中文
点击率（CTR）的预测在网络广告中至关重要[McMahan等人，2013[1]；Juan等人，2016[2]；Wen等人，2019[3]]，其中的任务是估计用户点击推荐广告或物品的概率。在在线广告中，广告商向出版商付费，在出版商的网站上展示他们的广告。一种流行的支付模式是每次点击成本（CPC）模式[Zhou等人，2018[4]；Zhou等人，2019[5]]，广告商只有在点击发生时才会被收费。因此，出版商的收入在很大程度上依赖于准确预测CTR的能力[Wang等人，2017[6]] 。
如今，各种CTR模型层出不穷，从 Linear到 TreeBased ，再到Embedding和MLP，随着深度学习网络的推进，CTR模型也得到了充分的发展。每个模型都有其优点，例如自适应因子化网络（AFN）可以从数据中自适应地学习任意等级的交叉特征，双输入感知因式分解机（DIFM）能在矢量级有效地学习输入感知因子（用于重新加权原始特征表示）。但是CTR预测的情况总是多种多样，有时我们会面临大量的用户数据需要快速处理，有时又会缺乏用户历史信息而面临冷启动的问题。没有一种CTR模型会很好地适应所有的情况。
基于自动机器学习的启发，我们将创建一个CTR库，里面包含着目前世界上表现优异的各种CTR模型。主要根据预测时所面临的情况，根据传入的参数，来自适应地判断并且选择适当的CTR模型进行预测，以此来提高预测精度，缩短预测时间，最大化企业效率。
[1] [McMahan et al., 2013] H Brendan McMahan, Gary Holt, David Sculley, Michael Young, Dietmar Ebner, Julian Grady, Lan Nie, Todd Phillips, Eugene Davydov, Daniel Golovin, et al. Ad click prediction: a view from the trenches. In Proceedings of the 19th ACM SIGKDD international conference on Knowledge discovery and data mining, pages 1222–1230. ACM, 2013.
[2] [Juan et al., 2016] Yuchin Juan, Yong Zhuang, Wei-Sheng Chin, and Chih-Jen Lin. Field-aware factorization machines for ctr prediction. In Proceedings of the 10th ACM Conference on Recommender Systems, pages 43–50. ACM, 2016.
[3] [Wen et al., 2019] Hong Wen, Jing Zhang, Quan Lin, Keping Yang, and Pipei Huang. Multi-level deep cascade trees for conversion rate prediction in recommendation system. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 33, pages 338–345, 2019
[4] [Zhou et al., 2018] Guorui Zhou, Xiaoqiang Zhu, Chenru Song, Ying Fan, Han Zhu, Xiao Ma, Yanghui Yan, Junqi Jin, Han Li, and Kun Gai. Deep interest network for click-through rate prediction. In Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, pages 1059–1068. ACM, 2018.
[5] [Zhou et al., 2019] Guorui Zhou, Na Mou, Ying Fan, Qi Pi, Weijie Bian, Chang Zhou, Xiaoqiang Zhu, and Kun Gai. Deep interest evolution network for click-through rate prediction. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 33, pages 5941–5948, 201
[6] [Wang et al., 2017] Ruoxi Wang, Bin Fu, Gang Fu, and Mingliang Wang. Deep & cross network for ad click predictions. In Proceedings of the ADKDD’17, page 12. ACM, 2017.
英文
The prediction of click-through rate (CTR) is crucial in online advertising [McMahan et al., 2013[1]; Juan et al., 2016[2]; Wen et al., 2019[3]], where the mission is to estimate the probability that users click on a recommended ad or item. In online advertising, advertisers pay publishers to display their ads on publishers’ sites. One popular payment model is the cost-per-click (CPC) model [Zhou et al., 2018[4]; Zhou et al., 2019[5]], where advertisers are charged only when a click occurs. As a consequence, a publisher’s revenue relies heavily on the ability to predict CTR accurately [Wang et al., 2017[6]].
Nowadays, various CTR models have emerged, from Linear to TreeBased , to Embedding and MLP. With the advancement of deep learning networks, the CTR model has also been fully developed. Each model has its merits. For Instance, Adaptive Factorization Network (AFN) can adaptively learn cross features of any level from data, and Dual Input Perceptual Factorization Machine (DIFM) can effectively learn input perception factors at vector level (used to reweight original feature representations). Nevertheless, there are always various situations for CTR prediction. We are faced with a large amount of user data that needs to be processed quickly at times, and a cold boot due to the lack of user history information at others. There is no CTR model that fits well in all situations.
Inspired by automated machine learning, we will create a CTR library containing a variety of CTR models that are currently performing well in the world. Based on the incoming parameters, we will self-adaptively determine and select the appropriate CTR model for forecasting according to the situation, in order to improve forecasting accuracy, shorten forecasting time, and maximize business efficiency.
[1] [McMahan et al., 2013] H Brendan McMahan, Gary Holt, David Sculley, Michael Young, Dietmar Ebner, Julian Grady, Lan Nie, Todd Phillips, Eugene Davydov, Daniel Golovin, et al. Ad click prediction: a view from the trenches. In Proceedings of the 19th ACM SIGKDD international conference on Knowledge discovery and data mining, pages 1222–1230. ACM, 2013.
[2] [Juan et al., 2016] Yuchin Juan, Yong Zhuang, Wei-Sheng Chin, and Chih-Jen Lin. Field-aware factorization machines for ctr prediction. In Proceedings of the 10th ACM Conference on Recommender Systems, pages 43–50. ACM, 2016.
[3] [Wen et al., 2019] Hong Wen, Jing Zhang, Quan Lin, Keping Yang, and Pipei Huang. Multi-level deep cascade trees for conversion rate prediction in recommendation system. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 33, pages 338–345, 2019
[4] [Zhou et al., 2018] Guorui Zhou, Xiaoqiang Zhu, Chenru Song, Ying Fan, Han Zhu, Xiao Ma, Yanghui Yan, Junqi Jin, Han Li, and Kun Gai. Deep interest network for click-through rate prediction. In Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, pages 1059–1068. ACM, 2018.
[5] [Zhou et al., 2019] Guorui Zhou, Na Mou, Ying Fan, Qi Pi, Weijie Bian, Chang Zhou, Xiaoqiang Zhu, and Kun Gai. Deep interest evolution network for click-through rate prediction. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 33, pages 5941–5948, 201
[6] [Wang et al., 2017] Ruoxi Wang, Bin Fu, Gang Fu, and Mingliang Wang. Deep & cross network for ad click predictions. In Proceedings of the ADKDD’17, page 12. ACM, 2017.