A Consistency-Based Paradigm for Belief Change


COBA 2.0 is a consistency-based belief change system based on the belief change algorithms developed by James Delgrande and Torsten Schaub.

The COBA 2.0 applet (which opens in a new browser window) was written by Daphne Liu.


Instructions for Using the Applet

The COBA 2.0 applet provides a menu-based graphical user interface that allows you to easily navigate to any one of its three screens, import sentences from files, save output to files, and view the COBA 2.0 user manual and javadoc documentation.

The first screen is the main user interface through which you may (1) add sentences to, or remove sentences from, the knowledge base(s), the revision list, and the contraction list; (2) change the settings for the type of belief change, the search algorithm for equivalence sets, and the type of merging; (3) carry out parallel change on a belief change scenario, and subsequently reject or commit a resultant knowledge base.

The second screen is the integrity constraints (IC) interface through which you may add sentences to, or remove sentences from, the entailment-based (EB) IC list and the consistency-based (CB) IC list.

The third and final screen presents three lists of older snapshots you may revert to. You will find the log of changes for three lists: (1) KB: the removals, replacements, and changes committed; (2) EB IC and (3) CB IC: the removals and replacements made.

The syntax of the input sentences is strict. A top-level sentence with a connective operator & (and), + (or), > (implies), or = (is equivalent to) must be enclosed in parentheses. Parentheses inside a sentence, however, are optional and are used only to enforce parsing preference. For example, (a&b+c) is a valid input sentence and is different from (a&(b+c)), whereas a top-level sentence like a&b is syntactically malformed. It is recommended that you take a look at the COBA 2.0 user manual to familiarize yourself with the syntax of input sentences.

Once sentences have been added to a list, you can use the radio buttons to view them in three formats:

  1. Default: This is the same format in which you must enter them, with strict parenthesizing.

  2. CNF: Conjunctive Normal Form.

  3. DNF: Disjunctive Normal Form.

When you have set the parameters - knowledge base(s), revision list, contraction list, EB IC list, and CB IC list – for a belief change scenario you'd like to test, click the “Get Parallel Change” button to preview the resultant knowledge base(s). Next you may either reject the resultant knowledge base(s) to keep the current knowledge base(s), or commit a resultant knowledge base to replace the current knowledge base(s). Then you can start over with your next belief change scenario.


User Manual
Look at the COBA 2.0 user manual for compilation, installation, and execution instructions.

Javadocs
Look at the COBA 2.0 javadoc documentation for the coba UI and the revision logic packages.

Implementation Documentation
Look at the COBA 2.0 implementation documentation for high-level pseudo-code and data structures used.

Download
All of the class files and source files are available as one zip file. Feel free to take a look at it and make improvements.

Read the paper
The belief change algorithms underlying COBA 2.0 are described in a paper written by James Delgrande and Torsten Schaub.


Last updated January 2006
Daphne Liu