讲座:http://bdai.ruc.edu.cn/?p=118
Modeling User Engagement for Ad and Search
ppt 链接:
Dr. Ke(Adam) Zhou
work 1: Modeling User Engagement for Ad and Search,文章链接:http://dl.acm.org/citation.cfm?id=2883053
user engagement
一,how to measure:
1, 分析学:CTR,absence time
2, 自我报告:调查问卷
3,生理学:实验室设备测量
二,maximizing short term vs long-term engagement
三,modeling user engagement (predicting) focus on ad.
online advertising: search ads, display ads, native ads(原生广告):facebook suggested post, twitter promote twitters.
ad ranking:maxmizing the profile but for long time it damage the profile, so we should take nattive advertising quality into consideration.
pre-click ad quality:点击之前
(1) how to measure?
1,CTR(click through rate): relevance, quality
2, ads feedback(offensive feedback rate OFR)
correlation: high CTR maybe high OFR
so, to measure the ads quality, it is better to use OFR.
(2) what makes an ad preferred by user?
美观》产品》品牌
(3)engineer ad quality features
Post-click quality metrics
dwell-time
features
work 2:personalized pre-click quality model(个性化),文章链接:http://dl.acm.org/citation.cfm?id=2883053