1: \begin{abstract}

2: Ranked data appear in many different applications, including voting and consumer surveys.

3: There often exhibits a situation in which data are partially ranked.

4: Partially ranked data is thought of as missing data.

5: This paper addresses parameter estimation for partially ranked data under a (possibly) non-ignorable missing mechanism.

6: We propose estimators for both complete rankings and missing mechanisms together with a simple estimation procedure.

7: Our estimation procedure leverages a graph regularization in conjunction with the Expectation-Maximization algorithm.

8: Our estimation procedure is theoretically guaranteed to have the convergence properties.

9: We reduce a modeling bias by allowing a non-ignorable missing mechanism.

10: In addition, we avoid the inherent complexity within a non-ignorable missing mechanism by introducing a graph regularization.

11: The experimental results demonstrate that the proposed estimators work well under  non-ignorable missing mechanisms.

12: \end{abstract}

13: