本論文研究廣告點擊預測方法，採用類神經網路架構為基礎，建立全體預測模型及分群預測模型，並比較採用四種不同類神經網路模型架構之預測效果。第一種架構是單屬性模型，以線性迴歸方法為基礎而建立的類神經網路模型架構；第二種架構是跨屬性模型，結合不同廣告欄位的屬性值，運用內積運算建立對應的特徵值；第三種架構是屬性權重因子分解機模型，為相關研究所提出的模型；第四種是FwFMs改良版模型，採用第三種模型的架構，但將部分參數固定採用跨欄位互信息值為權重值。本論文並對大量資料的資料分群提出兩種分割處理後再合併的分群方法，用來對測試資料選取適用的分群預測模型。第一種是雅卡爾相似分數群集法，第二種是餘弦相似分數群集法。實驗評估顯示，全體預測模型於FwFMs改良版模型架構，準確度可達76.40%。在分群預測模型中，採用四種類神經網路模型架構皆可提升準確度，最高可達76.58%。此外，採用餘弦相似分數群集法，能快速有效的對測試資料選取適當的分群預測模型。

The purpose of this study is to investigate the prediction method of advertising clicking, which adopts the foundation of neural network models. Four different neural network architectures are constructed for solving this task to compare their predicting effects. The first one is the Single Field Model, which was designed based on the linear regression method. The second one is the Cross Field Model, which combines various features of advertisement fields and performs inner product to build the corresponding features. The third one is the Field-weighted Factorization model, denoted as FwFMs, which is proposed in the related work. We modified the model of FwFMs by assigning the cross field mutual information to be the fixed weights of a hidden layer to propose the Modified-FwFMs. Furthermore, the global predicting model and group predicting model learned from the total training data sets and the clusters of similar data sets are constructed, respectively. Due to the large quantity of data, we proposed two data clustering methods, based on partitioning then merging, to construct the appropriate group predicting model for the data set. The first clustering method is designed based on the Jaccard similarity score. The second one is to perform data clustering based on the cosine similarity score. The results of the experiments show that the predicting accuracy of the global predicting model based on the Modified-FwFMs achieves 76.40%. In addition, the group predicting model can improve the prediction accuracy for all the four neural network neural network architectures. Furthermore, the cosine similarity scoring method can select appropriate group predicting model for data effectively and efficiently.

[1] Barbieri, N., Silvestri, F., & Lalmas, M. (2016). Improving Post-Click User Engagement on Native Ads via Survival Analysis. in Proceedings of the 25th International Conference on World Wide Web(WWW).
[2] Chen, J., Sun B., Lu H., & Hua, X. (2016). Deep CTR Prediction in Display Advertising. in Proceedings of the 2016 ACM on Multimedia Conference(MM).
[3] Dalessandro, B., Chen, D., Raeder, T., Perlich, C., Williams, H. M. & Provost, F. (2014). Scalable Hands-Free Transfer Learning for Online Advertising. in Proceedings of the 20th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining(KDD).
[4] Edizel, B., Mantrach, A., & Bai, X. (2017). Deep Character-Level Click-Through Rate Prediction for Sponsored Search. in Proceedings of the 40th International ACM SIGIR Conference on Research and Development in Information Retrieval(SIGIR).
[5] Graepel, T., Candela, J., Borchert, T., & Herbrich, R. (2010). Web-scale Bayesian Click-Through Rate Prediction for Sponsored Search Advertising in Microsoft’s Bing Search Engine. in Proceedings of the 27th International Conference on Machine Learning(ICML).
[6] He, X., Pan, J., Jin, O., Xu, T., Liu, B., Xu, T., Shi, Y., Atallah, A., Herbrich, R., Bowers, S. & Candela, J. (2014). Practical Lessons from Predicting Clicks on Ads at Facebook. in Proceedings of the Eighth International Workshop on Data Mining for Online Advertising(ADKDD).
[7] Huang, Z., Pan, Z., Lin, Q., Long, B., Ma, H. & Chen, E. (2017). An Ad CTR Prediction Method Based on Feature Learning of Deep and Shallow Layer. in Proceedings of the 2017 ACM on Conference on Information and Knowledge Management(CIKM).
[8] Ji, W., Wang, X. & Zhang, D. (2016). A Probabilistic Muliti-Touch Attribution Model for Online Advertising. in Proceedings of the 25th ACM International on Conference on Information and Knowledge Management(CIKM).
[9] Juan, Y., Lefortier, D. & Chapelle, O. (2017). Field-aware Factorization Machines in a Real-world Online Advertising System. In Proceedings of the 26th International Conference on World Wide Web Companion(WWW).
[10] Lee, K., Orten, B., Dasdan, A. & Li, W. (2012). Estimating Conversion Rate in Display Advertising from Past Performance Data. in Proceedings of the 18th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining(KDD).
[11] McMahan, H., Holt, G., Sculley, D., Young, M., Ebner, D., Grady, J., Nie, L., Phillips, T.,
 Davydov, E., Golovin, D., Chikkerur, S., Liu, D., Wattenberg, M., Hrafnkelsson, A., Boulos, T. & Kubica, J. (2013). Ad Click Prediction: A View from the Trenches. in Proceedings of the 19th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining(KDD).
[12] Oentaryo, R., Lim, E., Low, J., Lo, D. & Finegold, M. (2014). Predicting Response in Mobile Advertising with Hierarchical Importance-Aware Factorization Machine. in Proceedings of the 7th ACM International Conference on Web Search and Data Mining(WSDM).

[13] Pan, J., Xu, J., Ruiz, A.L., Zhao, W., Pan, S., Sun, Y. & Lu, Q. (2018). Field-weighted Factorization machines for click-Through Rate Prediction in Display Advertising. in Proceedings of the 2018 World Wide Web Conference(WWW).
[14] Ren, K., Zhang, W., Rong, Y., Zhang, H., Yu, Y. & Wang, J. (2016). User Response Learning for Directly Optimizing Campaign Performance in Display Advertising. in Proceedings of the 25th ACM International on Conference on Information and Knowledge Management(CIKM).
[15] Richardson, M., Dominowska, E. & Ragno, R. (2007). Predicting Clicks: Estimating the Click-Through Rate for New Ads. in Proceedings of the 16th International Conference on World Wide Web(WWW).
[16] Shirkhorshidi, A.S., Aghabozorg, S., Wah, T.Y. & Herawan, T. (2014). Big Data Clustering: A Review. in Proceedings of International Conference on Computational Science and Its Applications(ICCSA).
[17] Shah, P., Yang, M., Alle, S., Ratnaparkhi, A., Shahshahani, B. & Chandra, R. (2017). A Practical Exploration System for Search Advertising. in Proceedings of the 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining(KDD).
[18] Su, Y., Jin, Z., Chen, Y., Sun, X., Yang, Y., Qiao, F., Xia, F. & Xu, W. (2016). Improving Click-Through Rate Prediction Accuracy in Online Advertising by Transfer Learning. in Proceedings of the International Conference on Web Intelligence(WI).
[19] Wang, R., Fu, B., Fu, G. & Wang, M. (2017). Deep & Cross Network for Ad Click Predictions. in Proceedings of the ADKDD’17(ADKDD).
[20] Yang, H., Zhu, Y. & He, J. (2017). Local Algorithm for User Action Prediction Towards Display Ads. in Proceedings of the23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining(KDD).
[21] Yan, J., Liu, N., Wang, G., Zhang, W., Jiang, Y. & Chen, Z. (2009). How Much can Behavioral Targeting Help Online Advertising?. in Proceedings of the 18th International Conference on World Wide Web(WWW).
[22] Zhou, K., Redi, M., Haines, A. & Lalmas, M. (2016). Predicting Pre-Click Quality for Native Advertisements. in Proceedings of the 25th International Conference on World Wide Web(WWW).
[23] Zhai, S., Chang, K., Zhang, R. & Zhang, Z. (2016). DeepInttent: Learning Attention for Online Advertising with Recurrent Neural Networks. in Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining(KDD).