Abstract
Interactive gender inference aims to infer the gen- ders of the two involved users in a communication from the interactive text.In this paper,we address this task by proposing a joint inference approach which well incorporates label correlations among the instances.Specifically,an Integer Linear Pro- gramming(ILP)approach is proposed to achieve global optimization with various kinds ofintra-task and extra-task constraints.Empirical studies demon- strate the effectiveness of the proposed ILP-based approach to interactive gender inference