Is there a way in Bonsai to auto classify elements based on the properties and their predefined values defined in a bsdd dictionary? f.e. Wall (IfcWall) + IsExternal=true > "exterior wall"
Hi,
That’s interesting, I was thinking about that a few months ago but drop it. I was willing to deduce class on BSDD to avoid describing all combinations.
My main idea was to convert the BSDD shema into a RDF file:
Class will be class in RDF
Property will be property in RDF (I don’t know if it’s the best idea)
BIM object would be instance in RDF
This can work but it still need to solve a few challenge
1. find the best way to convert BSDD to RDF
2. Find a good query engine
3. Find a way to deal with error in logic
What if you found inconsistencies in your BSDD ?
BSDD can give you range of property, this can be an issue to manage logic.
4. How to deal with un unnecessary property ?
Your wall will come the 100 parameters and only 3 or 4 of them will matters with classification.
We have IDS to control model, but there is still a uncanny thing between IDS and BSDD. Maybe that’s not the best place to use it.
Thank you both for your response @Moult and @Elobo . It took me some time to set up a test dictionary on bSDD, but now I think I can explain my intention better:
The test dictionary is meant to represent our government standard for tendering of concrete building elements in a machine-readable version. Every class in the dictionary (in my demo, just one) defines a cost item e.g. "all concrete walls with the following characteristics: Width ≤ 20 cm, StrengthClass = C25/30 and Height ≤ 3.20 m" and includes all required properties and their valid values (ranges or allowed values) for filtering.
An IFC file in that project phase normally already carries all the planning properties needed for filtering and quantity takeoff according to that standard. Therefore, in bonsai, it would be great to automatically classify all walls that meet these requirements. Afterwards it would be easy to filter for the classified elements and sum up e.g. the gross volume of these elements.
At the moment, we use the same process in Solibri to check the tendering documents with its classification/filter feature. The advantages of a bSDD-based process would be the ability to use the classified IFC file for documentation and handover to tenderer - and that the fundamentals (filters) would be publicly available through the bSDD and traceable.
I think there are also other usecases where an autoclassification feature could be very useful.
@Rio
wow, quite an impressive job by your government, very detailed.
As posted above having a universal tool to automagically convert elements' attributes/properties & qto to any dictionary part of bSDD sounds like a huge task due to the variables involved.
My 2 cents is that, maybe, it would be approachable if a dictionary could be mapped into criteria used to select elements, from there assigning relevant classification could be a matter of a short script?
I understand that your wall 070201H is defined by 3 properties.
What about others properties that are usefull ? You may need to have the color of the wall, but you will use the same class.
How do you differentiate property that define versus property that describe the object ?
I see that we can use the PropertyType (Property or Depencency). The property would be descriptive and Dependency would be defining.
I didn’t find documentation on that, and I can only guess what dependency mean.
@Elobo in my bSDD dictionary every class (like 070201H) would only contain properties I need for identifying the corresponding class. I also marked these properties in bSDD as "Dependency" (I only guessed its meaning too). All other interesting properties could be required somewhere else - that dictionary is only intended for classification according to "LBHB LG07 - standardisierte Leistungsbeschreibung". https://github.com/buildingSMART/bSDD/blob/master/Documentation/bSDD JSON import model.md
Comments
Nope, but that's a great idea. It'll need some logic for dealing with ambiguous classifications though.
Hi,
That’s interesting, I was thinking about that a few months ago but drop it. I was willing to deduce class on BSDD to avoid describing all combinations.
My main idea was to convert the BSDD shema into a RDF file:
This can work but it still need to solve a few challenge
1. find the best way to convert BSDD to RDF
2. Find a good query engine
3. Find a way to deal with error in logic
What if you found inconsistencies in your BSDD ?
BSDD can give you range of property, this can be an issue to manage logic.
4. How to deal with un unnecessary property ?
Your wall will come the 100 parameters and only 3 or 4 of them will matters with classification.
We have IDS to control model, but there is still a uncanny thing between IDS and BSDD. Maybe that’s not the best place to use it.
Thank you both for your response @Moult and @Elobo . It took me some time to set up a test dictionary on bSDD, but now I think I can explain my intention better:
https://identifier.buildingsmart.org/uri/DEMO/LG07/1.0
The test dictionary is meant to represent our government standard for tendering of concrete building elements in a machine-readable version. Every class in the dictionary (in my demo, just one) defines a cost item e.g. "all concrete walls with the following characteristics: Width ≤ 20 cm, StrengthClass = C25/30 and Height ≤ 3.20 m" and includes all required properties and their valid values (ranges or allowed values) for filtering.
An IFC file in that project phase normally already carries all the planning properties needed for filtering and quantity takeoff according to that standard. Therefore, in bonsai, it would be great to automatically classify all walls that meet these requirements. Afterwards it would be easy to filter for the classified elements and sum up e.g. the gross volume of these elements.
At the moment, we use the same process in Solibri to check the tendering documents with its classification/filter feature. The advantages of a bSDD-based process would be the ability to use the classified IFC file for documentation and handover to tenderer - and that the fundamentals (filters) would be publicly available through the bSDD and traceable.
I think there are also other usecases where an autoclassification feature could be very useful.
@Rio
wow, quite an impressive job by your government, very detailed.
As posted above having a universal tool to automagically convert elements' attributes/properties & qto to any dictionary part of bSDD sounds like a huge task due to the variables involved.
My 2 cents is that, maybe, it would be approachable if a dictionary could be mapped into criteria used to select elements, from there assigning relevant classification could be a matter of a short script?
As @Elobo rightly put it
Very interesting challenge though ;)
I understand that your wall 070201H is defined by 3 properties.
What about others properties that are usefull ? You may need to have the color of the wall, but you will use the same class.
How do you differentiate property that define versus property that describe the object ?
I see that we can use the PropertyType (Property or Depencency). The property would be descriptive and Dependency would be defining.
I didn’t find documentation on that, and I can only guess what dependency mean.
@Elobo in my bSDD dictionary every class (like 070201H) would only contain properties I need for identifying the corresponding class. I also marked these properties in bSDD as "Dependency" (I only guessed its meaning too). All other interesting properties could be required somewhere else - that dictionary is only intended for classification according to "LBHB LG07 - standardisierte Leistungsbeschreibung".
https://github.com/buildingSMART/bSDD/blob/master/Documentation/bSDD JSON import model.md