SOFTCARDINALITY: Hierarchical Text Overlap for Student Response Analysis

In this paper we describe our system used to participate in the Student-Response-Analysis task-7 at SemEval 2013. This system is based on text overlap through the soft cardinality and a new mechanism for weight propagation. Although there are several official performance measures, taking into account the overall accuracy throughout the two availabe data sets (Beetle and SciEntsBank), our system ranked first in the 2 way classification task and second in the others. Furthermore, our system performs particularly well with “unseendomains” instances, which was the more challenging test set. This paper also describes another system that integrates this method with the lexical-overlap baseline provided by the task organizers obtaining better results than the best official results. We concluded that the soft cardinality method is a very competitive baseline for the automatic evaluation of student responses.