280 likes | 553 Views
FAST-PPR: Personalized PageRank Estimation for Large Graphs. Peter Lofgren (Stanford ) Joint work with Siddhartha Banerjee (Stanford), Ashish Goel (Stanford), and C. Seshadhri (Sandia). Motivation: Personalized Search. Motivation: Personalized Search. Re-ranked by PPR.
E N D
FAST-PPR: Personalized PageRank Estimation for Large Graphs Peter Lofgren (Stanford) Joint work with Siddhartha Banerjee (Stanford), Ashish Goel (Stanford), and C. Seshadhri (Sandia)
Motivation: Personalized Search Re-ranked by PPR
Result Preview 1.2 hour 6 min 2 sec Local-Update Fast-PPR Monte-Carlo
Previous Algorithm: Monte-Carlo Previous Algorithm: Monte-Carlo [Avrachenkov, et al 2007]
Previous Algorithm: Local Update [Anderson, et al 2007]
Bidirectional PageRank Algorithm u Reverse Work (Frontier Discovery) Forward Work (Random Walks)
Empirical Running Time Log Scale
Thank You • Paper available on Arxiv • Code available at cs.stanford.edu/~plofgren
Frontier is Important Frontier Aided Significance Thresholding
Average Running Time Reverse Work (Local Update) Forward Work (Monte-Carlo)
Local Update Algorithm v1 u t v2 U U U v3 U