Intelligent Information and Communication Systems
English version deutsche Version|

Knowledge representation and acquisition in MultiNet

Ingo Glöckner (Ingo.Gloeckner@FernUni-Hagen.DE)

Introduction

The knowledge representation paradigm MultiNet has shown its suitability for the semantic representation of natural language in many years of use at the University of Hagen. The core of the MultiNet representation is formed by semantic networks, which can be represented by directed graphs with nodes representing concepts from the discourse area and edges expressing relations between these concepts. Nodes and edges are enriched with various sets of attributes which establish a multi-layered structure of the network (each attribute partitions the nodes into classes or "layers" of nodes with the same attribute value). This multi-layered organization of the paradigm is reflected by the name MultiNet.

The work reported on this page was aimed at providing methods for representing knowledge in MultiNet, and at developing tools to support knowledge engineering tasks on the represented knowledge.

A reference implementation for MultiNet

The programming library for building MultiNet applications was implemented in the Scheme programming language. It comprises the following functionality:

The knowledge engineering tool MWR

Figure 1: Graphical user interface of MWR. Please click for a guided tour
MWR-Screenshot

The knowledge engineering tool MWR (Multinet WissensRepräsentation, i.e. "MultiNet knowledge representation") was designed to aid the knowledge engineer in creating and editing MultiNet knowledge bases, and to simplify the development of applications in the MultiNet framework.

See the guided tour of the MWR system for screenshots of the graphical user interface which illustrate the main components of MWR and its use as a platform for application development.

References

[Publications of the IICS Group]


Valid ISO-HTML! The first version of MWR was implemented by Carsten Gnörlich. For more information on MWR, please contact Ingo Glöckner (Ingo.Gloeckner@FernUni-Hagen.DE).