Identifying Customer Needs using Deep Learning & Big Data: Autonomous Interactive Artificial Agent Design & Use

Lars Vangsted Bugge

Student thesis: Master thesis


In a consumer-centric business world, innovation is derived by identifying customer needs, and is therefore of crucial importance in marketing and product/service development. Despite this fact, it is still a challenge to identify valuable customer perceptions. New innovations most often fail, and there is reason to continue to locate new ways of gathering usable data. Established ethnographic and anthropological research methods have traditionally been concerned with qualitative and quantitative data gathering, conducting interviews, doing surveys, mapping customer journeys, etc. Bus as the amount of data increases at an exponential level, and consist of highly repetitive and often hard to grasp information, these methods are no longer sufficient. Researchers concerned with CRM & algorithm design/data mining have been trying to cope with this matter by using algorithms to make sense of large amounts of data, but these studies have mostly been conducted from an algorithm engineering standpoint, where model accuracy and a comparison of the performance of different known algorithms, have been of prominent importance. This report proposes a comprehensive literature review, covering a broad examination of the foundation for designing Artificial Intelligence, a definition of customer needs and customer relationship management, and which type of algorithm and research design have previously been used within data mining and customer relationship management. This research design has been chosen in order to gather a broad understanding of the field, and thereby seek for new methods to identify a state-of-the-art way of gathering customer needs using deep learning and big data. This report finds that only very few research papers within CRM and data mining has used the possibilities of deep learning and big data to develop models capable of identifying customer needs. And no studies have yet to build completely autonomous interactive agents in anticipation of gathering customer needs. Further, based on the findings of this report, a suggestion for how such a framework could be build, is presented.

EducationsMSc in Management of Innovation and Business Development, (Graduate Programme) Final Thesis
Publication date2019
Number of pages77
SupervisorsClaus Varnes