A Corpus of Commonsense Knowledge about Solid Objects

Ernest Davis
Department of Computer Science
Courant Institute of Mathematical Sciences
New York University

This web-site is a corpus of commonsense knowledge about simple systems of solid objects.

Structure in the Large

The corpus is organizes as a collection of web pages written in HTML. The web pages are divided into four categories: (Pages marked in brackets are under construction, but not included in the current version.)

The home page (here).
Foundational issues: Ontology, Geometry, [Syntax and logic], Probability.
Higher-level theories: Fitting relations, Holes, Kinematics, Dynamics, [Abstraction, Approximation, and MetaRules].
Domain theories: Boxes, Rings, Heaps, [Doors], [Deadbolts], [Desks].

For stable reference, successive, comparatively presentable, states of the corpus will be archived. The default corpus (i.e. here) is the most recent archived version. Currently, no previous versions are yet archived. We also link the the active version under construction

Structure in the Small

The web pages consist mostly of a sequence of definitions and facts. Each of these has a number of fields. The key fields of a definition are an English language description; a declaration of new symbols and their sorts; and, when possible, a formal representation of the definition.

The key fields of a fact are the English language statement, the formal representation, and the following features:

The theory the fact belongs to: Topological, geometric (with or without a temporal component); kinematics; statics; or dynamics.
Whether it is deterministic (the default) or probabilistic.
Whether the objects involved in fact are individually enumerated (the default) or whether the fact involves sets of object of indeterminate cardinality.
Whether the fact involves an external agent or does not (the default).

Other fields for both definitions and facts include diagrams, comments, and citations.

Publications

Commonsense Knowledge about Boxes and Heaps (PDF)\ Submitted to KR-04.

Bibliography

This research has been supported by NSF grant IIS-0097537.