Developing an Automated Response Characterization Tool Based on Affect, Content, and Linguistic Features of the Casper Test
Open-ended questions are where respondents consider hypothetical scenarios that they might encounter in real life and describe what action(s) they might take in response to each situation. To assess a test-taker’s responses to these questions, a human scoring process (i.e., human raters marking each response individually) is followed. Despite following clear scoring criteria, human raters are likely to take additional factors (e.g., word choices) into account when scoring open-ended questions. Therefore, to what extent how someone writes influence the scoring needs to be carefully evaluated. Fortunately, new techniques and tools in natural language processing (NLP) allows researchers to analyze large amounts of textual data efficiently and achieve state-of-the-art results. This project aims to develop a tool that can analyze responses given to open-ended questions.