Name (abbrev)

Name (full)

Category

Last update

Spatial Layout Data

Miscellaneous

b D, Y

Application domain

Further specifications

Spatial Layout Data for GKS System

data + background knowledge

Type

Format

Complexity

ILP

Prolog

174 positive and 214 negative examples, 1500 clauses of background knowledge

WWW / FTP

ftp://mizo01.ia.noda.sut.ac.jp/incoming/Floor_data.tar

Contact person(s)

Related group(s)

Optional contact address

{mizo,ohwada}@ia.noda.sut.ac.jp

References

Mizoguchi, Fumio and Ohwada, Hayato (1995)
"Using Inductive Logic Programming for Constraint
Acquisition in Constraint-based Problem Solving"
Proceedings of the ILP Conference, K.U. Leuven 1995, pp. 297-322

Annotations

The examples provide symbolic and numerical information about spatial layout of Japanese houses, e.g. position and size of different rooms, their spatial relations or the total size of the house. There are two types of background knowledge - one of them allows to transform the numerical information into a qualitative form. The goal is to induce rules which can be regarded as constrained clauses containing both symbolic and numerical information, e.g. "If the space X is in the interval [Value1,Value2], it is concluded with the likelihood P% that the living room is adjacent to the dining room." This dataset has been generated to test GKS [MizoFumOhw95]: an ILP system for induction of constraint logic programs.

Comments