Research of Perception and Behavior in Architectural Space

Japanese


Rule Extracting in Space Composition Process

< Experiment of ASMT >

->

< Reproduction of space composition process using CAD >

->

< Analysis using ILP system >

# Input #
00Positive example e+E+,
00Negative Example e-E-,
00Background Knowledge BK

# Output #
00Classification rule R ,
00where R or BK |= e+
00and R or BK not|= e-

->

< Rules of space composition process >

Rule1:
placement(A) :- relation(A,irregular,right_face,B,C),
relation(C,semi_regular,right_face,D,E),
relation(E,i_attach,top,F,G).

Rule2:ハ
placement(A) :- angle(A,90),relation(A,close,none,B,C).
00000000000000:ハ


Title:

ANALYSIS OF ARCHITECTURAL SPACE COMPOSITION USING INDUCTIVE LOGIC PROGRAMMING

PDF file of this paper

Author(s):
NORITOSHI SUGIURA, SHIGEYUKI OKAZAKI

Publication/Publisher:
John S. Gero and Fay Sudweeks (eds.), Artificial Intelligence in Design '02, Kluwer Academic Publishers, July 2002

Abstract:
In this paper, we summarized an application of the ILP technique to the analysis of the human design process in architectural design. We focused particularly on discovering the patterns in the spatial composition process in the framework of design analysis called Architectural Space Montage Technique. Spatial composition data used in this paper are complexly structured data, which are collections of architectural objects with geometric relationships. The objects and relationships are constrained within a given set of attributes (such as object type, object angle, time stamp, relation type) together with background knowledge on the class hierarchies of object types and re-lation types. We modeled these spatial composition process data in first-order logic and applied the ILP system Progol to discover design patterns that characterized the given data.


Title:

Analysis of Spatial Composition process by Inductive Logic Programming
-Extracting Peculiar Rules in Process of Spatial Composition on Space Montage Technique-

PDF file of this paper

Author(s):
SUGIURA Noritoshi, OKAZAKI Shigeyuki

Publication/Publisher:
Journal of Archit. Plann. Environ. Engng. AIJ, No.546, Aug, 2001

Abstract:
The latent rules of space composition process in the experiment of Architectural Space Montage Technique (ASMT) was extracted by Inductive Logic Programming (ILP: one of machine learning methodology). The process of miniature disposition in ASMT is described as 5 attributes of miniatures: kind, direction, position, difference in order of disposition and geometric binomial relation, which is expressed in first order predicate logic. Four results of experiment of ASMT were analyzed by ILP. And the peculiar rules extracted in each experiment were L-shaped enclosure, composition with detached miniatures, lengthwise parallel, composition with mat, respectively.


Title:

System Extracting Rules in the Process of Spatial Composition on Architectural Space Montage Tehcnique

PDF file of this paper

Author(s):
SUGIURA Noritoshi, HOZUMI Teruaki, OKAZAKI Shigeyuki, YANAGISAWA Kazuhiko

Publication/Publisher:
Summaries of technical papers of Annual Meeting Architectural Institute of Japan, 1999

Abstract:
When man designs an architectural space, he refers some patterns about attributes of constituent elements and geometric relation between constituent elements. Usage of the patterns makes a style of the designer and spatial characteristics. As man learns the patterns through experience and training, it is difficult to express by words. In this paper, we applied a machine learning methodology called Inductive Logic Programming(ILP) and proposed a system that extracts latent patterns objectively from space composition process on Architectural Space Montage Technique. By this system, we can get some rules correspond to characteristics felt intuitively through clinical observation.


Title:

System Extracting Rules in the Process of Spatial Composition by Schizophrenic Patients on Architectural Space Montage Tehcnique

PDF file of this paper

Author(s):
SUGIURA Noritoshi, OKAZAKI Shigeyuki, YANAGISAWA Kazuhiko, HOZUMI Teruaki

Publication/Publisher:
Summaries of technical papers of Annual Meeting Architectural Institute of Japan, 1999

Abstract:
In this paper, we applied a machine learning methodology called Inductive Logic Programming(ILP) and proposed a system which extracts latent patterns objectively from space composition process on Architectural Space Montage Technique. We have conducted experiments with schizophrenic patients, and extracted some rules about placement of miniatures from composition process of two examples, which looks to be similar. We examined whether extracted rules express spatial characteristics common to both examples. We could get some rules by the system partly common to both examples. And from extracted rules, some latent differences between similar examples were found.


Title:

A method of description of spatial composition process and a method of extracting rules by Inductive Logic Programming
A study of spatial composition process with Architectural Space Montage Tehcnique Part1

PDF file of this paper

Author(s):
SUGIURA Noritoshi, OKAZAKI Shigeyuki, SUGAI Shigeyoshi

Publication/Publisher:
Summaries of technical papers of Annual Meeting Architectural Institute of Japan, 2000

Abstract:


Title:

Clustering works with Architectural Space Montage Tehcnique based on similarity of spatial composition process
A study of spatial composition process with Architectural Space Montage Tehcnique Part2

PDF file of this paper

Author(s):
SUGIURA Noritoshi, OKAZAKI Shigeyuki, SUGAI Shigeyoshi

Publication/Publisher:
Summaries of technical papers of Annual Meeting Architectural Institute of Japan, 2000

Abstract:


Title:

Analysis of Architectural Space Composition Using Inductive Logic Programming

PDF file of this paper (in japanese)

Author(s):
SUGIURA Noritoshi, OKAZAKI Shigeyuki

Publication/Publisher:
Summaries of technical papers of 45th Annual Meeting Japanese Society for Artificial Intelligence, 2001

Abstract:
  In this paper, we summarize an application of ILP technique to the analysis of human design process in the domain of architechtural design. We forcuse partcularly on the discovering patterns in spacial composition process in a context of an framework of design analysis, called Architectural Space Montage Technique. Spacial composition data used in this paper are complexly structured data, which are collections of architechtural objects with geometric relationships. We model these spacial composition process data in first-order logic and apply an ILP system Progol to discover design patterns that characterize the given data.


Index of Research


Acrobat Reader Download