Assessment of Personality from Interview Answers using Machine Learning Approach

Abstract

Advancement in technology have a fast outcome on every area of life, it may be personality development or any other area. Machine learning conveyed the encouraging result in personality prediction by analysing the data through its decision-making process. To judge one’s personality is an important part of interview which helps recruiters to analyse the reliability of the candidate. It is imperative to develop a system which identifies individuals’ personality. From various study it has been found that by the interview answers in the form of textual form is a good metric for predicting someone’s personality trait. To achieve this goal, machine learning models will be used. From past many years Machine Learning models and NLP are gaining acceptance in the area of topic modelling and text classification. Here, using Machine Learning methods and Natural Language Processing (NLP) to infer HEXACO trait values from textual information, a regression model is used. In this study various methods are compared and will come up with optimized method for personality prediction using ensembling machine learning algorithms. Text representation in the form of Term-Frequency Inverse Document Frequency and Word Embedding methods are compared to check which gives best performance as per the work. The goal is to derive a candidate's personality from the textual substance of interview responses, which distinguishes the personalized assumptions inherent in individual candidate persona judgement in less time. With lesser time requirement for successful personality prediction, our proposed methodology is energy efficient because of less power utilization and have also shown better accuracy while working on unknown data. Deep learning can be used for classification tasks to achieve finer performance, and feature engineering can be improved.