Paramenides : web-based expert system for problem solving and diagnosis

Abstract

Imagine a world where human expertise can be stored in a database and can easily be accessed by anyone through the Internet. Knowledge has always fascinated humans. Humans that possess vast knowledge in a domain are extremely valuable and experts in certain domains are quite scarce. However, human experts have various limitations:- they are not permanent (they die, retire), they can only service a limited number of users, they suffer from moods, and cannot always explain their reasoning. An expert system is a computer program whose main aim is to serve as the role of an expert. Similar to an expert, it asks questions related to a goal defined by the user, in order to reach a conclusion. It has various benefits:- It can be reproduced easily, can service a very large number of users, can clearly explain its reasoning and is permanent. The knowledge contained by an expert is stored in a knowledge base as a set of rules, based on an ontology. The expert system’s job is to reason about the stored knowledge and come up with conclusions as an expert would. It also takes care of uncertainty elements in the reasoning, and be able to produce results based on their expected belief. The aim of Paramenides is to bring expert systems to the public and create a globally accessible online resource for expert knowledge and problem solving. The main objective is to create a web-based expert system (www.paramenides.com) that is accessible publicly via the Internet where both users and experts can meet to share knowledge. Experts can define knowledge bases which are mapped to their domain of expertise, while the expert system can help users in solving/diagnosing problems. This is all done through an easy-to-use interface that does not require the users to be particularly computer-savvy to use. Paramenides works by using an inference engine that implements both backward chaining and forward chaining modules. It works on knowledge bases that are created by users themselves. A knowledge base consists of a series of IF ... THEN rules, which are then used for problem solving and diagnosis. Backward chaining involves setting a goal and working backwards in order to reach a conclusion. Forward chaining involves processing facts inputted by the user to find out what can be the problem. It also uses certainty factors to allow elements of uncertainty in its reasoning and thus be able to provide a list of reasons based on the sorted by their most probable outcome. An explanation module allows the system to provide a reasonable explanation on how it came to such a conclusion, in a visual manner. This allows the user to easily follow its reasoning, thus increasing its credibility. The knowledge bases defined in the expert system can be created by the users. Thus it is difficult to achieve test results on all possible knowledge bases. However, if correct rules are defined in the knowledge base, the system mimics an expert’s reasoning with 100% similarity. The results are quite subjective since even in real-life, reasoning and conclusion can vary greatly from one expert to another.