Background Exploration of the Adverse Event Reporting System (AERS) data by a wide scientific community is limited due to several factors. First, AERS data must be intensively preprocessed to be converted into analyzable format. Second, application of the currently accepted disproportional reporting measures results in false positive signals. Methods We proposed a data mining strategy to improve hypothesis generation with respect to potential associations. Results By numerous examples, we illustrate that our strategy controls the false positive signals. We implemented a free online tool, AERS spider (www.chemoprofiling.org/AERS). Conclusions We believe that AERS spider would be a valuable tool for drug safety experts.