Show simple item record

dc.contributor.authorSawan, Aktham
dc.contributor.authorJayousi, Rashid
dc.date.accessioned2018-10-18T18:58:22Z
dc.date.available2018-10-18T18:58:22Z
dc.date.issued2018-06-26
dc.identifier.citationAnalysis of social network for telecommunication companies A Sawan, R Jayousi - ICFNDS '18 Proceedings of the 2nd International Conference on Future Networks and Distributed Systems Article No. 22en_US
dc.identifier.isbn978-1-4503-6428-7
dc.identifier.urihttps://dspace.alquds.edu/handle/20.500.12213/3737
dc.description.abstractSocial Network Analysis (SNA) is defined as the science of grouping members and finding influencers member inside each group by utilizing advanced set of algorithms. Most specialized data mining software firms, such as IBM, SAS, python and R, create their own predefined algorithms for generating the SNA groups, but none of them is dedicated for the telecom industry. The aim of this paper is to develop a customized SNA algorithm for the telecom industry, since the predefined commercial algorithm failed to generate satisfactory results when used to generate the SNA groups for the Palestinian mobile service provider company (Jawwal), such as a low capture rate (only 55%), and failed to even capture high value customers generating and receiving hundreds of calls and SMS's. In addition to customizing the SNA algorithm for the telecom industry, relation strength and extenders have been used to enhance results in this paper. In order to reach the finest telecom SNA model, oracle SQL-PL/SQL software have been utilized, and various experiments have been tested based on different specific telecom parameters, such as group size, call duration, call count, and the ratio between duration and count. To test the new developed algorithm, 300 million call detailed record (CDR) for 4 million user in three consecutive months have been collected and used, and a result comparison with the IBM SNA model is added. Results for the new algorithm have increased coverage of network to be 75.9% instead of around 55% for IBM algorithm; moreover, all high value customers have included in the results for the new algorithm. We believe that this paper is relevant to track two cloud Distributed and Parallel systems.en_US
dc.language.isoen_USen_US
dc.publisherACMen_US
dc.subjectSocial network analysisen_US
dc.subjectData miningen_US
dc.subjectTelecommunication industryen_US
dc.titleAnalysis of social network for telecommunication companiesen_US
dc.typeArticleen_US


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record