Another progress report on my AI journey.
The month begins with my receiving Dr. Ivan Bratko’s Prolog book (2nd ed.). I saw the Advice Language section that he used to developed AL0. So I typed all the code example for AL0. However, I discovered a run time execution error. This probably due to the changes between true Edinburgh Prolog to ISO Prolog used by SWI Prolog compiler. As a result, I went to a site containing the ISO Prolog standard. I have found some interesting differences. In addition, I have reviewed three other sources for Expert Systems – Luger, U. Nilsson, and AMZI. Basically, these sources do not use IF condition THEN action approach of Dr. Bratko; instead they use the standard prolog syntax for the production/knowledge-based rules.
Regarding further with AL3, based on what Dr. Bratko wrote in his articles and his Prolog book, the programmer would have to develop the IF <condition> THEN <action> rules. But what if the machine could learn to develop these rules instead. Apparently this question has been answered by Dr. Eduardo F. Morales. He developed the PAL system, which is a first-order logic pattern analyzer which develop rules using Inductive Logic Programming – this is so cool. His articles answered my questions about machine learning in the field of chess. His articles are inspiring that I ordered a book on Inductive Logic Programming. However, before I read these articles, I was on google searching if anyone has done any work with Dr. Bratko’s Advice Language in recent years. I found a posting from 2000 in rec.games.chess in which the author was asking for any leads for the AL3 code since Dr. Bratko has not worked with it in years. In fact, he recommended for the author to contact a Mexican Professor who had AL3 running. The author mentioned an article from Jussi Tella regarding planning in chess where the AL3 and Paradise are discussed. Then a search for Bratko’s work in google lead to the Morales papers.
Because of my aforementioned about chess and machine learning, I was reading in Russell and Norving the section on Machine Learning and temporarily skipped the section on Uncertain Knowledge and Reasoning. My goal was to develop a learning agent or agents that could learn an environment and discover patterns in the environment. One idea was to use the wumpus world to show the predicate calculus and to convert the predicate calculus into logic programming syntax, and finally into prolog code. However, Larry Holder has completed a prolog version of the Wumpus World of Chapter 6.
Another item to inform is my paper called A Brute Force Approach to Solving the Knight’s Tour Problem Using Prolog was accepted by an International AI conference. This was a surprise since I did not expect to get selected – the conference acceptance rate was 27 percent. Much effort has been done to comply with the conference paper layout rules and format. Although I am entitled to use seven pages, the paper is only five pages in length including the one page appendix. A 20 minute presentation is required for the conference. Due date is May 12.
The most popular blog entries are as follows:
People are interested in searching for Wumpus World code.
This completes this month’s report.
