Three Java software components were built in ACL for the ASPIC project corresponding to three distinct aspects of the project : inference, decision and dialogue. These components can be downloaded from this page in separate zip files. Downloading these files indicates accepting the following disclaimer. As well as accepting the disclaimer, would you please register an interest in the components, if you havent already done so.
Each zip file contains:
- a README file,
- a single "bundled" Java jar that contains all relevant java classes in a single file,
- javadocs in the /javadoc directory,
- example files and assorted documentation in the /etc directory,
- a Java compiled jar and it's dependency jars in the /lib directory,
- a copy of the software disclaimer.
The Inference Engine is a software implementation of the ASPIC agreed argument game algorithms for defining the status of arguments. It provides a structure to capture defeasible knowledge and reason over the status of query matches (defeated or undefeated). The component allows a user to examine the yes (undefeated) or no (defeated) status of each match or to view a graph visualisation of the proof argument network associated with the reasoning argument game. It also provides a machine readable version of the proof and results via AIFXML.
The Decision Component is a software implementation of the ASPIC decision model that consumes the ASPIC Argumentation Engine. It has no GUI but does have a command line interface.
The Dialogue Component allows users to mediate an argumentation based persuasion dialogue between two partners. The component includes a GUI demo of the default dialogue protocol and was successfully demonstrated in the CARREL large scale demonstrator.
ArgKit is an open source follow-on project that builds on the lessons learnt from the ASPIC project. It's main module, Dungine, is a cut down version of the Inference Engine. It uses similar argumentation algorithms but makes a clean separation between the argument valuation and the argument generation. Dungine does not yet have a system for generating the tree arguments that the inference engine does.