From a data perspective, the primary purpose of the Internet in the limited network environment of the past was information acquisition. As the network environment improved, people started posting their thoughts in blogs in reverse-chronological order. Furthermore, with the popularization of smart devices, it has become possible for users to quickly and easily share their feelings through social networking messages. However, to extract meaningful values from the social network service (SNS), sentiment analysis is required for handling a user’s emotional data because SNS data include a large amount of personal feelings rather than facts. Furthermore, user profile analysis is restricted by data granularity, media diversity, and privacy invasion in mobile environments. In this paper, in order to overcome these limitations, we propose a sentiment user profile system capable of extracting a user profile from individual data by comparing the available external data for the features of a sentiment ontology tree (SOT) based on polarity comparison. The proposed system is constructed based on a sentiment hierarchy with unstructured mobile data, and it compensates for the concentration of a single medium. Furthermore, it is able to analyze individual data without an invasion of privacy on mobile devices.
Unfortunately, ACM prohibits us from displaying non-influential references for this paper.
To see the full reference list, please visit http://dl.acm.org/citation.cfm?id=2938597.