Association Rules Mining Analysis of App Usage Based on Mobile Traffic Flow Data
Abstract
With the rapid development of mobile Internet, more and more Apps emerge in people’s daily life. It is important to analyze the relations a Apps, which is helpful for network management and control. In this paper, we utilize network footprint data which consists of DPI data from ISPs and Crawler data from Web for App usage analysis. Focusing on the most popular Apps in China, we propose a distributed NFP data collection and processing framework. We do association rules mining on NFP data by using Apriori and MS-Apriori algorithm. Experimental results validate our proposed method and present some interesting association rules of Apps.
Existing Work
There are many studies on the Web usage mining and association rules mining is one of the major research directions. discussed association rules within the scope of their WHOWEDA (warehouse of web data) project which is supported by a web data model and a set of algebraic operators. proposed a new association rules algorithm based on Apriori which is suitable for dynamic database mining and applied it to the web log mining. used two intelligent algorithms for predicting the user behavior’s namely Apriori and Eclat and also did the performance comparison of the two algorithms in terms of time and space complexity for the filtered data. the authors proposed a hybrid model combining the advantages of neighborhood models and latent factor models based on collaborative filtering to recommend Apps by predicting users’ behavior of App usage. a Next-App prediction service was built to demonstrate the use of the framework and to evaluate the usefulness of such a prediction service. However their data comes from server of specific Web or Apps and application market. Then the analysis is limited to the inside of the Web or few Apps. Sometimes, the App analysis is limited to only downloading of users, not the usage behavior. In order to analyze users’ App usage pattern, we use the NFP data to do association rule mining and find the hidden rules a Apps.
Proposed System
In this paper we focus on App usage analysis by association rule analysis for the NFP data. Our main task is to obverse which Apps are together used. We collect data from ISP traffic data and App detail information by Crawler of the most popular Apps in China. And we used Apriori and MS- Apriori mainly do association rules analysis.MS-Apriori is a algorithm based on Apriori to solve the rare item problem that some item appears rarely. Through grouping by users of NFP data in specific period, we can get users visitation of Apps in this period. Then using Apriori and MS-Apriori, we generate rules and discuss how to select interesting rules. Finally, we analyze the rules we get. In this paper, we propose an App usage association rules analysis system. Based on this, App developers can recommend their App to targeted crowd to achieve better results.
CONCLUSION
In this paper, we propose a data model NFP and collect NFP from DPI of ISPs and Crawler data of App detailed information. Then, we do association rule mining for the NFP data. We find many association rules which are reasonable or look like incredible and the rules are different at different time periods. The rules provide insight for App developers to recommend other Apps to their users. And developers also can have knowledge of their users’ interest and usage pattern. For the next step we want to use the same method to do association rules for the categories instead of the specific App or for the people set who use the same App in same time period to enrich our experiment results.
REFERENCES
[1] R. Antonello, S. Fernandes, D. Sadok, and J. Kelner, “Characterizing signature sets for testing dpi systems,” in GLOBECOM Workshops (GC Wkshps), 2011 IEEE, pp. 678–683, IEEE, 2011.
[2] T. J. Parvat and P. Chandra, “Performance improvement of deep packet inspection for intrusion detection,” in Wireless Computing and Network- ing (GCWCN), 2014 IEEE Global Conference on, pp. 224– 228, IEEE, 2014.
[3] S.K.Madria,C.Raymond,S.Bhowmick,andM.Mohania,“Association rules for web data mining in whoweda,” in Digital Libraries: Research and Practice, 2000 Kyoto, International Conference on., pp. 227–233, IEEE, 2000.
[4] R. Luan, S. Sun, J. Zhang, F. Yu, and Q. Zhang, “A dynamic improved apriori algorithm and its experiments in web log mining,” in Fuzzy Systems and Knowledge Discovery (FSKD), 2012 9th International Conference on, pp. 1261–1264, IEEE, 2012.
[5] B. Kotiyal, A. Kumar, B. Pant, R. Goudar, S. Chauhan, and S. Junee, “User behavior analysis in web log through comparative study of eclat and apriori,” in Intelligent Systems and Control (ISCO), 2013 7th International Conference on, pp. 421–426, IEEE, 2013.
[6] J. Lovinger and I. Valova, “Scrubbing the web for association rules: An application in predictive text,” in Machine Learning and Applications (ICMLA), 2015 IEEE 14th International Conference on, pp. 439–442, IEEE, 2015.
[7] T.-P. Hong, C.-S. Kuo, and S.-C. Chi, “Trade-off between computation time and number of rules for fuzzy mining from quantitative data,” International Journal of Uncertainty, Fuzziness and Knowledge-Based Systems, vol. 9, no. 05, pp. 587–604, 2001.
[8] S. Matthews, M. Gongora, and A. Hopgood, “Evolving temporal fuzzy association rules from quantitative data with a multi-objective evolu- tionary algorithm,” Hybrid Artificial Intelligent Systems, pp. 198–205, 2011.
[9] H. Han and R. Li, “A practical compartmentation approach for the android app coexistence,” in Proceedings of the 2017 International Conference on Cryptography, Security and Privacy, pp. 49–55, ACM, 2017.
[10] M. Hashemi and J. Herbert, “A pro-active and dynamic prediction assistance using baranc framework,” in Proceedings of the International Workshop on Mobile Software Engineering and Systems, pp. 269–270, ACM, 2016.
[11] C. Zhang, Y. Sun, J. Guo, and T. Xiong, “Mining dynamic association rules from multiple time-series data based on data of power plant,” in Advanced Information Management, Communicates, Electronic and Automation Control Conference (IMCEC), 2016 IEEE, pp. 1968–1972, IEEE, 2016.
[12] H.-Aˆ.Cao,T.K.Wijaya,K.Aberer,andN.Nunes,“Temporalassoci- ation rules for electrical activity detection in residential homes,” in Big Data (Big Data), 2016 IEEE International Conference on, pp. 3097– 3106, IEEE, 2016.
[13] V. Kadappa, S. Guggari, and A. Negi, “Teacher recruitment data analytics using association rule mining in indian context,” in Data Science and Engineering (ICDSE), 2016 International Conference on, pp. 1–7, IEEE, 2016.
[14] B. F. Momin and V. S. Kande, “Mining semantic content using association rules in video surveillance,” in Advanced Communication Control and Computing Technologies (ICACCCT), 2016 International Conference on, pp. 705–708, IEEE, 2016.
[15] R. Agrawal, R. Srikant, et al., “Fast algorithms for mining association rules,” in Proc. 20th int. conf. very large data bases, VLDB, vol. 1215, pp. 487–499, 1994.
[16] B. Liu, W. Hsu, and Y. Ma, “Mining association rules with multiple minimum supports,” in Proceedings of the fifth ACM SIGKDD interna- tional conference on Knowledge discovery and data mining, pp. 337– 341, ACM, 1999. 60