SKESD
Structured Knowledge: Entry, Search and Display

The aim of the project is to create a cross between a Wiki and an AI knowledge representation system.

Like a Wiki it will have a web interface and support multiple users editing the data it stores. Like (at least some) AI knowledge representation systems that data will consist of an ontology (way of splitting up the world) and a set of frames which represent the relationships which exist between the sets and instances defined in the ontology.

The data so entered can be used either by computer programs as they are in current knowedlge representation systems or (at least to me, more interestingly) as a means of communicating between people.

Use by people will be supported by the definition of transforms which will convert the internal frame/set/instance representation into natural language expressions. At least initially it is expected that these expressions will be "unnatural" and "robotic". However if the transform to a particular natural language for a particular frame type, set or instance has been defined, any use of that object type can be translated to that natural language. I expect that automatic translation into multiple natural languages will be an acceptable trade-off for lack of fluency.

If typical member of set
Variable X1
Set Mice
Predicate
Weight (kilos)
Object X1
Number 0.2
  If [X1] is a typical member of set [Mice] then
  [X1] weighs [0.2 Kilos]

SKESD @ SourceForge.net Logo