A bio text mining workbench combines natural language processing tools and machine learning methods to extract biological information from texts. It includes tools for named entity recognition of genes, proteins and other entities. It also extracts biological events involving interactions between entities. An active learning approach is used to improve the named entity recognition by selecting the most informative texts for human annotation. This helps enhance the model with less human effort. The workbench provides interfaces for corpus management, annotation and evaluation of the named entity and event extraction systems.