Description
For this project, you will design and build a knowledge-based intelligent system that collects
user preferences and reasons about them.
1 Requirements
1. The system should have an easy-to-use GUI (using the Python Tkinter module1
) for
collecting names of attributes and their values, hard constraints, and preferences. The
system should also allow for reading in these input from files. (See section 3 for formats
of these files.)
• Attributes (A) in this project are going to be binary.
• Hard constraints (H) are represented as propositional formulas in the Conjunctional Normal Form (CNF).
• The system should support preferences (T) in the preference languages we discussed in class: Penalty Logic and Possibilistic Logic. Formulas involved in the
preference theories are of CNF as well.
2. The system should support the following reasoning tasks:
• Existence of feasible objects: decide whether there are feasible objects w.r.t H,
that is, whether there are models of H that are truth assignments making H true.
• Exemplification: generate, if possible, two random feasible objects, and show the
preference between the two (strict preference, equivalence, or incomparable).
• Optimization: find an optimal object w.r.t T.
• Omni-optimization: find all optimal objects w.r.t T.
3. The system should take advantage of the clasp system, a SAT solver that takes a propositional formula in CNF and computes its models. It can be used to compute feasible
objects for H, check if a truth assignment satisfies a formula, etc. A short tutorial will
be posted shortly.
4. For testing, the system should solve an instance, developed by you, that contains at
least 6 hard constraints and at least 6 preference rules over at least 8 attributes. Also
use this instance when demonstrating your system.
1See https://pythonspot.com/tag/tkinter/ and https://www.python-course.eu/
tkinter_labels.php for helpful references.
5. By Mar. 16, you will need to meet me to discuss the progress. You will make individual
appointments with me by email. Failure of this will result in deduction in the project
grade.
2 Deliverables
Zip the following to name [your-last-name] Project3.zip and submit to Canvas.
1. A text file with description of the instance (attributes and their values, hard constraints,
and preferences) you used for testing.
2. A directory that contains all your source codes.
3. A README file that contains instructions to build and run your system.
4. A PDF report that describes how your system works and shows the testing results using
the test instance (e.g., screen shots of various steps).
3 File Formats
3.1 Attributes File
appetizer: soup, salad
entree: beef, fish
drink: beer, wine
dissert: cake, ice-cream
…
3.2 Hard Constraints File
NOT soup OR NOT beer
NOT soup OR NOT wine
…
3.3 Preferences File (Penalty Logic)
fish AND wine, 10
wine OR cake, 6
…
