Exploitation and exploration in a performance based contextual advertising system

Contextual advertising Search advertising Click-through rate
DOI: 10.1145/1835804.1835811 Publication Date: 2010-07-27T14:10:11Z
ABSTRACT
The dynamic marketplace in online advertising calls for ranking systems that are optimized to consistently promote and capitalize better performing ads. streaming nature of data inevitably makes an system choose between maximizing its expected revenue according current knowledge short term (exploitation) trying learn more about the unknown improve (exploration), since latter might increase future. exploitation exploration (EE) tradeoff has been extensively studied reinforcement learning community, however, not paid much attention until recently. In this paper, we develop two novel EE strategies advertising. Specifically, our methods can adaptively balance aspects by automatically optimal incorporating confidence metrics historical performance. Within a deliberately designed offline simulation framework apply algorithms industry leading performance based contextual conduct extensive evaluations with real event log data. experimental results detailed analysis reveal several important findings behaviors demonstrate perform superiorly terms ad reach click-through-rate (CTR).
SUPPLEMENTAL MATERIAL
Coming soon ....
REFERENCES (12)
CITATIONS (43)
EXTERNAL LINKS
PlumX Metrics
RECOMMENDATIONS
FAIR ASSESSMENT
Coming soon ....
JUPYTER LAB
Coming soon ....