More Related Content
Decision generalisation from game logs in no limit 'Multilayer Networks' Section3 Yet Another ShortShort AI AI Wolf Contest -Development of Game AI using Collective Intelligence- 人工知能は人狼の夢を見るか-日本デジタルゲーム学会年次大会2013@函館 Viewers also liked (7)
人工知能は人狼の夢を見るか~人狼知能セミナー20141119 脱初心者!経験は人狼力を向上させるのか?~データから見るベテランの実力~ The bursty dynamics of the twitter information network
- 1. The Bursty Dynamics of the
Twitter Information Network
D1 臼井翔平
第一回とりらぼ輪読会 2014/5/31
Seth Myers Jure Leskovec
Stanford University
- 5. Present work
• Information causes bursts in network
evolution
• Bursts of edge creations and deletions
• Modeling and predicting bursts
- 6. Present work
• Information causes bursts in network
evolution
• Bursts of edge creations and deletions
• Modeling and predicting bursts
- 18. Present work
• Information causes bursts in network
evolution
• Bursts of edge creations and deletions
• Modeling and predicting bursts
- 30. Present work
• Information causes bursts in network
evolution
• Bursts of edge creations and deletions
• Modeling and predicting bursts
- 35. Follow確率
• 𝑃𝑗,𝑖 = 𝑃 𝑗 𝑓𝑜𝑙𝑙𝑜𝑤 𝑖 𝑌𝑖𝑗)
≡ 𝐶 ∙ exp 𝛼 ∙ 𝑌𝑖𝑗
= 𝐶 ∙ exp
𝛼
𝜎𝑖
∙ 𝑙𝑛 𝑆 𝑖, 𝑗 − 𝜇𝑖
= 𝐶 ∙
exp(𝑙𝑛 𝑆 𝑖, 𝑗 )
exp(𝜇𝑖)
𝛼
𝜎 𝑗
= 𝐶 ∙
𝑆(𝑖, 𝑗)
exp(𝜇𝑖)
𝛼
𝜎 𝑗
- 37. Follow burstの確率
• 新規followの期待値: 𝑗∈𝑁2(𝑖) 𝑃𝑗,𝑖
• Follow burstはそれまで知らなかったノードを
retweetによって知る
𝑁 𝑅𝑇(𝑖, [𝑡, 𝑡 + ∆𝑡)):区間[𝑡, 𝑡 + ∆𝑡)の間にfollowし
ている誰かがiのtweetをretweetしたノード集合