Automated Essay Scoring in the Presence of Biased Ratings

Proceedings of NAACL-HLT 2018, pages 229–237 New Orleans, Louisiana, June 1 - 6, 2018. c⃝2018 Association for Computational Linguistics Automated Essay Scoring in the Presence of Biased Ratings Evelin Amorim Computer Science Department UFMG Brazil evelin@dcc.ufmg.br Marcia Canc¸ado Linguistics Department UFMG Brazil mcancado@ufmg.br Adriano Veloso Computer Science Department UFMG Brazil adrianov@dcc.ufmg.br Abstract Studies in Social Sciences have revealed that when people evaluate someone else, their eval- uations often reflect their biases. As a re- sult, rater bias may introduce highly subjec- tive factors that make their evaluations inaccu- rate. This may affect automated essay scoring models in many ways, as these models are typ- ically designed to model (potentially biased) essay raters. While there is sizeable literature on rater effects in general settings, it remains unknown how rater bias affects automated es- say scoring. To this end, we present a new annotated corpus containing essays and their respective scores. Different from existing cor- pora, our corpus also contains comments pro- vided by the raters in order to ground their scores. We present features to quantify rater bias based on their comments, and we found that rater bias plays an important role in auto- mated essay scoring. We investigated the ex- tent to which rater bias affects models based on hand-crafted features. Finally, we propose to rectify the training set by removing essays associated with potentially biased scores while learning the scoring model. 1 Introduction Automated Essay Scoring (AES) aims at develop- ing models that can grade essays automatically or with reduced involvement of human raters (Page, 1967). AES systems may rely not only on gram- mars, but also on more complex features such as semantics, discourse and pragmatics (Davis and Veloso, 2016; Song et al., 2014; Farra et al., 2015; Somasundaran et al., 2014). Thus, a prominent approach to AES is to learn scoring models from previously graded samples, by modeling the scor- ing process of human raters. When given the same set of essays to evaluate and enough graded sam- ples, AES systems tend to achieve high agreement levels with trained human raters (Taghipour and Ng, 2016). While research in AES has focused on design- ing scoring models that maximize the agreement with human raters(Chen and He, 2013; Alikan- iotis et al., 2016), there is a lack of discussion on how biased are human ratings. Despite mak- ing judgments on a common dimension, raters may be influenced by their attitudes, their cul- tural background, and their political and economic views (Guerra et al., 2011). Since AES models are designed to learn by analyzing human-graded essays, AES models could inherit rating biases present in the scores from human raters, and this may result in systematic errors. Thus, our ob- jective in this paper is to examine the extent to which rater bias affects the effectiveness of state- of-the-art AES models. A deeper understanding of such factors may help mitigating the effects of rater bias, enabling AES models to achieve greater objectivity. In order to study the effects of rater bias …

Content truncated to 500 words