placeholder image

Modeling Graph-of-Graphs for Solving Document Categorization problems.

Grant


Scheme


  • Discovery Projects

Abstract


  • Dependencies among documents, and the document structure can contain information which is essential for solving document categorization problems. Existing methods are unable to encode more than one type of structure at a time. This causes limitations to the classification accuracy that can be obtained by existing methods. This project proposes data driven modelling techniques capable of encoding graphs which contain nodes that are labelled by other graphs. This is referred to as the Graph-of-Graphs (GoG) model. We will achieve this by building on the Graph Neural Network method which we will adapt to allow a stage based learning of GoG models. It will be shown that real world problems can be solved by the proposed approach.

Date/time Interval


  • 2011 - 2013

Sponsor Award Id


  • DP110103376

Local Award Id


  • 11834

Scheme


  • Discovery Projects

Abstract


  • Dependencies among documents, and the document structure can contain information which is essential for solving document categorization problems. Existing methods are unable to encode more than one type of structure at a time. This causes limitations to the classification accuracy that can be obtained by existing methods. This project proposes data driven modelling techniques capable of encoding graphs which contain nodes that are labelled by other graphs. This is referred to as the Graph-of-Graphs (GoG) model. We will achieve this by building on the Graph Neural Network method which we will adapt to allow a stage based learning of GoG models. It will be shown that real world problems can be solved by the proposed approach.

Date/time Interval


  • 2011 - 2013

Sponsor Award Id


  • DP110103376

Local Award Id


  • 11834