Dataflow visualization systems enable flexible visual data exploration byallowing the user to construct a dataflow diagram that composes query andvisualization modules to specify system functionality. However learningdataflow diagram usage presents overhead that often discourages the user. Inthis work we design FlowSense, a natural language interface for dataflowvisualization systems that utilizes state-of-the-art natural languageprocessing techniques to assist dataflow diagram construction. FlowSenseemploys a semantic parser with special utterance tagging and special utteranceplaceholders to generalize to different datasets and dataflow diagrams. Itexplicitly presents recognized dataset and diagram special utterances to theuser for dataflow context awareness. With FlowSense the user can expand andadjust dataflow diagrams more conveniently via plain English. We applyFlowSense to the VisFlow subset-flow visualization system to enhance itsusability. We evaluate FlowSense by one case study with domain experts on areal-world data analysis problem and a formal user study.