en
0.25
0.5
0.75
1.25
1.5
1.75
2
Measuring Tie-Strength in Implicit Social Networks
Published on Sep 27, 20132660 Views
Given a set of people and a set of events attended by them, we address the problem of measuring connectedness or tie strength between each pair of persons. The underlying assumption is that attendance
Related categories
Chapter list
Measuring Tie-Strength in Implicit Social Networks00:00
Problem Definition00:12
Motivation00:46
Tie Strength01:35
Macbeth02:10
Decisions, Decisions02:58
Outline03:30
Running Example04:08
Axioms04:36
Axiom 1: Isomorphism04:49
Axiom 2: Baseline05:10
Axiom 3: Frequency & Axiom 4: Intimacy06:04
Axiom 5: Popularity06:50
Axioms 6 & 7: Conditional Independence of People and of Events07:21
Axiom 8: Submodularity08:36
Example - Mapping to Axioms09:19
Observations on the Axioms09:57
Inherent Tension Between Frequency & Intimacy10:48
Observations on the Axioms (cont.)11:28
Preamble to the Characterization Theorem12:12
Characterizing Tie-Strength13:15
Many Measures of Tie-Strength14:41
Non Self-Referential Tie-Strength Measures15:00
Self-Referential Tie-Strength Measures16:19
Measures of Tie-Strength that Satisfy All the Axioms17:01
Measures of Tie-Strength that Do Not Satisfy All the Axioms18:10
Tie-Strength and Orderings18:35
Tie-Strength and Orderings: Details18:54
Data Sets19:24
Degree Distributions20:26
Completeness of Axioms 1‐820:48
Take-away point #121:56
Two Tie-Strength Functions that Do Not Satisfy the Axioms22:14
Soundness of Axioms 1‐822:49
More on Soundness23:19
Even More on Soundness - 124:00
Even More on Soundness - 224:15
Take-away point #225:02
Take-away point #325:12
Tie-Strength Measures Used in Rank Correlation Experiments25:38
Kendall τ Coefficient25:49
Adamic‐Adar, Delta, & Linear produce TS rankings that are highly correlated26:00
Common Neighbor & Max produce TS rankings that are mostly uncorrelated26:51
Take-away point #427:27
Scalability Issue27:52
Related Work28:14
Conclusions29:48
Take-away point #530:27
Thank You!31:01