I'm taking my first AI course this semester at uni and we were asked to do a project on any AI topic that we find interesting and to impliment the code in "Prolog". One example is Implementing 4 or 5 search algorithms and comparing the effeciency in different selected special cases. I just need a topic in the same level. plz HELP!
ps. A*, Sales Man, Search Akgorithms are done by other students ..
I really enjoyed learning, implementing and tweaking A* and its variants (manhattan distance, etc). I'm using it in a game to do path finding on a node based map, with simple weight calculations.
I'd recommend you pick something that will keep your interest so you can do a stellar job on the project!
Good luck!
Prolog is really good at finding solutions to problems. Maybe you could write a solver for Sudoku, or for other common logic puzzles?
If Prolog is the language, I doubt that search algorithms are the best choice. To really use the powers of Prolog, you should write a knowledge engine - i.e. a program into which you insert data and it makes conclusions from that data.
SICP has a great section about this: 4.4 - Logic Programming. Their style is "lispy" since it's their own Prolog implementation, but you could easily adapt that to standard prolog. Building upon their example and extending it is a great project.
Try doing a simple SHRDLU style parser and world model of some sort.
I think it really depends on what you're interested in otherwise, after a few (inevitable) hurdles, you'll give up and lose any motivation. Is there a particular area of AI that interests you? How about a system which solves games such as Sudoku? Or one that interacts with users (Chess, Checkers etc..)
How about implementing a Simple Genetic Algorithm (SGA)? It is well described in Goldberg's seminal work. For now, you could just implement a simple Binary Coded GA, and demonstrate its working on some simple search/optimization problems. If you want code reference, you may take a look at the Pascal source in the above book.
Mitchell's work 'Introduction to Genetic Algorithms' is also highly recommended for getting the basic ideas.
How about doing something in AI planning or Robotics?
For instance you can do a project to program the Robot Arm to move certain Blocks from one configuration to the other with certian constraints..Also called the Block World Problem.
Implementing TicTacToe is a classic AI homework or my choice would be a trading bot according to financial portfolio momentum, stoploss rulebased reasoning and learning from mistakes and successes written in your preferred lagnguage.
Automated theorem proving would be a good bet. it is a subset of automated reasoning, a very interesting, and very cool part of AI (at least in my opinion). You could start just by proving classical propositional logic statements, this can then be used as a basis for proving higher order logic statements.