The toughies and remainders are pressed each night and later identified by graduate students and botanists at herbariums.

I don't think anyone ever notes which key is used. My impression is that each person uses whichever key is most familiar and/or most recent at that date (or the one they authored!). This is probably a failure of most people and programs and it is good for you to call attention to it. I agree with you that the reference should either be explicitly referenced within the record or as part of the sampling protocol .

Organism observation: Are observations right for freebase?

2008-06-12T05:22:11.0007Z

Riffing off your post from yesterday, I don't quite understand how your suggestion about using a date field works. What I want to be able to do is see for a given observation what names have been applied to it by whom, when they were applied, and why that person thinks this is the correct name. In my own site I allow other users to vote on the names so over time you develop a consensus.

The concern I have with a "Revised oragnism observation" is that it seems to assume a strict discrete linear ordering to the namings. As the VegBank example shows the scientific consensus process can take years to settle out. But perhaps I don't really understand your suggestion. It seems like we need some concrete examples.

What you observed about the old names being useful data for misidentification is correct and we use it for exactly that purpose on Mushroom Observer.

I haven't gotten my head around CVT's yet. I'll have to try creating some types in my sandbox to get a feel for it, but I'll take your word for it.

Your Classification code type is interesting. I can see how it works well for what you're describing, but I don't see the same value for the "Overlaps" concept in a taxonomic DAG. In my, you can't know a prior that two taxa overlap. It is just a side effect of each them containing overlapping children. However, the general concept seems very similar.

I would definitely be interested in exploring confidence ranking ideas. What's the best forum to do that in? I'm also interested in taking this conversation to the data modeling list. I've been lurking there for the last week or so to get a feel for the way the discussion goes. It's also been really helpful having a chance to discuss it more in depth with just one other voice. Thanks for the time you've put in!

Organism observation: Are observations right for freebase?

2008-06-12T04:57:46.0000Z

Thanks for the VegBank reference. I hadn't heard the term "concept-base taxonomy", but it makes a great deal of sense to me. I haven't quite figured out how they actually deal with cases where the observer has a scientific name and either doesn't know or doesn't record the original reference. I haven't been able to actually access the database.

The common case where this comes up are people who have learned or been given latin names through essentially oral tradition. As the VegBank discussion points out, you can generally find the original publication for a particular name and get that author, but you don't really know if that's the concept you are working with or not. In some cases it may actually not be possible (yet) to efficiently track down any solid concept for a name. However, even if that's possible, people often don't bother to be that precise. In my opinion, you need as best as possible to actually describe what information people are using to make an id. Thus a naming should carry with it as much detail as the user is willing to provide. This can include a specific reference to a concept as well as notes indiciating the specific features that were used to make the determination.

Motorcycle part number (manufacturer ): Link to Consumer product?

2008-06-11T20:38:15.0018Z

I bit the bullet and co-typed consumer_product. Seems there's a couple benefits and no real harm.

Motorcycle part number (manufacturer ): Link to Consumer product?

2008-06-11T19:32:46.0000Z

Ah, I see. That's definitely more complex than the consumer_product retailer links. About the only thing that you would really gain by including consumer_product is the ability to have non-blog or -forum reviews (are there webzines that review motorcycle parts?), so maybe you don't really need the co-type, unless you want the "is-a" assertion.

Organism observation: Are observations right for freebase?

2008-06-11T18:14:06.0000Z

I just remembered that the creators of VegBank have given the integration of multiple taxon concepts quite a bit of thought. Check out their webpage that describes VegBank's included databases. In particular, see VegPlants.

Motorcycle part number (manufacturer ): Link to Consumer product?

2008-06-11T06:10:54.0000Z

Consumer_product has a link for purchase links. My approach was to make these links through Motorcycle_part_number_(retailer), which is often different than the manufacturers number. This approach allows users to look up and cross-reference manufacturer and retailer numbers AND give retailers a place to advertise without sullying the actual product page (i.e., spam). Co-typing with consumer_product wouldn't be that big of a problem but it does add some redundancy. I've also started subdividing "reviews" into separate blog and forum links. Perhaps this is overkill, but I bet a lot of this info could be scraped, and the subdivisions are informative.

Motorcycle part number (manufacturer ): Link to Consumer product?

2008-06-10T18:36:13.0007Z

My understanding of the type (and correct me if I'm wrong) is that the instances of "motorcycle part number" are themselves the things you buy. The short description seems more generic, and often seems to describe a class of part, rather than a specific part. If that's right, then "consumer product" should just be a co-type. I don't see any duplication of properties between "consumer product" and "motorcycle part number", so I wouldn't expect that you would need to do any refactoring.

Literature: Merge with the other "Literature" instance

2008-06-10T09:29:54.0012Z

Duplicate of Literature. Merge please.

Organism observation: Are observations right for freebase?

2008-06-10T06:39:15.0007Z

Wouldn't the date field handle most of your concerns about dynamic namings? For example, I may want to summarize J.J. Audubon's initial bird species observations and classifications. Many of those have changed. By linking the date of the observation, the name at that time is maintained and can eventually be linked to the current taxon definition(s) given future taxonomic crosswalks (i.e., links between synonyms and reasons for the links). It may be useful to link the classification system explicitly, but this is already done indirectly through the organism classification (i.e., currently every taxon is identified based on the ITIS classification system). I'm fine with that as long as Freebase will accept multiple taxonomies.

A "Revised organism observation" type could be modeled to track changes in taxon identification due to literature and picture reviews by the observer and others. There could be utility in tracking these changes in organism identification. For example, which species are most commonly misidentified?

One of the annoying constraints of Freebase is an inability to nest CVTs. This limitation has some implications for your proposed amendments to the types.

Regarding multiple higher classifications, I think this screams for the need for multiple taxonomic classification systems that are linked at every node (i.e., crosswalks). For an example, check out my Classification_code type for land cover classification systems. An attributed topic is Woody wetlands.

Regarding confidence rankings... cool. I would be happy to help you refine it. One of my colleagues is a social scientist who may be able to provide some guidance. Someone else will need to chime in on whether something like this already exists in Freebase and answer your question about reasons for priority of image display.

Something to consider is that you might get more feedback by posting these questions to the data modeling list.

Organism observation: Are observations right for freebase?

2008-06-10T05:38:46.0000Z

Interesting discussion in the meta-contributors list. Seems like the Mushroom Observer observations would be reasonable then. However, to me this argues even more strongly for separating the naming from the observation. The facts of the observation are known for sure. The namings are a matter of opinion and can shift over time both due to new eyes looking at the observation and from changes in the literature.

For example, the observation you've labelled Amanita muscaria is currently accurate, but it is expected to change in the not too distant future (perhaps this year) due to new DNA studies on the various species of red capped Amanitas. Ideally your observation would get the new name applied to it once it is published and some determines which one you found, but your old naming would remain for the purpose of search.

It might be reasonable in Freebase to handle this by making the 'Observer organism' field allowing multiple values. However, I'd want to associate a naming with a particular person since different people have different levels of authority with respect to a given group. Also having the date of the naming is useful as is the ability to indicate which naming is currently considered the most appropriate.

Regarding the Organism Classification type, it's better than some I've seen but doesn't have as much data as I'd like. In particular, I would like to see the Synonym scientific name reflect back to other Organism Classifications in some way and an option for multiple Higher Classifications.

Regarding synonyms, I've found that treating the concept of a taxon as distinct from a name is really useful. Names are the strings published in a book (e.g., Amanita muscaria (L. : Fr.) Lam.). Taxa are the actual organisms that a Name is applied to. In a database, a taxon essentially ends up being a list of names with a subset being marked as 'valid'. Most often that subset only has one member, but there are cases of long standing debate where a shared collaborative space should really allow both opinions to stand on equal footing.

Regarding multiple higher classifications, as much as people like to think of the current taxonomic relationships as a tree, it is really a directed acyclic graph. Especially when you get into things like plant hybrids and even more extreme genetic engineering, the tree metaphor breaks down. It is also fundamentally based on a clear concept of species which simple doesn't exist in some groups. Even without explicit examples of such violations, the current state of knowledge is incomplete and any presentation of the data as a tree necessarily follows a particular opinion which may be under significant debate. Interestingly, a DAG option is already available for the more abstract Organism Classification Rank, but the data and the images definitely need some work.

Regarding confidence ranking systems, I've come up with something on Mushroom Observer that's doing a reasonable job, but I'm not completely happy with it. It is also related to ideas like quality and interestingness that I haven't had a chance to really focus on yet. Is there anything along those lines that you've come across in Freebase? How does it decide which image to show in cases where multiple images are provided?

Motorcycle part number (manufacturer ): Link to Consumer product?

2008-06-10T04:28:42.0000Z

Do you think it should be a co-type or linked in as the "Short description" property? The former will require additional refactoring here and on other types. I also like my properties better than those on Consumer_product (of course!). :)

Motorcycle part number (manufacturer ): Link to Consumer product?

2008-06-09T17:59:23.0007Z

I think it would make perfect sense to include "consumer product" on this type. If you want to retroactively add "consumer product" to all the existing instances of "motorcycle part number", let me know -- we should be able to do that pretty easily.

Organism observation: Are observations right for freebase?

2008-06-09T07:03:55.0001Z

I posted a somewhat similar question in meta-contributors. IMO individual observations are useful if they are placed within a well defined spatial and temporal context. That allows users to lump observations in various ways to describe other things (e.g., what is the diversity of bird species in Umstead Park?). To me, the semantic web is useful for what users can collaboratively put into it as well as what they can individually pull out of it.

I modeled Organism_observation as a compound value type (CVT). This type is essentially a container for a list of grouped topics. I've linked this type as a property of Naturalist's observations, which can include topics like "velosa's fungi life list". See "Spatial Ed's mushroom observations".

I named the property linked to Organism_observation as "Observed organism". My use of the word organism was somewhat keeping in line with Freebase terminology. If I wanted to be completely consistent I would have named it "Observed organism classification" but that seemed more confusing. Organism in Freebase actually refers to individual organisms (e.g., Seabiscuit).

I think Organism_classification is founded solely on ITIS taxonomy definitions. I've made several suggestions for allowing users to crosswalk the ITIS classification scheme to other taxonomic systems. No luck yet but I think it is just a matter of time.

Observer confidence is something I thought about adding as a property linked to an enumerated type (i.e., list) but I didn't have a confidence ranking system to use so I didn't addone. Feel free to do so. I would add it as another property of the Organism_observation CVT. Make sure to check the boxes for disambiguation and single value.

User name is linked once to all observations as a property of topics of the Naturalist's_observation type.

"Unknown" or "Unknown fungi" or "Unknown Cantharellus", etc. can be added in as the observed taxon in the list of observations in the Naturalist'_observation type if users are not sure of the species.

Organism observation: Are observations right for freebase?

2008-06-09T05:31:26.0012Z

This is my first question, since a single observation is a far smaller (and therefore even more plentiful) unit of knowledge than a taxon. It is also generally a personal event rather than a group topic that can be improved by many users over time. However, assuming that it is, I would describe it a bit differently from what you have at the moment.

First, the property 'Observation' should be something like 'Life Lists' or 'Species Lists'. Second, I would not associate it directly with an 'Observer organism'. Rather I would associate it with a set of 'Namings'. A Naming in turn is associated with a Oragnism Classification (or what I would call a taxon) and would have a confidence and the name of the person proposing the given name.

The big advantage of this extra layer of indirection is it allows unknowns to be observered and it deals with the evolution of taxonomic naming over time.

I could also get into what I think a taxon should be, but I need to do my homework on Organism Classification and see if I like it or not.

-Nathan

Motorcycle part number (manufacturer ): Link to Consumer product?

2008-06-08T10:11:27.0012Z

I created this type as a way to cross reference motorcycle parts with their use on various levels of motorcycle descriptions (e.g., model year, body style). One of the properties is "Short description", which I've set as text. However, this short description is really a name that people typically call the part unless they need to order it. I'm wondering if it would be better to link this property to the Consumer_product type. This refactoring might help advance the Consumer_product type's schema development too. Thoughts?

Kawasaki Heavy Industries Consumer Products and Machinery Company: Merge then split

2008-06-04T02:49:45.0012Z

The Kawasaki Heavy Industries and Kawasaki Motorcycles topics have overlapping descriptions. I think the actual name should be Kawasaki Motors Corp, USA for the motorcycle manufacturer. I don't see any mention of Kawasaki Heavy Industries on that page... not sure how they are related...

HAT-P-2 b: Wrong name: should be HAT-P-2 b

2008-06-02T16:26:12.0000Z

No, do not delete, merge other topic to this...

HAT-P-2 b: Wrong name: should be HAT-P-2 b

2008-06-02T13:10:57.0005Z

Please see HAT-P-2 b instead

Spaceflight: refactoring

2008-05-22T21:30:42.0007Z

Heh. I am admin in this group so I can move them myself if people approve. Maybe over weekend.

Anyone else wants to enter some data? FOr now I am slowly typing data from here:

http://www.k204.ru/books/Aviadvigatel.pdf

as owner of astronautix.com unfortunately is not willing to share his data. Well, he probably has reasons.

Spaceflight: refactoring

2008-05-22T17:54:56.0000Z

Sounds like you need some help from the data architecture group. Are these the only 4 types you'd like moved?

Spaceflight: refactoring

2008-05-18T21:52:50.0000Z

I like rocket enngine, rocket stage, rocket famliy, and rocket engine family. Unless we here something surprising from NASA, I think we should put these into the Spaceflight domain.

Spaceflight: refactoring

2008-05-18T16:08:18.0000Z

Great! Will do. We are waiting to hear back from some people at NASA who have been developing metadata schema so we can try to incorporate those too. After several forwarded emails we last heard that "NASA has done quite a lot of work in this area", but my primary contact is traveling for a couple weeks so we don't yet know what that work was.

Spaceflight: refactoring

2008-05-18T02:20:31.0000Z

Danny, pls look at some my space related types in http://freebase.com/view/user/hedgehog/default_domain

P.D.Q. Bach: Should be split into two topics

2008-05-05T22:04:19.0005Z

One for the person:

Peter Schickele

One for the Fictional Person:

P. D. Q. Bach

Communications satellite: Commercial satellite transmission

2008-05-05T05:19:54.0000Z

Communications satellite: Commercial satellite transmission

2008-05-04T16:37:35.0012Z

There is a lot of great information in this document that could be used to develop some data models that describe commerical satellite transmission:

http://www.fas.org/spp/military/docops/army/comsat/Csfinweb.htm

Academic Library: Thanks for creating this type!

2008-05-01T18:02:28.0007Z

Yes, I think we shoudl probably model it on how hte public libraries work: Academic Library System (with associated institution), which has associated Academic Libraries.

Academic Library: Thanks for creating this type!

2008-05-01T15:26:38.0000Z

I'd like to see a field for linking it to the affiliated academic institution. A complexity that needs to be addressed is that on many university campuses, there are a number of different libraries, each of which may have its own name. I'd recommend that there be topics for each library on campus and field that links to the overall library system topic page and a field that links to the affiliated academic institution.

I'd do this myself, but I have no clue how to.

Spaceflight: refactoring

2008-04-21T19:10:29.0012Z

Spaceflight needs some refactoring in light of what we have learned about data modeling since this early experiment. In hope of getting some help on this, I would like to ask spatialed to join as an admin and help fix satellite as discussed.

Now is a good time to speak up if you are depending on something no being changed.

Academic Library: Thanks for creating this type!

2008-04-18T18:24:30.0012Z

Thanks so much for creating this type. I've added a couple of topics: the Bodleian Library and Baillieu Library at the University of Melbourne.

Can we help you at all with building this out a bit further? We could add properties for collection size, affiliated academic institution, etc. And then I think we'd probably like to promote it into /library if you'd be interested in that.

spatialed: Cyclical included types

2008-04-17T03:03:35.0000Z

Good catch. Thanks. I was still figuring out how to work with Included types when I created those schema. They probably still need some work but I removed the cyclical relationships. Let me know if you see any other problems. I plan to get back to these schema and start adding a lot more data for them May-June.

spatialed: Cyclical included types

2008-04-16T20:00:49.0007Z

you have a cyclical relationship:

http://freebase.com/view/schema/user/spatialed/default_domain/satellite_image_product

i click on the first included type "Land cover classification system" of "Satellite Image Product"

http://freebase.com/view/schema/user/spatialed/default_domain/land_cover_classification_system

and "Land cover classification system" has as its first included type "Satellite Image Product"

spatialed: Cyclical included types

2008-04-16T18:30:13.0000Z

i'm seeing a cyclical relationship in:

http://freebase.com/view/schema/user/spatialed/default_domain/satellite_image_product

a satellite image product has itself as an included type.

is this an error?

Spacecraft: Is a satelite a spacecraft?

2008-04-15T22:49:29.0007Z

This all sounds right to me. I don't feel strongly about renaming naming Satellite, as long we make it clear in the description what it means. If you do rename, I prefer Artificial to Man-made.

Spacecraft: Is a satelite a spacecraft?

2008-04-15T04:55:45.0007Z

I forgot to mention that Earth observing satellite should then be a co-typed with Man-made satellite (co-typed with satellite and spacecraft) and Sensor vehicle.

I'm not sure where Satellite mission fits into all of this. Perhaps it stays as it is - an independent tag to group Man-made satellites of the same mission.

Spacecraft: Is a satelite a spacecraft?

2008-04-15T03:45:47.0000Z

A man-made satellite would include properties of both a Spacecraft type and a Satellite type. So, Satellite should probably be renamed Man-made satellite (or something equivalent) and keep all properties but those that refer to orbit.

Satellite should have the orbit properties of my Earth orbiting satellite type, which are a little more comprehensive and less redundant than Satellite's two orbit properties (mine were modeled using data presented on a NASA page).Satellite could therefore be used to describe anything that orbits another thing.

The two "Operational life" and "Operational" properties of my Earth orbiting satellite type should be moved to Spacecraft.

The sensor property of Earth observing satellite should probably be moved to a more generic type that connotes a sensor vehicle, regardless of its location.

How does this sound?