[OHM] OHM data model - can it be changed? Re: Thesis about Open Historical Map
Thomas Schwotzer
thomas.schwotzer at HTW-Berlin.de
Wed Feb 6 10:04:37 UTC 2019
Hello Jeff,
thank you so much for the quick answer.
My status is this:
We have a data model that covers the mentioned problems.
Have we solved it? Yes.
That model runs on a Ubuntu machine with PostGIS in our
lab. Is it a stable and productive systems? No.
We use GeoServer to render maps. We are far from a thing
that could be called a map. But we know, that Geoserver
can handle that amount of data. Since last semester, thanks
to a student, we even have a date-sensitive WMS cache.
We have good experiences with parliament (GeoSPARQL).
Thanks to another student, we have a running engine in
the lab that works with data from our data model, see above.
I'm software-engineer for a quarter century. I'm pretty
aware what it means to have a productive systems and what
efforts are necessary. With that in mind and with all the
necessary prudence, I can say:
Yes, we have a the model and the technology in the lab
running as prototypes.
That really works:
https://www.researchgate.net/publication/316033070_Open_Historical_Data_Map_-_a_prototype/download
Code is free and open
https://github.com/OpenHistoricalDataMap
We have several prototypes of editors but implementing an OHDM
specific editor cannot be they way.
I'm looking for a way the integrate our stuff into OSM and hopefully
OHM. I claim, our data model is better. Our rendered map is far
from good. (no offense to my students who might read that list).
We have no clear plan how to deal with editors.
Our next steps are:
a) Active OSM archive: We plan store OSM changes every year. But
as a continuum. A changed geometry will not change the object id.
We are nearly finished with that.
b) Proceed working with parliament to offer an LOD source.
c) We have to take care of editors and that rendering. We need
at least a plan - hugest problem.
d) We use PostGIS - we can integrate nearly any standardized geometry.
We have not yet a Web (better REST) interface for that. We have no
user concept etc.
Next semester, we continue working on a) and b). We work slowly, but but
steady.
May we could join our efforts?
Best regards,
Thomas
PS: And as I can see: Our server is down... Semester break - maintenance
work. Will be online soon. I have to add this to my list:
e) Stable server infrastructure.
Am 05.02.19 um 22:25 schrieb Jeff Meyer:
> Yep - I think you've hit the nail on the head & it's a problem we've
> been pushing further down the road in order to focus on the time-enabled
> mapping problems.
>
> Clearly, you've been thinking about this more than we have (& have been
> solving it?).
>
> The crudest way we've thought about solving it (and we haven't thought
> through the problems this might create) is by linking to an "upstream"
> data source of some sort with unique identifiers.
>
> Working with LOD is a top priority, for sure... OHM would be pretty
> useless without connections to other, richer data sources.
>
> So, in short, yes, we know we need to tackle that problem at some point!
> We'll take a look at your work and hopefully, there will be a way to
> find a good fit!
>
> Regards,
> Jeff
>
>
> On Mon, Feb 4, 2019 at 7:29 AM Thomas Schwotzer
> <thomas.schwotzer at htw-berlin.de <mailto:thomas.schwotzer at htw-berlin.de>>
> wrote:
>
> Hello Jeff,
> thank you very much for the OHM insights.
>
> As you know, we play around with a similar project
> since 2015.
>
> https://github.com/OpenHistoricalDataMap
>
> We learned a big deal about working with huge
> data but also how GeoSPARQL could be used as bridge
> into the wonderful world of Linked Open Data.
>
> As I could see, that is also project of OHM.
>
> To make things short: I wonder, if you have any plans
> to change the OHM data model?
>
> In my eyes, there is one crucial problem:
>
> OSM does not separate between objects and their geometry.
> In OSM, that causes no problem - in OHM it will.
>
> I'm sure, you are aware of that issue: Objects can change.
> A trail can become a highway and vice versa. A fortress can
> be extended or partly destroyed. The object is not changed
> in any case but its features or geometry.
>
> We have split both concepts (object and geometry) in our project.
>
> https://github.com/OpenHistoricalDataMap/OHDM-Documentation/wiki/OHDM-Data-Model
>
> https://www.researchgate.net/publication/303818952_Open_Historical_Data_Map_OHDM_-_work_in_progress
>
> My office is in a building that used to be a factory a decade ago.
> Geometry hasn't changed. Feature set has.
>
> There are start_day and end_day tags in OHM. OHM would need a *copy*
> of the geometry to model my office and the previous factory building,
> though.
>
> Each OHM object would have it own id and the apparent relation
> between both would get lost. (Not to mention the wast of memory space)
> That's a real pity when it comes to semantic relations like LOD. A
> time line, a changing object in time is is of major interest in
> systems like OHM. That relation would get lost because each point
> in time would be represented by another OHM ID.
>
> Did I understand something wrong?
>
> I see two solutions:
>
> 1) The OHM model is to be changed in a way that this n-m
> relation is made explicit.
>
> Drawback: It would become incompatible with OSM. That
> is a huge drawback.
>
> 2) An OHM 2.0 database is produced. That database works with
> a changed data model but can produce the OHM format as usual.
>
> Actually, that is what I would like to do with our OHDM project:
> Using our data model but drop all efforts for building editors
> etc. We would ex- and import OHM data. But we could attach to
> our data model with e.g. Parliament
> (http://parliament.semwebcentral.org/) to offer a LOD source.
>
> I just wonder if you plan to tackle that issue.
>
> Best regards,
> Thomas
>
> PS: Have I mentioned that I still love strolling through
> old maps. I really like the idea having a running OHM full
> of data :)
>
> Am 29.12.2018 um 18:21 schrieb Jeff Meyer:
> > Hello Volkan!!
> >
> > Thanks for your great questions! It's very exciting to know there
> are
> > others working on similar / helpful projects.
> >
> > First, there are a few things we're working on for OHM that will
> make
> > the future OHM slightly different from what's available today.
> I'll try
> > to distinguish between the two where appropriate. Hopefully, this
> won't
> > be too confusing.
> >
> > 1. What are the exact goals of OHM ? Which of these have already
> been
> > implemented ? Which functionalities are planned for OHM ?
> > The goals of OHM are:
> > a. Create a cool online time machine where people can come and view
> > depictions of high- and low-level geographic information (maps!)
> > throughout history.
> > b. Create a common, open, shared repository for vector-based map
> > information that users can improve.
> > c. I'm sure there are other goals and purposes that others on
> this list
> > might add.
> >
> > 2. Which functionalities have already been implemented ?
> > Because we are borrowing most/much of our technology from
> OpenStreetMap
> > (OSM), almost anything OSM has implemented we can currently use.
> > These include:
> > - Map database
> > - Raster tile service for rendering maps
> > - API for import / export
> > - Editing tools, both in-browser and external
> >
> > 3. Is there a way to get the map data, like an API ?
> > Yes, a very basic API that takes a geo bounding box for both
> reading and
> > writing to the central database.
> >
> > 4. Can I import maps in OHM ?
> > Certainly, although you may need to convert your GIS files into OSM
> > formats and we ask that there be start_date and end_date and
> source tags
> > added to all your ways.
> >
> > 5. Can I filter the map for a specific year, like I would like to
> see a
> > map from 1907, if not, will this functionality implemented soon ?
> > Yes! Sometime in early 2019, we hope to make this more widely
> available,
> > but an early prototype is available here:
> >
> https://openhistoricalmap.github.io/openhistoricaltiles/datefilter/#14.00/40.78411/-119.21285/2008-08-15,2008-09-15/
> >
> > 6. What does the data model of OHM look like ?
> > We use the OSM data model, which you can learn more about at the OSM
> > wiki, but it loosely consists of basic GIS data types that have
> > associated key:value pairs.
> >
> > 7. How far is the development of OHM (in percent) ?
> > Hard to say & it depends on whether you'd include what OSM has
> already
> > built into that estimate. Regardless, we have a fair amount of
> > customization we'd like to do.
> > What are your specific needs?
> >
> > Thanks,
> > Jeff
> >
> >
> >
> > On Thu, Dec 27, 2018 at 8:40 AM "Volkan Özgen"
> <volkan.oezgen at web.de <mailto:volkan.oezgen at web.de>
> > <mailto:volkan.oezgen at web.de <mailto:volkan.oezgen at web.de>>> wrote:
> >
> > Hello Open Historical Mapper :),
> > my name is Volkan, I'm from Berlin, Germany and I currently study
> > applied computer science at HTW-Berlin (Hochschule für
> Technik und
> > Wirtschaft).
> > I'm currently working on my thesis and the title is "Usability of
> > Open Historical Map for Open Historical Data Map".
> > Open Historical Data Map is like OHM a open source project
> by Prof.
> > Dr. Thomas Schwotzer. He is my professor at the university.
> > The goal of OHDM is to create historical maps.
> > There are a few things I need to know about OHM for my thesis:
> > 1. What are the exact goals of OHM ? Which of these have
> already
> > been implemented ? Which functionalities are planned for OHM ?
> > 2. Which functionalities have already been implemented ?
> > 3. Is there a way to get the map data, like an API ?
> > 4. Can I import maps in OHM ?
> > 5. Can I filter the map for a specific year, like I would
> like to
> > see a map from 1907, if not, will this functionality
> implemented soon ?
> > 6. What does the data model of OHM look like ?
> > 7. How far is the development of OHM (in percent) ?
> > I know that's a lot of questions but I hope you guys can help
> me out :)
> > Wish you all a great day and merry christmas :)
> > Greetings from Germany
> > _______________________________________________
> > Historic mailing list
> > Historic at openstreetmap.org <mailto:Historic at openstreetmap.org>
> <mailto:Historic at openstreetmap.org <mailto:Historic at openstreetmap.org>>
> > https://lists.openstreetmap.org/listinfo/historic
> >
> >
> >
> > --
> > Jeff Meyer
> > 206-676-2347
> > osm: Open Historical Map (OHM)
> > <http://wiki.openstreetmap.org/wiki/Open_Historical_Map> / my OSM
> user
> > page <http://www.openstreetmap.org/user/jeffmeyer>
> > t: @OpenHistMap
> >
> >
> >
> >
> >
> > _______________________________________________
> > Historic mailing list
> > Historic at openstreetmap.org <mailto:Historic at openstreetmap.org>
> > https://lists.openstreetmap.org/listinfo/historic
> >
>
> --
> Prof. Dr.-Ing. Thomas Schwotzer
> Hochschule für Technik und Wirtschaft (HTW) Berlin
> University of Applied Sciences
> FB 4 | Angewandte Informatik | WH C 616
> Wilhelminenhofstraße 75A, 12459 Berlin
> Phone +49 30 5019-2604 | Fax +49 30 5019-482604
> Thomas.Schwotzer at htw-berlin.de
> <mailto:Thomas.Schwotzer at htw-berlin.de> | www.htw-berlin.de
> <http://www.htw-berlin.de>
>
>
>
> --
> Jeff Meyer
> 206-676-2347
> osm: Open Historical Map (OHM)
> <http://wiki.openstreetmap.org/wiki/Open_Historical_Map> / my OSM user
> page <http://www.openstreetmap.org/user/jeffmeyer>
> t: @OpenHistMap
>
>
>
>
--
Prof. Dr.-Ing. Thomas Schwotzer
Hochschule für Technik und Wirtschaft (HTW) Berlin
University of Applied Sciences
FB 4 | Angewandte Informatik | WH C 616
Wilhelminenhofstraße 75A, 12459 Berlin
Phone +49 30 5019-2604 | Fax +49 30 5019-482604
Thomas.Schwotzer at htw-berlin.de | www.htw-berlin.de
More information about the Historic
mailing list