mobile phones & social networks Jari Saramki Dept. of - PowerPoint PPT Presentation

mobile phones & social networks Jari Saramäki Dept. of Biomedical Engineering & Computational Science Aalto University Finland EIB Seminar, Feb 11, 2014, Luxembourg

Why Study Social Networks? 1 • The basic need to form & maintain social relationships has been the largest technological driver for a few decades! • Still new technologies take everyone by surprise (including their providers!) 2004 1988 1995

Big Data & Computational Social Science • Massive electronic records: • Mobile operators: calls, text messages, WiFi, ... • Online social networks: Facebook, Twitter, ... • Online purchasing and browsing histories • Allow studies of human behaviour on an unprecedented scale

From Calls to Network Analysis call detail records call operator’s billing system network: people linked if they have called one another network analysis apply mathematical & B C event sequence temporal temporal motif computational tools to subgraph repeated contact returned contact ( � t =3 ) 1.,2. 2. anonymized t =1, 1. understand network t =2, t =1, 1., 2. t =7 t =2 causal chain non-causal chain a a b b structure , its evolution , 1. 2. 2. 1. network data a c a c t =3 t =3 and its effects on out-star in-star 2. t =1 t =4 t =1 t =4 1. 3. 1. 1. b d b 2. 2. t =8 dynamical processes on networks pattern detection δ = 0 δ = 0 . 1 δ = 0 . 5 δ = 1 modelling

Features of human social networks • inhomogeneity: different social habits • homophily: similar people like to connect • assortativity: highly connected people like to connect • group structure: circles of friendship • limited personal network size: time and cognitive constraints

This talk Why Study Social Networks? 1 Weak Ties, Strong Ties 2 Persistent Social Habits 3 Patterns of Conversation 4

Weak Ties, Strong Ties 2

The weak ties hypothesis • The weak ties hypothesis (Granovetter 1973): 1) 5 min The relative overlap of two 8 min individual’s friendship networks varies 7 min directly with the strength of their tie to one another. • 2) We used anonymized call records to investigate this 20 min • Call data for 7 million people over 18 weeks • network: 3) • nodes = people, • two people linked if mutual calls, • tie strength = total duration of calls

Verification of the Weak Tie Hypothesis • Define the overlap O ij of a link as the fraction of common friends • Calculate average overlap as a function of tie strength tie strength = duration of calls • Increasing tendency observed tie strength = - hypothesis verified number of calls Onnela, Saramäki, et al. , Proc. Natl. Acad. Sci. (USA) 104 , 7332 (2007), New Journal of Physics 9 , 179 (2007)

Weak ties are crucial for connectivity! small network sample 80% of strongest links removed 80% of weakest links removed diluted fragmented Onnela, Saramäki, et al. , Proc. Natl. Acad. Sci. (USA) 104 , 7332 (2007), New Journal of Physics 9 , 179 (2007)

Weak ties act as bottlenecks for information diffusion Onnela, Saramäki, et al. , Proc. Natl. Acad. Sci. (USA) 104 , 7332 (2007), New Journal of Physics 9 , 179 (2007)

Persistent Social Habits 3

Limits to numbers of relationships time: we only have brain power: the same applies! so much of it!

Networks in flux • Data on 24 volunteer students • Students finished high school & went to university • All outgoing calls for 18 months • Three social surveys

Social signatures 10 0 a) b) A J A B C 1 0 5 B I fraction of calls D 10 − 1 5 7 2 3 E 56 4 F H C ego G 6 H 3 5 I 10 − 2 8 J 1 3 G D F E 10 − 3 0 2 4 6 8 10 rank 1) count calls to everyone 2) rank everyone, see what % of in a 6-month interval calls goes to #1, what % to #2, etc

Average signatures very large numbers of calls concentrated at top ranks top 3: males 40% of calls females 48% of calls signatures do not change over time

I : M a r - A u g 1 10 0 One in, one out average for all 10 -1 24 students fraction • 10 -2 Communication mainly with a small a ) number of others 10 -3 0 5 10 15 2 0 I : S e p - F e b 2 • 10 0 Very persistent pattern, even when friends are replaced by newcomers 10 -1 • Individual-level persistence: 10 -2 b ) 10 -3 0 5 10 15 2 0 “If you like to have two best friends, I : M a r - A u g 3 10 0 this never changes, irrespectively of alter I 1 alter I 2 who those friends are” alter I 3 10 -1 kin 10 -2 c ) 10 -3 0 5 10 15 2 0

Patterns of Conversation 4

Call & text message sequences green lines = social ties call sms

Burstiness: universal feature of human dynamics all calls by one person calls to each friend correlated call sequences Karsai et al, Phys. Rev. E 83, 025102(R) (2011) time

temporal motifs B C event sequence temporal temporal motif subgraph repeated contact returned contact ( � t =3 ) 1.,2. 2. t =1, 1. t =1, t =2, 1., 2. t =2 t =7 causal chain non-causal chain a a b b 1. 2. 2. 1. a c a c t =3 t =3 out-star in-star 2. t =4 t =4 t =1 t =1 1. 3. 1. 1. b d b 2. 2. t =8 Kovanen, Kaski, Kertész, Saramäki, PNAS 2013 • We want to detect temporal patterns, where links activate within short time periods • Patterns should be grouped into equivalence classes (motifs) based on the order of events (calls in this case) • Study pattern frequency vs properties of involved nodes

temporal motifs in call sequences • there is temporal homophily : patterns where participants are similar (age, gender) are overexpressed Kovanen, Kaski, Kertész, Saramäki, PNAS 2013

temporal motifs in call sequences • there is temporal homophily : patterns where participants are similar (age, gender) are overexpressed • there are gender differences : � causal chain non-causal chain . . 2 1 1 2 . . out-star in-star repeated contact returned contact � 1.,2. . 2. 1 1 . . 2 2 1. . females : chains & stars males : “ping-pong” Kovanen, Kaski, Kertész, Saramäki, PNAS 2013

temporal motifs in call sequences • there is temporal homophily : patterns where participants are similar (age, gender) are overexpressed • there are gender differences : � causal chain non-causal chain . . 2 1 1 2 . . out-star in-star repeated contact returned contact � 1.,2. . 2. 1 1 . . 2 2 1. . females : chains & stars males : “ping-pong” • there are group talk patterns: chains & stars within social groups Kovanen, Kaski, Kertész, Saramäki, PNAS 2013

Summary 5 • We have a small number of close relationships and many “weak links” • Those weak links are important! • Our closest friends (who resemble us!) are typically also friends • Much of our communication is with closest friends & family only • Our social patterns change only slowly, if at all

Summary 5 • Call records provide information that cannot be obtained with traditional methods (e.g. surveys) • This allows statistical detection of behavioral patterns • Also a lot of commercial interest: “data scientist” is one of the hottest professions currently