《海量数据挖掘-王永利》ch08-advertising.pptVIP

* * * * * * * * Is greedy deterministic or randomized (it is deterministic!!) * * * Why exhausts budget: -- remember, optimal exhausts both budgets GENERAL PROOF * Whatever is not used should be assigned to A2 (since if we could assign to A1 we would since we still have the budget) CASE 1) Less than half of A1’s queries got assigned to A2. So yB/2 CASE 2) If more than half of A1’s queries got assigned to A2. Consider last of A1s queries assigned to A2. At that time B2 B1. At that time B2 ?. Since more than half of A1’s queries got assigned to A2 (this means the remaining budget had to be 1/2). So B11/2. * * * * * * * * J. Leskovec, A. Rajaraman, J. Ullman: Mining of Massive Datasets, * Advertiser Bid CTR Bid * CTR A B C $1.00 $0.75 $0.50 1% 2% 2.5% 1 cent 1.5 cents 1.125 cents Two complications: Budget CTR of an ad is unknown Each advertiser has a limited budget Search engine guarantees that the advertiser will not be charged more than their daily budget J. Leskovec, A. Rajaraman, J. Ullman: Mining of Massive Datasets, * CTR: Each ad has a different likelihood of being clicked Advertiser 1 bids $2, click probability = 0.1 Advertiser 2 bids $1, click probability = 0.5 Clickthrough rate (CTR) is measured historically Very hard problem: Exploration vs. exploitationExploit: Should we keep showing an ad for which we have good estimates of click-through rate or Explore: Shall we show a brand new ad to get a better sense of its click-through rate J. Leskovec, A. Rajaraman, J. Ullman: Mining of Massive Datasets, * * J. Leskovec, A. Rajaraman, J. Ullman: Mining of Massive Datasets, Two advertisers A and B A bids on query x, B bids on x and y Both have budgets of $4 Query stream: x x x x y y y y Worst case greedy choice: B B B B _ _ _ _ Optimal: A A A A B B B B Competitive ratio = ? This is the worst case! Note: Greedy algorithm is deterministic – it always resolves draws in the same way * J. Leskovec, A. Rajaraman, J. Ullman: Mining of Massive Dataset