A joint-based paradigm identifies the triggers and arguments according to candidate triggers and entities. From there, trigger classification and argument role classification are realized simultaneously, which helps minimize error propagation