Yang proposes a new graphical method based on deduction graphs to solve dependency, rule, and fact inference problems in relational databases, rule-based expert systems, logic programming, and general AI. The paper presents an algorithm for constructing a deduction graph and investigates some properties of the deduction graph.
The author demonstrates two simple applications of deduction graphs. The first application is an improvement in handling extraneous, superfluous, and redundant entities in relational databases. The second application is a technique for rule and fact inferences in function-free Horn clauses.
This paper does not develop more realistic applications of the method. The author himself admits that the use of deduction graphs is not as powerful a technique as, for example, the resolution approach, and that the application of deduction graphs to less restrictive classes of clauses needs further research.
This paper is a research note in content and style. The style is concise and the language is sometimes difficult to follow. Researchers working in relational databases should find some ideas here, but the benefit to logic programming and general AI audiences is very limited at this stage.