Identifying Predictive Structures in Relational Data Using Multiple Instance Learning
Amy McGovern, David D. Jensen · 2003
This paper introduces an approach for identify-ing predictive structures in relational data using the multiple-instance framework. By a predictive structure, we mean a structure that can explain a given labeling of the data and can predict labels of unseen data. Multiple-instance learning has previously only been applied to flat, or proposi-tional, data and we present a modification to the framework that allows multiple-instance tech-niques to be used on relational data. We present experimental results using a relational modifica-tion of the diverse density method (Maron, 1998; Maron & Lozano-Pérez, 1998) and of a method based on the chi-squared statistic (McGovern & Jensen, 2003). We demonstrate that multiple-instance learning can be used to identify predic-tive structures on both a small illustrative data set and the Internet Movie Database. We compare the classification results to a k-nearest neighbor approach. 1.