Semi-autonomous methodology to validate and update customer needs database through text data analytics

To develop highly competitive products, companies need to understand customer needs (CNs) by effectively gathering and analysing customer data. With the advances in Information Technology, customer data comes not only from surveys and focus groups but also from social media and networking sites. Few studies have focused on developing algorithms that are devised exclusively to help to understand customer needs from big opinion data. Topic mining, aspect-based sentiment analysis and word embedding are some of the techniques adopted to identify CNs from text data. However, most of them do not consider the possibility that part of the customer data analysed is already known by companies. With the aim to continuously enhance company understanding of CNs, this paper presents an autonomous methodology for automatically classifying a set of text data (customer sentences) as referring to known or unknown CN statements by the company. For verification purposes, an example regarding a set of customer answers from an open survey questionnaire regarding the climate system of a car is illustrated. Results indicate that the proposed methodology helps companies to validate and update the customer need database with an average of 90 % precision and 60 % recall.