在不同推薦App的方法中, 大部份是根據使用使用者下載App的記錄和使用者對於App的評分和評論進行推薦, 但使用者下載App的記錄是一個較弱的指標, 而使用者對於App的評分和評論會造成冷起始問題, 因此我們提出下載傾向的概念進行推薦, 第一步是先預測使用者的下載傾向, 當預測使用者有下載傾向時才到第二步推薦相似的App,以上的預測和推薦都是藉由監督式機器學習方法來實現, 而我們只需要收集使用者的 App列表就可以達成, 且App列表具有方便收集和高安全性的好處。
在實驗結果中預測不同的下載傾向平均的AUC分數是0.65, 而也證明可以有效地推薦相似的App, 並且以上演算法都有高度的擴展性皆可運行在Spark平台上。

In App recommendation, many methods are that leverage the user's App download history or their ratings and comments as the basis for App recommendations. The user's App download history is a weak indicator and the user's ratings and comments have the cold-start problem. So we propose the concept of downloading tendency ( tendency ). The first step is that predicting user's tendency of downloading the App in the near future, and then the second step is recommending the lookalike App to the user by supervised machine learning. We only need the user's App list, describing what App installed on user's mobile, which has fewer security risks and more collected conveniently.
In our experiment, we can identify average AUC scores at 0.65 for predicting different genre tendency and prove of retrieving lookalike App effectively and our algorithm has high scalability running on Spark.

[1] App Store. https:// www.appstore.com/.
[2] Google Play. https://play.google.com/.
[3] Apache Spark. http://spark.apache.org/.
[4] Apache Spark MLlib. http://spark.apache.org/mllib/.
[5] Zaharia, M., Chowdhury, M., Das, T., Dave, A., Ma, J., McCauley, M., ... & Stoica, I. (2012, April). Resilient distributed datasets: A fault-tolerant abstraction for in-memory cluster computing. In Proceedings of the 9th USENIX conference on Networked Systems Design and Implementation (pp. 2-2). USENIX Association.
[6] Meng, X., Bradley, J., Yuvaz, B., Sparks, E., Venkataraman, S., Liu, D., ... & Xin, D. (2016). Mllib: Machine learning in apache spark. JMLR, 17(34), 1-7.
[7] Breiman, L. (2001). Random forests. Machine learning, 45(1), 5-32.
[8] Walker, S. H., & Duncan, D. B. (1967). Estimation of the probability of an event as a function of several independent variables. Biometrika, 54(1-2), 167-179.
[9] Cox, D. R. (1958). The regression analysis of binary sequences. Journal of the Royal Statistical Society. Series B (Methodological), 215-242.
[10] Yan, B., & Chen, G. (2011, June). AppJoy: personalized mobile application discovery. In Proceedings of the 9th international conference on Mobile systems, applications, and services (pp. 113-126). ACM.
[11] Costa-Montenegro, E., Barragáns-Martínez, A. B., Rey-López, M., Mikic-Fonte, F. A., & Peleteiro-Ramallo, A. (2011). Which App? A recommender system of applications in markets by monitoring users' interaction. In 2011 IEEE International Conference on Consumer Electronics (ICCE).
[12] Zheng, V. W., Cao, B., Zheng, Y., Xie, X., & Yang, Q. (2010, July). Collaborative Filtering Meets Mobile Recommendation: A User-Centered Approach. In AAAI (Vol. 10, pp. 236-241).
[13] Lin, J., Sugiyama, K., Kan, M. Y., & Chua, T. S. (2013, July). Addressing cold-start in app recommendation: latent user models constructed from twitter followers. In Proceedings of the 36th international ACM SIGIR conference on Research and development in information retrieval (pp. 283-292). ACM.
[14] Lin, J., Sugiyama, K., Kan, M. Y., & Chua, T. S. (2013, July). Addressing cold-start in app recommendation: latent user models constructed from twitter followers. In Proceedings of the 36th international ACM SIGIR conference on Research and development in information retrieval (pp. 283-292). ACM.
[15] Review app permissions thru Android 5.9 .
https://support.google.com/googleplay/answer/6014972.
[16] Luhn, H. P. (1957). A statistical approach to mechanized encoding and searching of literary information. IBM Journal of research and development, 1(4), 309-317.
[17] Sparck Jones, K. (1972). A statistical interpretation of term specificity and its application in retrieval. Journal of documentation, 28(1), 11-21.
[18] VMFive. https://vmfive.com/.