Sentiment Classification and Prediction of Job Interview Performance

Attracting and hiring talented employees is a challenge for companies. The job interview process is a very critical step for both employer and candidate. Having a smooth hiring process in a company will increase future employees' satisfaction. Candidates tend to share their feedback and experience of interviews and company's hiring process with others. Having a negative experience can affect its brand image and reputation as an employer. This will make it hard to attract talented employees. In this research, machine learning and neural network models, such as support vector machines, logistic regression, Naive Bayes, and long short–term memory (LSTM), were trained to predict the candidates' sentiments after a job interview. Each model was trained using several data representations and weighting approaches, such as term binary, term frequency, and term frequency–inverse document frequency (TF–IDF). As a result, training logistic regression with TF–IDF and unigram word representation achieved an F1-measure of 0.814.