近年來，由於行動裝置的普及，人類的生活習慣逐漸地被改變。運行在行動裝置上的行動應用程式(App)，也對行動裝置使用者帶來很大的影響。根據官方的統計，至今已釋出了三百多萬的行動應用程式。由於大量的行動應用程式，持續地被釋出，使用者要從中找到自己需要且感興趣的行動應用程式，成為一個急需克服的難題。
本研究從使用者回饋中萃取出使用者的需求，以此為基礎建立行動應用程式推薦系統。我們利用文字探勘方法中的主題模型，來處理使用者所留下文字訊息，並從中萃取出隱藏在眾多文字訊息中的主題。此外，我們也藉由觀察使用者過去安裝的行動應用程式，來找出使用者的喜好。以主題的分布機率來代表每個行動應用程式中的功能或特色，並結合使用者的喜好，來估算某行動應用程式對不同使用者的推薦分數。最後，依據推薦分數降冪排序，產生推薦清單給每位目標使用者。
本研究以Recall@M作為指標來評估個人化推薦系統之品質。首先，本研究的推薦效能，得以顯著地提升行動應用程式商場所提供的推薦清單之效能。此外，和其他文字訊息相比，如行動應用程式的開發者描述及更新公告，使用使用者回饋來標記行動應用程式的功能，是更為有效的變數。最後，實驗結果顯示，將本研究所提出的方法，結合現存文獻中的技術，能明顯提升推薦效能。

Recently, the explosive development of mobile device has dramatically changed human life, mobile application becomes pervasive as well. Nowadays, there have released about 3 million mobile applications. Due to the tremendous and still increasing number of mobile application, user get harder to find needed apps.
To tackle this problem, we propose a personalized recommender system based on the features of textual data. Specifically, we apply LDA to extract hidden topics of user reviews and use the probability of topic distribution to represent the features of app. Further, construct user profile based on his or her consumed apps in order to realize user requirements and preference. Eventually, for each app, we take account both the topic distribution and user preference to estimate a recommended score for target user and sort candidate apps by descending score to come out a personalized app recommended list.
For the evaluation, we crawl the real-world dataset and adopt Recall@M as the measurement of performance. The experimental result shows our proposed mechanism outperforms the baseline and is able to enhance the performance of the state-of-the-art recommender systems. We then concluded that the user feedback is an effective variable to represent the features of app and plays significant role on app personalized recommender system.

CHAPTER 1-Introduction 1
1.1. Background and Motivation 1
1.2. Results and Contribution 5
1.3. Overall Architecture 5
CHAPTER 2-Literature Review 6
2.1. Semantic Mining Approach 6
2.2. Mobile Application Recommender System 8
CHAPTER 3-Methodology 11
3.1. Research Process 11
3.2. Data Collection 13
3.3. Data Preprocess 15
3.4. Generate latent topics 17
3.5. Construct user profile 18
3.6. Estimate weight of user topic 19
3.7. Calculate the probability of a user like an app 21
CHAPTER 4-Empirical Evaluation 22
4.1. Dataset description 24
4.2. Experimental Settings 26
4.3. Evaluation metric 27
4.4. Result and discussion 28
CHAPTER 5-Conclusion and Future work 39
References 40

[1] G. Linden, B. Smith, and J. York, “Amazon.com recommendations: item-to-item collaborative filtering,” IEEE Internet Comput., no. February, 2003.
[2] J. Lin, K. Sugiyama, M.-Y. Kan, and T.-S. Chua, “Addressing Cold-Start in App Recommendation : Latent User Models Constructed from Twitter Followers Categories and Subject Descriptors,” Proc. 36th Int. ACM SIGIR Conf. Res. Dev. Inf. Retr., pp. 283–292, 2013.
[3] N. Chen, S. C. H. Hoi, S. Li, and X. Xiao, “SimApp : A Framework for Detecting Similar Mobile Applications by Online Kernel Learning,” Proc. 8th ACM Int. Conf. Web Search Data Min., pp. 305–314, 2015.
[4] J. Lin, K. Sugiyama, M.-Y. Kan, and T.-S. Chua, “New and Improved: Modeling Versions to Improve App Recommendation,” Proc. 37th Int. ACM SIGIR Conf. Res. Dev. Inf. Retr., pp. 647–656, 2014.
[5] P. Yin, P. Luo, and W. Lee, “App Recommendation : A Contest between Satisfaction and Temptation,” Proc. 6th ACM Int. Conf. Web search data Min., vol. 7, no. 1, pp. 395–404, 2013.
[6] R. Baeza-yates, D. Jiang, and B. Harrison, “Predicting The Next App That You Are Going To Use Categories and Subject Descriptors,” Proc. 8th ACM Int. Conf. Web Search Data Min., pp. 285–294, 2015.
[7] D. Yankov and P. Berkhin, “Interoperability Ranking for Mobile Applications Categories and Subject Descriptors,” Proc. 36th Int. ACM SIGIR Conf. Res. Dev. Inf. Retr., no. Section 3, pp. 857–860, 2013.
[8] B. Fu, J. Lin, L. Li, and C. Faloutsos, “Why people hate your app: making sense of user feedback in a mobile app store,” Proc. 19th ACM SIGKDD Int. Conf. Knowl. Discov. data Min., pp. 1276–1284, 2013.
[9] D. Pagano and W. Maalej, “User feedback in the appstore: An empirical study,” 21st IEEE Int. Requir. Eng. Conf., pp. 125–134, 2013.
[10] C. Sangani, “Sentiment Analysis of App Store Reviews,” Methodol. 4.1, 2013.
[11] G. A. Miller, “WordNet : A Lexical Database for English,” Commun. ACM, vol. 38, no. 11, pp. 39–41, 1995.
[12] D. M. Blei, A. Y. Ng, and M. I. Jordan, “Latent Dirichlet Allocation,” J. Mach. Learn. Res., vol. 3, pp. 993–1022, 2003.
[13] E. Guzman and W. Maalej, “How do users like this feature? a fine grained sentiment analysis of app reviews,” 22nd Int. Requir. Eng. Conf., pp. 153–162, 2014.
[14] J. Herlocker, J. Konstan, A. Borchers, and J. Rield, “An algorithmic Framework for Performing Collaborative Filtering,” Proc. 22nd Annu. Int. ACM SIGIR Conf. Res. Dev. Inf. Retr., pp. 230–237, 1999.
[15] K. Shi and K. Ali, “GetJar Mobile Application Recommendations with Very Sparse Datasets,” Proc. 18th ACM SIGKDD Int. Conf. Knowl. Discov. data Min., pp. 204–212, 2012.
[16] S. Bird, “NLTK : The Natural Language Toolkit,” Proc. COLING/ACL Interact. Present. Sess., no. July, pp. 69–72, 2006.
[17] S. Panichella, A. Di Sorbo, E. Guzman, C. Visaggio, G. Canfora, and H. Gall, “How Can I Improve My App? Classifying User Reviews for Software Maintenance and Evolution,” Proc. 31st Int. Conf. Softw. Maint. Evol. (ICSME 2015). Bremen, Ger., no. 1, pp. 281–290, 2015.
[18] D. Ramage, E. Rosen, J. Chuang, C. D. Manning, and D. A. Mcfarland, “Topic Modeling for the Social Sciences,” Neural Inf. Process. Syst. Work. Appl. Top. Model. Text Beyond, pp. 2–5, 2009.
[19] C. Lin, “Projected Gradient Methods for Non-negative Matrix Factorization,” Neural Comput., vol. 19, no. 10, pp. 2756–2779, 2007.
[20] B. Liu and N. Z. Gong, “Personalized Mobile App Recommendation : Reconciling App Functionality and User Privacy Preference,” Proc. Eighth ACM Int. Conf. Web Search Data Min., pp. 315–324, 2015.
[21] T. L. Griffiths and M. Steyvers, “Finding scientific topics,” Proc. Natl. Acad. Sci., pp. 5228–5235, 2004.
[22] C. Wang and D. M. Blei, “Collaborative Topic Modeling for Recommending Scientific Articles,” Proc. 17th ACM SIGKDD Int. Conf. Knowl. Discov. data Min., pp. 448–456, 2011.
[23] X. Luo, M. Zhou, Y. Xia, and Q. Zhu, “An Ef fi cient Non-Negative Matrix-Factorization-Based Approach to Collaborative Filtering for Recommender Systems,” IEEE Trans. Ind. Informatics, vol. 10, no. 2, pp. 1273–1284, 2014.