On evaluation of customer dissatisfaction index based on sentiment analysis: a statistical approach

A customer dissatisfaction index based on the fusion of sentiment analysis as a part of natural language processing and the statistical confidence measure is proposed. To extract the polarity of a text verbatim we introduce a new concept of qualifier and qualified which leads to the overall process of evaluating a measure of customer dissatisfaction (referred to here as the customer dissatisfaction index) being more robust and less subjective. We also propose a simple method for testing the degree of customer sentiments. The method is illustrated through the real life complaint data for automobile repairs.