User Profile for Provisioning Information and Communication Services Based on User Influence
Total Page:16
File Type:pdf, Size:1020Kb
FACULTY OF ELECTRICAL ENGINEERING AND COMPUTING Vanja Smailović USER PROFILE FOR PROVISIONING INFORMATION AND COMMUNICATION SERVICES BASED ON USER INFLUENCE DOCTORAL THESIS Supervisor: Assistant Professor Vedran Podobnik, PhD Zagreb, 2016 FAKULTET ELEKTROTEHNIKE I RAČUNARSTVA Vanja Smailović KORISNIČKI PROFIL ZA PRUŽANJE INFORMACIJSKIH I KOMUNIKACIJSKIH USLUGA ZASNOVANIH NA UTJECAJNOSTI KORISNIKA DOKTORSKI RAD Mentor: doc. dr. sc. Vedran Podobnik Zagreb, 2016. Doktorski rad izrađen je na Sveučilištu u Zagrebu Fakultetu elektrotehnike i računarstva, na Zavodu za telekomunikacije. Mentor: doc. dr. sc. Vedran Podobnik Doktorski rad ima: 153 stranice Doktorski rad br.: __________ About the Supervisor: Vedran Podobnik was born in Zagreb in 1982. He received M.Eng. (2006, Electrical Engineering) and Ph.D. (2010, Computer Science) degrees from the University of Zagreb, Faculty of Electrical Engineering and Computing (FER), Zagreb, Croatia, as well as M.Phil. (2013, Technology Policy) degree from the University of Cambridge, Judge Business School, Cambridge, UK. From 2011 he is an Assistant Professor at the Department of Telecommunications at FER. He is the founder and Director of the “Social Networking and Computing Laboratory (socialLAB)”. He led several national and international scientific and industrial projects. Currently he is a project leader of the research project "Managing Trust and Coordinating Interactions in Smart Networks of People, Machines and Organizations" financed by the Croatian Science Foundation and the industrial project "A Platform for Context-Aware Social Networking of Mobile Users" financed by Ericsson Nikola Tesla. He is also a project coordinator of the ERASMUS+ project "Innovative ICT Solutions for the Societal Challenges" financed by the EU. His teaching and research activities are in a transdisciplinary field of smart networks, social computing and technology policy. He co-authored over 70 papers, including publications in Information Sciences, AI Magazine and Cybernetics and Systems journals. Asisst. Prof. Podobnik is a member of IEEE, ACM, INFORMS and KES International associations, as well as Cambridge Union Society. He participated in 25 conference international programs committees, and he serves as a technical reviewer for various international journals. He was a leader of an interdisciplinary team which was awarded the highest national award for notable achievements in the education activity (2015, awarded by the Croatian Parliament). As a junior researcher, he received the Croatian Annual National Award for Science in the field of technical sciences (2011, awarded by the Croatian Parliament), as well as the Silver Medal „Josip Loncar“ award for outstanding doctoral dissertation and particularly successful scientific research (2010, awarded by FER). O mentoru: Vedran Podobnik rođen je u Zagrebu 1982. godine. Diplomirao je u polju elektrotehnike te doktorirao u polju računarstva na Sveučilištu u Zagrebu Fakultetu elektrotehnike i računarstva (FER), 2006. odnosno 2010. godine. Također je magistrirao u području upravljanja tehnologijom 2013. godine na Sveučilištu u Cambridgeu, Judge Business School (Ujedinjeno Kraljevstvo). Od 2011. godine radi na Zavodu za telekomunikacije FER-a. Utemeljitelj je i voditelj "Laboratorija za društveno umrežavanje i društveno računarstvo (socialLAB)". Bio je voditelj nekoliko domaćih i međunarodnih znanstvenih projekata te projekata suradnje s gospodarstvom. Trenutno je voditelj istraživačkog projekta "Managing Trust and Coordinating Interactions in Smart Networks of People, Machines and Organizations" koji financira Hrvatska zaklada za znanost te industrijskog projekta "A Platform for Context-Aware Social Networking of Mobile Users" koji financira kompanija Ericsson Nikola Tesla. Također je koordinator projekta ERASMUS+ "Innovative ICT Solutions for the Societal Challenges" koji financira EU. Provodi nastavne i istraživačke aktivnosti u transdisciplinarnom području pametnih mreža, društvenog računarstva i upravljanja tehnologijom. Objavio je više od 70 radova, uključujući radove u časopisima Information Sciences, AI Magazine i Cybernetics and Systems. Doc. Podobnik član je stručnih udruga IEEE, ACM, INFORMS i KES International te društva Cambridge Union Society. Sudjelovao je u 25 međunarodnih programskih odbora znanstvenih konferencija, te sudjeluje kao recenzent u većem broju inozemnih časopisa. Bio je voditelj interdisciplinarnog tima koji je nagrađen najvišom hrvatskom državnom nagradom u području edukacije (2015., dodijelio Hrvatski sabor). Kao mladi istraživač primio je godišnju nacionalnu nagradu za znanost u području tehničkih znanosti (2011., dodijelio Hrvatski sabor), kao i srebrnu plaketu "Josip Lončar" za posebno istaknutu doktorsku disertaciju (2010., dodijelio FER). ACKNOWLEDGEMENTS First and foremost, I want to thank my supervisor Vedran Podobnik – it has been an honor to be his first PhD student. I appreciate the time, effort, motivation and guidance he has given me throughout my whole PhD pursuit. I am glad I have accepted his advice to apply and enroll in the study because it has led me onto a wonderful journey filled with positive experience. I gratefully acknowledge the support of Darko Huljenić, whose almost father- like wisdom and encouragement have helped me find funding for my PhD research. I appreciate the patience and support of Saša Dešić and other colleagues at Ericsson Nikola Tesla and Department of Telecommunications who helped funding my PhD as well. As for my friends and colleagues, I am grateful to Matija Šulc for being a true friend all these years and to Darko Štriga for his efforts with the development and coding phase. I am also grateful to my cousins Anes and Vedran for their encouragement and visionary inspiration with regard to engineering matters ever since I was little. The evaluation phase of my PhD would not have been so thorough without the knowledge and aid of Marko Katavić, Marko Lucić, Mara Louč and Sonja Šimpraga whose contribution I am truly thankful for. Additionally, I appreciate the expertise with which my friends Louč, Dino and Mislav have used their partying, socializing and musicianship skills needed to prolong my PhD pursuit as much as possible – jokes aside, I thank them for all the help and support in crafting my PhD as well. Lastly, I want to thank my mom Vesna, dad Snješko and sister Una for all of their love, inspiration and support – I owe you the encouragement to pursue my dreams because without you, all of this would not have been even remotely possible. Finally, I am thankful to my loving, patient, encouraging, and supportive Zrinka whose faithful efforts and contribution during the final stages of this PhD I am so appreciative of. "The real voyage of discovery consists not in seeing new sights, but in having new eyes." – Marcel Proust SUMMARY User profile enables collection, storage and interpretation of user data, which in turn enables analysis and reasoning upon such data. It is an ongoing research challenge to utilize vast amounts of available multi-source, heterogeneous user data with the goal of identifying key, socially influential actors for provisioning information and communication services. Novel approach to solving this challenge, as proposed in this thesis, includes several steps. First, a novel user profiling method is proposed in order to efficiently acquire, aggregate and consolidate user data from two data-sources – the telecom-operator's network and Interned-based social networks. Second, a novel user profile model is developed for programmatic inference upon such user data. Third, an improved algorithmic approach to reasoning upon profile data is proposed, resulting in new knowledge about each user – their social influence. The proposed algorithm improvements in calculating social influence are statistically validated and evaluated through an experiment with real-world profile data from real people. Substantial difference in results between the two data-sources is statistically proven and discussed, providing insight into their synergy once utilized together with the goal of having a thorough measure of user's social influence. Finally, out of several novel algorithm proposals stemming from their literature-based predecessor called the Limited Recursive Algorithm (LRA), the variant with Sample-Literature-Optimal Posting Frequency factor (SLOF) demonstrates significant impact and improvement, underlining the overall added value of thesis' scientific contribution. For example, churn-prevention, prioritizing customer care, digital advertising and disease-tracking algorithms could all benefit from the SLOF algorithm as exemplified in the thesis. Keywords: social networking service, user profile, user profiling, social networks, social network analysis, social influence calculation. PROŠIRENI SAŽETAK Naslov disertacije: Korisnički profil za pružanje informacijskih i komunikacijskih usluga zasnovanih na utjecajnosti korisnika Uvod Danas, svatko od nas istovremeno sudjeluje u više društvenih mreža koje prožimaju različite dijelove naših života – obitelj, prijatelji, hobiji i posao – su samo neke od njih. Navedene društvene mreže iz stvarnoga života su sve češće i češće prisutne i u virtualnoj digitalnoj dimenziji omogućenoj Internetom, gdje se pojavljuju u obliku informacijsko-komunikacijskih usluga, a korisnici im često pristupaju pomoću pametnih pokretnih uređaja ili drugih pokretnih uređaja spojenih na Internet. Za korištenje ovakvih usluga korisnik mora, implicitno ili eksplicitno,