3/01/2015

推特 (Twitter) 如何建立追蹤誰 (who-to-follow) 系統

Ashish Goel, Pankaj Gupta, John Sirois, Dong Wang, Aneesh Sharma, Siva Gurumurthy,
The Who-To-Follow System at Twitter: Strategy, Algorithms, and Revenue Impact, Interfaces, Volume 45, Issue 1, January-February 2015, pp. 98–107. (入圍 2014 Franz Edelman Award)

如何推薦使用者追蹤其他使用者的推文?如下圖所示,第一個是推薦贊助商,如果成功,可以幫助推特得到廣告收入;以下則是其他使用者 (楊又肇, Twitter:我們私底下其實有評分機制,聯合新聞網)




推薦贊助商使用餘絃相似度 (cosine similarity),詳細的方法則是利用拍賣 (auction) 和動態規劃(dynamic programming) 求解。推薦其他使用者使用 Personalized PageRank 和 SALSA (stochastic approach for link structure analysis),概念和 PageRank 類似,利用隨機漫步 (random walk) 和線性方程組,以決定最值得推薦的使用者,該文表一說明實驗的結果,理論證明則提供模擬次數的基礎和數值方法收斂速度。

影響?
More than one-eighth of all new connections on the Twitter network are a direct result of this system, and a substantial majority of Twitter’s revenue comes from its promoted products, for which this system was a foundation. To place this contribution into perspective, Twitter is now a publicly traded company with a market capitalization of more than $30 billion, projected annual revenue of close to $1 billion, and more than 240 million active users
Ashish Goel 教授於 2009 年到推特教授休假 (sabbatical leave),開始和推特的工程師發展演算法,取代原先的人工推薦系統。具體說明美國的產學合作方式之一。

沒有留言:

張貼留言