A Topic-Independent Method for Automatically Scoring Essay Content Rivaling Topic-Dependent Methods

Abstract

One of the major drawbacks to the conventional methods is that they are topic-dependent. They require a great number of human-reader scored essays that are written to the same essay topic as the target essay. This means that for every new essay topic, one has to collect a great number of essays that are written to the new essay topic and has to score them, which is costly and timeconsuming.This paper proposes a topic-independent method for automatically scoring essay content. Unlike conventional topic-dependent methods, it predicts the human score of a given essay without training essays written to the same topic as the target essay. To achieve this, this paper introduces a new measure called MIDF that measures how important and relevant a word is in a given essay. The proposed method predicts the score relying on the distribution of MIDF. Surprisingly, experiments show that the proposed method achieves an accuracy of 0.848 and performs as well as or even better than conventional topic-dependent methods.To solve the drawback, this paper proposes a topicindependent method for automatically scoring essay content. Here, topic-independent means that once obtained a set of human-reader scored essays written to a topic, one can apply the proposed method to new essays written to any essay topic. Surprisingly, it rivals the conventional topic-dependent methods as experiments will show later. To achieve this, this paper introduces a new measure called MIDF that measures how important and relevant a word is in a given essay. The proposed method predicts the score relying on the distribution of MIDF.