Master IAD Project - Knowledge Representation and Reasoning
Academic Year 2025-2026

Français | English

This project implements a solver for Argumentation Systems (AS). It computes and verifies different types of extensions according to preferred (PR) and stable (ST) semantics.

An argumentation framework is defined as F = ⟨A, R⟩ where:

• A is a set of abstract arguments
• R ⊆ A ×ばつ A is the attack relation between arguments

For more detailed information, please refer to the detailed report.

The program solves the following 6 problems:

• Python 3.8 or higher
• No external dependencies required (standard library only)

python3 --version

python3 program.py -p <PROBLEM> -f <FILE> -a <ARGUMENTS>

• -p: Problem type (VE-PR, DC-PR, DS-PR, VE-ST, DC-ST, DS-ST)
• -f: Path to the .apx file containing the argumentation framework
• -a: Query arguments
  • For VE-* problems: comma-separated list (e.g., a,c,d)
  • For DC-* and DS-* problems: single argument (e.g., b)

Assuming af.txt contains an AF with A = {a,b,c,d} and R = {(a,b), (b,c), (b,d)}:

# Verify if {a,c,d} is a preferred extension
python3 program.py -p VE-PR -f af.txt -a a,c,d
# Output: YES
# Check credulous acceptance of 'b' (preferred semantics)
python3 program.py -p DC-PR -f af.txt -a b
# Output: NO
# Check skeptical acceptance of 'a' (preferred semantics)
python3 program.py -p DS-PR -f af.txt -a a
# Output: YES
# Verify if {a,c,d} is a stable extension
python3 program.py -p VE-ST -f af.txt -a a,c,d
# Output: YES

Argumentation-Project/
├── README.md 
├── README.fr.md # French documentation
├── program.py # Main entry point
├── src/
│ ├── __init__.py
│ ├── cli.py # Command-line argument handling
│ ├── apx_parser.py # Parser for .apx files
│ ├── systeme_argumentation.py # Argumentation system class
│ ├── semantics.py # Semantics algorithms
│ └── queries.py # Query resolution
└── Fichiers-tests/ # Test files
 ├── test_af1.apx
 ├── test_af1_pr.txt
 ├── test_af1_st.txt
 └── ...

The .apx files follow this format:

arg(argument_name).
att(source_argument,target_argument).

• Each argument must be declared with arg() before being used in an attack
• No spaces in lines
• Argument names can contain letters, numbers, and _ (except arg and att which are reserved)

arg(a).
arg(b).
arg(c).
arg(d).
att(a,b).
att(b,c).
att(b,d).

This example represents the following graph:

a → b → c
 ↓
 d

The Fichiers-tests/ directory contains several test cases:

# Test with provided files
python3 program.py -p VE-PR -f Fichiers-tests/test_af1.apx -a a,c,d
python3 program.py -p DC-ST -f Fichiers-tests/test_af2.apx -a b

The solver uses a backtracking algorithm to compute all conflict-free, admissible sets, and then identifies maximal admissible sets (preferred extensions).

A stable extension is computed by finding conflict-free sets that attack all arguments outside the set.

• VE (Verification): Checks if the given set is an extension under the specified semantics
• DC (Credulous): Returns YES if the argument appears in at least one extension
• DS (Skeptical): Returns YES if the argument appears in all extensions

@thmsgo18 - Thomas Gourmelen