Automated Interview Scoring (AIS) by Integrating Wordnet with NLTK

Abstract

Manual rating of autobiographical interviews, the essence of the assessment of autobiographical memory in psychology, is a labour-intensive and time-consuming task. Here, narrative details have to be rated by human raters, that is, classified as internal (episodic) or external (non-episodic), which limits the scalability of such research efforts critically. This project looks at the application of NLP methods for automating the rating process. The objective of this research is to fine-tune a language model such as distilBERT toward the end of making its usage more efficient and relieving part of the burden from researchers without compromising the precision in content classification. The project further assesses the reliability of the automated system across various datasets and determines whether its general structure may present possible futures for advancing research in autobiographical memory. All these developments work towards focusing on efficient and scalable NLP tools development for mainstream psychological use.