Data Science Solutions to Enhance Pension Communication
Communication is crucial in the domain of pensions and insurance. The aim of the proposed project is to leverage the power of state-of-the-art data science methods to study and enhance communication in two modalities: the textual modality (e.g., email messages between clients and pension providers) and the visual modality (e.g., tracking eye movements of clients reading information on the websites of pension providers). Textual analytics focuses on the sources of information (e.g., webpages about pensions); eye tracking analytics focuses on the recipients of the information (e.g., clients reading the webpages). Recent innovations in data science methods, most notably “deep learning,” and other machine learning algorithms give rise to unprecedented performance in textual analytics tasks, such as highly accurate sentiment recognition (e.g., the automatic recognition of a message being from a satisfied client or an unsatisfied client) and reliable automatic content classification (e.g., the message contains a question about a specific pension option). The same innovations in data science give rise to
advances in video-based eye tracking to reliably estimate eye gaze from webcam images.
The proposed project will study the benefits of using machine learning for the two modalities separately and combined in an effort to facilitate the communication in the domain of pensions and insurance. The deliverables of the project will be academic papers, reports on the achieved results, and software tools for performing the tasks at hand. The results of the project will benefit pension funds and insurance
companies in enhancing the efficiency of communication with their clients. In addition, by conducting research using real life data, cases and data-science techniques we contribute to the machine learning literature (e.g., practical applications of machine learning and deep learning) and the academic pension literature. In addition, our findings will shed light on the usability in the field of and what kind of improvements need to be made in the machine learning algorithms and techniques in the field of data science.