PROFILING - CONCEPTS AND APPLICATIONS

Profiling is an approach to put a label or a set of labels on a subject, considering the characteristics of this subject. The New Oxford American Dictionary defines profiling as: “recording and analysis of a person’s psychological and behavioral characteristics, so as to assess or predict his/her capabilities in a certain sphere or to assist in identifying a particular subgroup of people”. This research extends this definition towards things demonstrating that many methods used for profiling of people may be applied for a different type of subjects, namely things. The goal of this research concerns proposing methods for discovery of profiles of users and things with application of Data Science methods. The profiles are utilized in vertical and 2 horizontal scenarios and concern such domains as smart grid and telecommunication (vertical scenarios), and support provided both for the needs of authorization and personalization (horizontal usage).:The thesis consists of eight chapters including an introduction and a summary. First chapter describes motivation for work that was carried out for the last 8 years together with discussion on its importance both for research and business practice. The motivation for this work is much broader and emerges also from business importance of profiling and personalization. The introduction summarizes major research directions, provides research questions, goals and supplementary objectives addressed in the thesis. Research methodology is also described, showing impact of methodological aspects on the work undertaken. Chapter 2 provides introduction to the notion of profiling. The definition of profiling is introduced. Here, also a relation of a user profile to an identity is discussed. The papers included in this chapter show not only how broadly a profile may be understood, but also how a profile may be constructed considering different data sources. Profiling methods are introduced in Chapter 3. This chapter refers to the notion of a profile developed using the BFI-44 personality test and outcomes of a survey related to color preferences of people with a specific personality. Moreover, insights into profiling of relations between people are provided, with a focus on quality of a relation emerging from contacts between two entities. Chapters from 4 to 7 present different scenarios that benefit from application of profiling methods. Chapter 4 starts with introducing the notion of a public utility company that in the thesis is discussed using examples from smart grid and telecommunication. Then, in chapter 4 follows a description of research results regarding profiling for the smart grid, focusing on a profile of a prosumer and forecasting demand and production of the electric energy in the smart grid what can be influenced e.g. by weather or profiles of appliances. Chapter 5 presents application of profiling techniques in the field of telecommunication. Besides presenting profiling methods based on telecommunication data, in particular on Call Detail Records, also scenarios and issues related to privacy and trust are addressed. Chapter 6 and Chapter 7 target at horizontal applications of profiling that may be of benefit for multiple domains. Chapter 6 concerns profiling for authentication using un-typical data sources such as Call Detail Records or data from a mobile phone describing the user behavior. Besides proposing methods, also limitations are discussed. In addition, as a side research effect a methodology for evaluation of authentication methods is proposed. Chapter 7 concerns personalization and consists of two diverse parts. Firstly, behavioral profiles to change interface and behavior of the system are proposed and applied. The performance of solutions personalizing content either locally or on the server is studied. Then, profiles of customers of shopping centers are created based on paths identified using Call Detail Records. The analysis demonstrates that the data that is collected for one purpose, may significantly influence other business scenarios. Chapter 8 summarizes the research results achieved by the author of this document. It presents contribution over state of the art as well as some insights into the future work planned