Analiza algoritama za određivanje povjerenja među prijateljima na društvenim mrežama

master's thesis

• Title (english): Analysis of Algorithms for Determining Trust among Friends on Social Networks
• Author: Mirjam Šitum
• Mentor: Vedran Podobnik (mentor)
• Committee member: Wolfdieter Merkl (predsjednik povjerenstva)
• Committee member: Jürgen Dorn (član povjerenstva)
• Committee member: Birgit Hofreiter (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2014-07-03, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Electrical Engineering

Abstract

Povjerenje je jedno od glavnih komponenti ljudske interakcije. U ovom predu prezentirana je analiza povjerenja medu prijateljima na društvenoj mreži Facebook, kao i teretski pregled najvažnijih teoretskih komponenti povjerenja. Razliciti algoritmi su implementirani i evaluriani, direktni algoritmi i algoritmi ravnopravnih sudionika. Direktni algoritmi racunaju povjerenje izmedu prijatelja na Facebooku na temelju njihovih interakcija, dok algoritmi ravnopravnih sudionika racunaju povjerenje... More izmedu bilo kojih korisnika u mreži. Tri adaptacije algoritama ravnopravnih sudionika su ucinjene u ovom radu: Tidal algoritam, Mole algoritham i Eigen algoritham. Rezultati ovih algoritama su kombinirani sa rezultatima direktnog algoritma u svrhu istraživanja mogucnosti poboljšanja rezultata. Svi algoritmi evaluirani su uz pomoc upitnika kojeg je ispunilo 100 korisnika. Korisnici su birali 5 najboljih prijatelja u 5 razlicitih konteksta i njihovi odgovori su korišteni pri evaluaciji algoritama sa cetiri razlicite mjere: Kendal Tau udaljenost, rang udaljenost, broj poklapanja i broj tocnih poklapanja. Na kraju su dani zakljucci i evaluacija rezultat, kao i mogucnosti daljnjeg rada. Less

Abstract (english)

Trust is the one of the major components of human interaction. This work gives an analysis of trust among friends on social network Facebook as well as the theoretical review of most important components of trust in general. Different algorithms are implemented and evaluated, both direct trust algorithms and peer-to-peer trust algorithms. Direct trust algorithm are the one that compute trust between friends on Facebook based on their interaction, and peer-to-peer trust algorithms compute... More trust between any two users in the network of Facebook users. Three adaptations of peer-to-peer algorithm are made in this work: Tidal trust algorithm, Mole trust algorithm and Eigen trust algorithm. Results of these algorithms are also combined with the results of direct trust algorithms for purposes of exploring new, more precise algorithm results. All of the algorithms were evaluated using the survey which was filled in by 100 users. Users choose their top 5 friends in 5 different contexts and their answers were used to evaluate algorithms with different measures: Kendal Tau distance, Rank difference, Count match and Exact match. In the end, some conclusions and evaluation of results was given with open possibilities for further work. Less

• Keywords: povjerenje; ravnopravni sudionici; Facebook; društvena mreža; interakcija; propagacija; kalibriranje težina; anketa; rangiranje; mjera udaljenosti
• Keywords (english): trust; peer-to-peer; Facebook; social network; interaction; propagation; weight calibration; survey; ranking; distance measures
• Language: english
• URN:NBN: urn:nbn:hr:168:953982
• Study programme: Title: Information and Communication Technology; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka informacijske i komunikacijske tehnologije (magistar/magistra inženjer/inženjerka informacijske i komunikacijske tehnologije)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2018-02-06 12:18:53