Traditional information retrieval systems based on Boolean logic suffer from two inherent problems: (1) inaccurate or incomplete query representation, and (2) inconsistent indexing; While many researchers have demonstrated that neural networks can solve the incomplete query problems for information retrieval, the inconsistent indexing problem still remains unsolved. In this paper, we present a hybrid methodology of integrating an inductive learning technique with a neural network (connectionist model) in order to solve both inconsistent indexing and incomplete query problems. Since an inductive learning technique has the ability to identify the most significant document index terms with various levels of relationship to their semantic significance, it provides a possible solution to the problem of inconsistent indexing. This paper reports the first phase of research that demonstrates how a neural network augmented by an inductive learning technique results in effective information retrieval performance in the areas that demand flexible inferencing and reasoning when incomplete queries and inconsistent indexing problems are present.